TD 07 96

HELSINKIUNIVERSITYOFTECHNOLOGY
FacultyofElectronics,CommunicationsandAutomation
DepartmentofCommunicationsandNetworking
TapioLev
SCENARIOANALYSISONFUTUREINTERNET
Thesis submitted in partial fulfillment of the requirements for the degree of Master of
ScienceinEngineering
Espoo,May15th,2009
Supervisor: ProfessorHeikkiHmminen
Instructor:
DocentKaleviKilkki
HELSINKIUNIVERSITYOFTECHNOLOGY
ABSTRACTOFMASTERSTHESIS
DepartmentofCommunicationsandNetworking
Author:
TapioLev
Title:
ScenarioAnalysisonFutureInternet
Date:
May15th,2009
Faculty:
Numberofpages: 6+94
Professorship:
S38NetworkingTechnology
Supervisor:
ProfessorHeikkiHmminen
Instructor:
DocentKaleviKilkki
TheInternethasgrownoutofitsoriginalscopeandscalewhileitsimportanceforthe
society has increased. Improved awareness of the challenges the Internet is
confronting has activated decisionmakers around the world, and various initiatives
have been established to study the Future Internet. All these research efforts share,
however,thesamechallenge,howtodirecttheresearchtothemostrelevanttopics.
This thesis eases the problem through scenarios which disclose the most significant
trendsanduncertaintieshavingimpactonInternetevolutionfor10yearsfromnow.
Before looking to the future, the historical milestones and current situation of the
Internet ecosystem are studied based on various written sources. With the help of
ideas and information gathered from brainstorming sessions and expert interviews,
four alternative evolution scenarios for the Internet are developed by using
Schoemakers scenario planning method. Use of PEST framework in identifying key
trends and key uncertainties ensures that all the important macroenvironmental
factorsaffectingtheFutureInternetaretakenintoaccount.Finally,thescenariosare
usedinanalyzingtheresearchstrategyoftheFinnishFutureInternetprogram.
ThescenariosshowthatthechallengestheInternetisfacingcanbesolvedinvarious
wayswhichleadtodifferentnetworkandbusinessarchitectures.Differencesbetween
the scenarios concerning the power relationships and value distribution between
stakeholders reveal the underlying tensions and differing interests of stakeholders.
Strategic analysis suggests that deployment of new solutions needs to be planned
carefully already from the beginning. Overall, the results are valuable in guiding the
discussionabouttheFutureInternet.
Keywords:Internet;scenarioanalysis;networkarchitecture;valuedistribution
TEKNILLINENKORKEAKOULU
DIPLOMITYNTIIVISTELM
Elektroniikan,tietoliikenteenjaautomaationtiedekunta
Tietoliikennejatietoverkkotekniikanlaitos
Tekij:
TapioLev
Otsikko:
SkenaarioanalyysitulevaisuudenInternetist
Pivys:
15.5.2009
Sivumr:
6+94
Tiedekunta:
Elektroniikan,tietoliikenteenjaautomaationtiedekunta
Professuuri:
S38Tietoverkkotekniikka
Valvoja:
ProfessoriHeikkiHmminen
Ohjaaja:
DosenttiKaleviKilkki
Internet on kasvanut ulos alkuperisist raameistaan, samalla kun sen

yhteiskunnallinen merkitys on kasvanut. Parantunut tietoisuus Internetin
kohtaamista haasteista on aktivoinut ptksentekijt ympri maailmaa, ja lukuisia
hankkeita on perustettu tutkimaan tulevaisuuden Interneti. Kaikilla
tutkimusprojekteilla on kuitenkin haasteena, miten suunnata tutkimus trkeimpiin
asioihin.Tmdiplomityhelpottaaongelmaaskenaarioidenavulla,jotkatuovatesille
Internetin evoluutioon seuraavan kymmenen vuoden aikana vaikuttavat
merkittvimmttrenditjaepvarmuudet.
Ennen tulevaisuuden tutkimista perehdytn Internetin historiallisiin

virstanpylvisiin ja nykytilanteeseen kirjallisuustutkimuksen keinoin. Aivoriihiss ja
asiantuntijahaastatteluissa kerttyjen ajatusten ja tietojen avulla luodaan nelj
vaihtoehtoista Internetin evoluutioskenaariota kyttmll Schoemakerin
menetelm. PESTmallin kytt trendien ja epvarmuuksien tunnistamisessa
varmistaa, ett kaikki trket Internetin makroympristn tekijt huomioidaan.
LopuksiskenaarioitakytetnanalysoitaessasuomalaisenTulevaisuudenInternet
tutkimusohjelmantutkimusstrategiaa.
Skenaariot esittvt, ett Internetin kohtaamat haasteet voidaan ratkaista monella

tavalla, jotka johtavat erilaisiin verkko ja liiketoimintaarkkitehtuureihin.
Skenaarioiden vliset erot sidosryhmien voimasuhteissa ja arvonjakautumisessa
kertovat taustalla olevista jnnitteist ja sidosryhmien erivist intresseist.
Strateginen analyysi ehdottaa, ett uusien ratkaisujen kyttnotto pit suunnitella
huolellisestialustaalkaen.Kaikenkaikkiaantuloksetovatarvokkaita,sillneohjaavat
keskusteluatulevaisuudenInternetist.
Avainsanat:Internet;skenaarioanalyysi;verkkoarkkitehtuuri;arvonjakautuminen
Preface
This Masters Thesis has been written as a partial fulfillment for the Master of Science
degree in Helsinki University of Technology. The work has been carried out between
June 2008 and May 2009 at the Department of Communications and Networking as a
deliverablefortheFutureInternetprogramofTivit(FinnishStrategicCentreforScience,
TechnologyandInnovationinthefieldofICT),fundedbytheFinnishFundingAgencyfor
Technology and Innovation (Tekes), and coordinated within the Econ@Tel (COST605)
network.
Iwishtoexpressmygratitudetothepeoplewhohavesupportedmeinthiswork.First
ofall,IwouldliketothankProfessorHeikkiHmminenfortheopportunitytowritethis
thesis in his team and for his extensive guidance and insights throughout the research
process.IamgratefultoKaleviKilkkiforvaluableadvicesduringthewritingprocess.I
alsothankmyothercoworkersintheNetworkingBusinessteam,aswellastheFuture
Internet program partners and fellow researchers. Without the contribution and
enthusiasm of the interviewees and the participants of the brainstorming sessions this
researchwouldnothavebeenpossible.ThusIoweanespeciallylargewordofthanksto
theexpertsfortheircommentsandopinions.
Withoutmyfamily,fellowstudentsandfriendsitwouldhavebeenmuchhardertoget
here. Special thanks to my dearest, Hanna, for the strength and wisdom you gave me
duringthecourseofmywork.
Finally, I want to thank my parents, Marja and Jorma, for excellent upbringing and
supportduringmystudies.
Espoo,May15th,2009
TapioLev
Tableofcontents
Tableofcontents...............................................................................................................................I
Listoffigures...................................................................................................................................III
Listoftables.....................................................................................................................................III
Acronymsandterms......................................................................................................................IV
1
Introduction...............................................................................................................................1
1.1
Researchquestionandobjectives.....................................................................................................2
1.2
Scope...................................................................................................................................................................3
1.3
Researchmethods.......................................................................................................................................4
1.4
Structureofthethesis..............................................................................................................................5
Theoreticalframeworks........................................................................................................7
2.1
Scenarioplanning........................................................................................................................................7
2.2
PESTanalysis.................................................................................................................................................9
2.3
Brainstorming............................................................................................................................................10
2.4
Interviews.....................................................................................................................................................11
HistoricalmilestonesoftheInternetevolution..........................................................13
3.1
Divisionintoresearchandcommercialeras............................................................................13
3.2
Structureofthemilestonestudy.....................................................................................................14
3.3
Infrastructuremilestones....................................................................................................................15
3.4
Servicemilestones...................................................................................................................................21
Internetecosystem................................................................................................................27
4.1
Internetinterconnectivity...................................................................................................................27
4.2
Stakeholders................................................................................................................................................29
4.3
SizeoftheInternet..................................................................................................................................30
4.4
Trafficcharacteristics............................................................................................................................33
4.5
Presentproblems.....................................................................................................................................35
4.6
FutureInternetresearch.....................................................................................................................38
Scenarioconstructionprocess..........................................................................................41
5.1
Brainstorming............................................................................................................................................41
5.2
Keytrends.....................................................................................................................................................42
5.3
Initialkeyuncertainties........................................................................................................................46
5.4
Expertinterviews.....................................................................................................................................47
5.5
Finalkeyuncertainties..........................................................................................................................50
Scenarios...................................................................................................................................59
6.1
Wild&Free..................................................................................................................................................60
6.2
IsolatedWalledGardens......................................................................................................................61
6.3
ContentdrivenOverlays......................................................................................................................62
6.4
DeviceContentBundles.......................................................................................................................63
Comparisonofscenarios.....................................................................................................65
7.1
Architecture.................................................................................................................................................65
7.2
Valuedistribution....................................................................................................................................66
Fromscenariostoresearchstrategy...............................................................................69
8.1
ResearchplanoftheFutureInternetprogram.......................................................................69
8.2
Identifiedresearchtopics....................................................................................................................70
8.3
FutureInternetprogramandscenarios.....................................................................................72
Conclusion................................................................................................................................74
9.1
Keyfindings.................................................................................................................................................74
9.2
Discussion.....................................................................................................................................................76
9.3
Futureresearch.........................................................................................................................................76
References........................................................................................................................................78
Appendices.......................................................................................................................................87
A)Summaryofbrainstormingresults........................................................................................................87
II
Listoffigures
Figure1:Structureofthethesis...................................................................................................................................5
Figure2:Scenarioplanningprocess...........................................................................................................................9
Figure3:HourglassarchitectureoftheInternet................................................................................................18
Figure4:GeneralizedInternetecosystem.............................................................................................................28
Figure5:StakeholdersoftheInternetecosystem..............................................................................................30
Figure6:Internetdomainsurveyhostcount.......................................................................................................31
Figure7:Internetusersandpenetrationratesintheworldbygeographicregions..........................32
Figure8:Listofkeywordsusedtofeedthebrainstorming..........................................................................41
Figure9:Flipchartmatrixusedinevaluatingtheimportanceanduncertaintyofidentifiedforces...43
Figure10:Initialscenarioaxes..................................................................................................................................47
Figure11:U3Wherewilltheintelligencebelocated?..................................................................................55
Figure12:U4WhatwillbethedominatingbusinessmodelintheInterneteconomy?..................55
Figure13:U5Howwillsolutionsfortrust,securityandauthenticationbeimplemented?..........56
Figure14:U6Willthetrafficbetreatedneutral?............................................................................................56
Figure15:U7Amountofstandardization:standardsvs.proprietarysolutions?..............................56
Figure16:U8Wherewillthestandardizationhappen?...............................................................................57
Figure17:Scenariomatrix...........................................................................................................................................59
Figure18:Uncertaintiesvaluedinfivepointscale...........................................................................................60
Figure19:Simplifiedbusinessandtechnicalarchitecturesinscenarios.................................................65
Figure20:Simplifiedvaluedistributioninscenarios.......................................................................................67
Figure21:WorkpackagesandcrossissuesofFIProgram...........................................................................70
Figure22:Explanationofthegroupsofidentifiedforces...............................................................................87
Listoftables
Table1:Evaluationcriteriaofthemilestones.....................................................................................................14
Table2:Infrastructuremilestones...........................................................................................................................15
Table3:Servicemilestones.........................................................................................................................................21
Table4:ProtocoltrafficdistributioninOC192backbonelinkfromSeattletoChicago....................34
Table5:FutureInternetresearchefforts..............................................................................................................40
Table6:Keytrends.........................................................................................................................................................42
Table7:Listofinterviewees.......................................................................................................................................48
Table8:Finalkeyuncertainties.................................................................................................................................50
Table9:U9Possiblecausesforcollapse.............................................................................................................58
Table10:Importanceofthechosenresearchtopicsinthescenarios.......................................................71
III
Acronymsandterms
3G
3GPP
4G
AdobeFlash
AS
BGP
CAIDA
CDN
CERN
CIDR
CSC
(D)ARPA
DCCP
DHCP
DNS
DoS
EGP
ESP
EU
FDM
femtocell
FI
FIA
FP7
FTP
GRE
GSM
HIIT
HIP
HMTL
HTTP
ThirdGeneration
theThirdGenerationPartnershipProject
FourthGeneration
A multimedia platform used to create animation, advertisements, and
various web page components, to integrate video into web pages, and to
developrichInternetapplications.
AutonomousSystem
BorderGatewayProtocol(RFC4271)
CooperativeAssociationforInternetDataAnalysis
ContentDeliveryNetwork
ConseilEuropenpourlaRechercheNuclaire(EuropeanOrganizationfor
NuclearResearch)
ClasslessInterDomainRouting(RFC4632)
ITCenterforScience
(Defence)AdvancedProjectsResearchAgency
DatagramCongestionControlProtocol(RFC4330)
DynamicHostConfigurationProtocol(RFC2131)
DomainNameSystem(RFC1035)
DenialofService(attack)
ExteriorGatewayProtocol(RFC904)
EncapsulatingSecurityPayload(RFC4303)
EuropeanUnion
FrequencyDivisionMultiplexing
A small cellular base station designed for use in residential or small
business environments. It connects to the service providers network via
broadband.
FutureInternet
FutureInternetAssembly
SeventhFrameworkProgramme
FileTransferProtocol(RFC959)
GenericRoutingEncapsulation(2784)
Global System for Mobile communications (originally Groupe Spcial
Mobile)
HelsinkiInstituteforInformationTechnology
HostIdentificationProtocol(RFC4423)
HyperTextMarkupLanguage(RFC1866)
HyperTextTransferProtocol(RFC2616)
IV
HTTPS
HW
ICANN
ICMP
ICT
IDN
IETF
IGP
IM
IMS
IoT
IP
IPR
IPsec
IPv4
IPv6
ISC
ISP
ITU
LAN
MP3
NAT
NIC
NSF
NCP
OECD
P2P
P2PTV
PEST
PSTN
QoS
R&D
RFC
RSVP
RTCP
RTP
RTSP
HyperTextTransferProtocolSecure(RFC2818)
Hardware
InternetCorporationforAssignedNamesandNumbers
InternetControlMessageProtocol(TFC792)
InformationandCommunicationsTechnologies
InternationalizedDomainNames
InternetEngineeringTaskForce
InternetGovernanceProject
InstantMessaging
IPMultimediaSubsystem
InternetofThings
InternetProtocol(RFC791)
IntellectualPropertyRights
InternetProtocolSecurity(RFC4301)
InternetProtocolversion4
InternetProtocolversion6
InternetSystemsConsortium
InternetServiceProvider
InternationalTelecommunicationUnion
LocalAreaNetwork
MPEG1AudioLayer3
NetworkAddressTranslator(RFC3022)
NewlyIndustrializedCountry
NationalScienceFoundation
NetworkControlProtocol(RFC36)
OrganizationforEconomicCooperationandDevelopment
PeertoPeer
PeertoPeerTelevision
Political, Economic, Social, Technological (framework for environmental
scanning)
PublicSwitchedTelephoneNetwork
QualityofService
Research&Development
RequestForComments
ResourceReSerVationProtocol(RFC2205)
RTPControlProtocol(RFC3550)
RealtimeTransportProtocol(RFC3550)
RealTimeStreamingProtocol(RFC2326)
SCTP
SIP
SMS
SMTP
SRA
SRI
SSH
SSL
SW
SWOT
TCP
telco
TDM
TLS
Ubicomp
UCLA
UDP
UI
UN
URI
URL
U.S.
VoIP
VPN
W3C
WDM
WWW
WLAN
XML
StreamControlTransmissionProtocol(4960)
SessionInitiationProtocol(RFC3261)
ShortMessageService
SimpleMailTransferProtocol(RFC5321)
StrategicResearchAgenda
StanfordResearchInstitute
SecureShell(RFC4251)
SecureSocketLayer
Software
Strengths,Weaknesses,Opportunities,Threats(SWOTAnalysis:astrategic
planning method used to evaluate the Strengths, Weaknesses,
Opportunities,andThreatsinvolvedinaprojectorinabusinessventure)
TransmissionControlProtocol(RFC793)
telephonecompany
TimeDivisionMultiplexing
TransportLayerSecurity(RFC5246)
Ubiquitouscomputing
UniversityofCalifornia
UserDatagramProtocol(RFC768)
UserInterface
UnitedNations
UniversalResourceIdentifier(RFC3986)
UniformResourceLocator(RFC1738)
UnitedStatesofAmerica
VoiceoverInternetProtocol
VirtualPrivateNetwork
WorldWideWebConsortium
WavelengthDivisionMultiplexing
WorldWideWeb(alsoWebandW3)
WirelessLocalAreaNetwork
eXtensibleMarkupLanguage
VI
1 Introduction
TheimportanceoftheInternetforthesocietyisconstantlyincreasing.Infourdecades
the Internet has grown from a network of computer science researchers to a global
backbone of the information society, and currently over one billion people use it to
communicate,searchandshareinformation,conductbusiness,andenjoyentertainment.
The Internet community lead by IETF (Internet Engineering Task Force) has been
remarkably successful in solving scalability bottlenecks caused by novel application
requirements and surprising growth in user base. Nevertheless, the Internet and its
architectural principles (Carpenter, 1996) were designed in the 1970s mostly for
purposes that resemble very little the current and foreseen usage scenarios. Thus the
Internet community is questioning the ability of the Internet to cope with the
forthcomingchallenges.
IncreasedawarenessofthechallengestheInternetisconfrontingandofthepossibilities
it is offering has activated decisionmakers and researchers around the world. Various
initiativesforexampleinEurope,U.S.,Japan,andSouthKoreahavebeenestablishedto
studytheFutureInternet.Theseresearcheffortsareseenimportantintheglobalspace
but also local interests are high since governments and regions try to maximize their
future competitiveness through strategically wellaimed investments to the Future
Internet. For example Vivian Reding, European Union Commissioner for information
societyandmedia,recentlystatedthatEuropemustbeakeyplayerinthefutureofthe
Internet (Reding, 2009). Consequently the financial investments are high; for instance
EuropeanUnionmemberstateshavealreadycommittedover9.1billionoffundingfor
ICT (Information and Communications Technologies) research (European Commission,
2008).
InvestingintheFutureInternetcontainshighlevelofrisksbecausetheresearchersare
missing the crystal ball that would tell which are the most relevant issues to research.
Besides, recognizing the technical shortcomings is not enough since the Internet
evolution is affected also by political, economic and social forces that need to be
understood so that technical solutions are accepted and adopted successfully.
Additionally,thevaryingstakeholderinterestsandincentivesaswellastheinertiaofthe
Internet caused by its sheer size need to be taken into account when planning the
deploymentofnewtechnicalsolutions.
Creating scenarios is one way to deal with the complex uncertainties related to the
evolution of the Internet ecosystem. For instance Future Internet Assembly (FIA) has
asked its crossdomain working groups to provide scenarios for the Future Internet to
directthediscussionandallocaterelatedfunding(Silva&Campolargo,2009).Thisthesis
anditsscenariowork,forone,areconductedintheframeoftheFinnishnationalFuture
Internet (FI) program which is aimed to bring together the key research resources to
develop Future Internet networking technologies and create new global ICT based
business ecosystems. The participants of the project represent broadly the Finnish
telecommunicationsindustrylandscapeincludingbothindustrialpartners:TheFinnish
IT center for science (CSC), Ericsson, Nokia, Nokia Siemens Networks, and Sonera; and
research partners: Helsinki and Tampere Universities of Technology, Universities of
Helsinki and Jyvskyl, Helsinki Institute for Information Technology (HIIT), and VTT
Technical Research Centre of Finland. The program is a part of the ICT cluster of the
Finnish Strategic Centres for Science, Technology and Innovation (TIVIT / ICT SHOK)
fundedbytheNationalTechnologyAgencyofFinland(TEKES).
1.1 Researchquestionandobjectives
WhiletheInternetisbecomingmoreandmoreintegralpartofthesocietytheinterestof
variousstakeholdersisincreasing.Thesestakeholderstrytogetcompetitiveadvantage
through strategically wellaimed research efforts. However, due to the size and
complexityoftheInternetecosystemdirectingexpensiveresearchtotherighttargets
isdifficult.Thusdesireforunderstandingandboundingtheuncertaintiesrelatingtothe
futuredevelopmentofInternetincreases.Thisthesistriestoeasetheproblemthrough
scenarios which disclose the most significant trends and uncertainties. The main
researchquestionsareexpressedasfollows:
Whicharethealternative(technological)scenariosforInternetover10years?
Whatarethekeytrendsanduncertaintiesthatproducethesescenarios?
ThekeystakeholderinthisstudyistheFinnishFutureInternetprogram.Theanswerto
the main research question is used in analyzing the strategic research agenda of the
program.Thisisformulatedinasupportingstrategicquestion:
WhatshouldbetheIETFresearchstrategyofTIVIT/FIprogramtocopewitheach
scenario?
To make the problem more convenient to handle, a handful of objectivesis recognized
andachievedduringtheresearch:
Identify historical milestone technologies and link them to IETF RFCs (Request
forComments).
Identify key trends and uncertainties of the Future Internet evolution by

organizingbrainstormingsessionsandexpertinterviews.
Createfourscenariospresentingalternativefutures.
Analyze and compare scenarios from the perspective of technical and business
architecture,andvaluedistribution.
Analyze the applicability of the strategic research agenda of Finnish Future

Internetprograminthescenarios.
1.2 Scope
Theresearchquestionisquitebroad,andsomefocusingisneededtomaketheproblem
more convenient to handle. The most important question is explaining the meaning of
thewordInternetthatistypicallydefinedveryvaguely.U.S.FederalNetworkingCouncil
definitionfromtheyear1995isusedasthebasisforourowndefinition:
Internet refers to the global information system that (i) is logically linked
togetherbyagloballyuniqueaddressspacebasedontheInternetProtocol(IP)or
its subsequent extensions/followons; (ii) is able to support communications using
the Transmission Control Protocol/Internet Protocol (TCP/IP) suite or its
subsequent extensions/followons, and/or other IP compatible protocols; and (iii)
providesormakesaccessible,eitherpubliclyorprivately,highlevelserviceslayered
onthecommunicationsandrelatedinfrastructuredescribedherein.(FNC,1995)
ThetwofirstcriteriaIPaddressspaceandTCP/IPsuitearepurelyarchitectural.The
thirdpointconsideringservicespresentstheusageperspective.Inthisthesistheseboth
aspects are taken into account. From technical perspective, the scope is only on the IP
layer and the layers above that, which means that data link and physical layers are
mostlyneglected.Likewise,differentaccessmethodsarenotstudiedinthisthesissince
theconcentrationismoreonthecorenetwork.Inadditiontoatechnicalpointofview,
theInternetisalsounderstoodasaphenomenonhavingeconomic,regulatoryandsocial
implications.
The time frame is a ten years period from 2009 to 2018. Although the scenarios are
studiedintheglobalspace,thestrategicimplicationsofresultingscenariosarediscussed
onlyfromtheperspectiveoftheFinnishFutureInternetprogram.Thefocusofstrategic
analysisisfurtherlimitedtoresearchstrategy,andespeciallytoIETFwork.
1.3 Researchmethods
Theresearchmethodsappliedinthisthesisare
literaturesurvey,
brainstorming,
interviews,and
analysisbasedontheoreticalframeworks.
Literature survey is used in analyzing the most important historical milestones in the
InternetevolutionandingettinggoodunderstandingofthecurrentstateoftheInternet
ecosystem. The chosen approach is qualitative and written sources include academic
publications, IETF RFCs, technical and management books, relevant future forecasts,
newsarticles,andwhitepapers.
After attaining sufficient grounding, brainstorming sessions and interviews are
organized to broaden the view. The participants of brainstorming sessions and
interviewscomefromFinnishacademiaandtelecommunicationsindustry,andmostof
them are closely related to the Finnish Future Internet program. The results from
brainstorming and interviews act as the basis during scenario construction and thus
havegreatinfluenceonthisthesis.Eventually,fewtheoreticalframeworksarecombined
withtheseresultsandtheyinclude
scenarioplanning,and
PESTframework.
1.4 Structureofthethesis
ThestructureofthethesisispresentedinFigure1.
1.Introduction
2.Theoreticalframeworks
3.Historicalmilestoneanalysis
Brainstorming
4.Internetecosystem
5.Scenarioconstruction
Interviews
6.Scenarios
7.Comparisonofscenarios
8.Fromscenariostostrategy
9.Conclusion
Figure1:Structureofthethesis
Thisfirstchapterintroducesthethesistothereader.ThenChapter2explainstheoretical
frameworks, including the scenario planning method and PEST analysis. Additionally,
theorybehindbrainstormingandinterviewsasdatacollectionmethodsisdescribed.
The IETF and RFCcentric view on historical milestones of the Internet is presented in
thethirdchapterandthenatureandchallengesofthecurrentInternetecosysteminthe
fourthchapter.
TheprocessofcreatingscenariosispresentedinChapter5.Thisincludesapplicationof
research methods (brainstormingand expertinterviews) as well asdescriptions of the
keytrendsanduncertainties.
Chapters6and7describethederivedscenariosfirstonebyoneandthencomparedwith
each other. After that in the eighth chapter, the scenarios are used in analyzing the
researchstrategyoftheFinnishFutureInternetprogram.
Finally,Chapter9concludesthefindingsandsuggestssometopicsforfurtherresearch.
2 Theoreticalframeworks
Thischapterintroducesthetheoreticalframeworksandresearchmethodsthatareused
inthisthesis,andexplainshowtheyrelatetoeachother.
2.1 Scenarioplanning
Scenario planning is an established tool for exploring complex situations with high
uncertainty. Modern scenario techniques stem from war game simulations at the Rand
Corporationinthe1950s(Schoemaker,1993).Inthe1970stheywereusedsuccessfully
inthepetrochemicalindustry(Wack,1985)tocopewiththeoilcrisis.
More recently, scenario planning has been used in dealing with high uncertainty of
emerging technologies in the ICT field. Karlson et al. (2003) took a holistic view and
createdfourpossiblescenariosfortheevolutionofwirelessindustryfrom2003to2015.
Nordlundetal.(2007)usedKarlsonetal.smethodtocreatescenariosfordigitalhome
management.Heikkinen(2008ab)hasusedscenariostounderstandtheusageofmobile
peertopeer services. Additionally, Smura and Sorri (2009) have studied the wireless
local area access market concentrating on indoor access and especially on rivalry
betweenWLANs(wirelesslocalareanetworks)andfemtocells.
Although practitioners have developed scenario planning to several directions during
thepastdecades(seecomparisoninBradfieldetal.,2005),theyallbuildonidentifying
drivingforcesconsistingofbothpredeterminedanduncertainelements.Predetermined
elements,oftencalledastrends,describethecollectiveknowledgeoftheindustry.They
do not depend on any particular chain of events nor scenario come to pass, and thus
applyinallthescenarios(Schwartz,1998).Uncertainelements,orsimplyuncertainties,
areforcesdeemedimportantbutwhoseoutcomesarenotverypredictable(Schoemaker
&Mavaddat,2000).Thustheycanbedescribedasvariablesorasthingsweknowwe
dont know. When uncertainties are studied, the interdependencies and relationships
betweenthemareofhighimportance,sincenotallcombinationsmayoccur.Especially
important is finding of the root causes or independent uncertainties that can then be
usedasbasisforseparatingthescenarios.
Van der Heijden (1996) lists following principles that can be taken as guidelines in
scenarioplanning.
Therehastobeatleasttwoscenarios.Morethanfourisoftentoocomplex.
Eachscenariomustbeplausible.
Thescenariosmustbeinternallyconsistent.
Thescenariosmustberelevanttotheissuesthatarebeingresearched.
Thescenariosmustprovidenewideasandinsightsusableinstrategicplanning.
If the scenario planner conforms to these rules, he can freely choose the scenario
constructionmethodandpresentationformat.
Frommanyalternatives,Schoemakersmethodwasseenasthemostsuitableoneforthis
study (Schoemaker, 1991, 1993, 1995; Schoemaker & Mavaddat, 2000). Schoemaker
(1993) defines scenarios as focused descriptions of fundamentally different futures
presentedincoherentscriptlikeornarrativefashion.Theysimplifytheavalancheofdata
into a limited number of possible states (Schoemaker, 1995). Nevertheless, scenarios
shouldnotbetreatedasforecastsbutratherasmeansforboundingandunderstanding
futureuncertainties.Ifsuccessfullyusedscenarioplanningcanpreventtunnelvisionby
revealing hidden or weak signals and by stimulating decision makers to consider
changestheywouldotherwiseignore(Schoemaker,1995).
The tenstep framework presented in (Schoemaker & Mavaddat, 2000) was used
although the last two steps requiring quantitative modeling were not carried out. The
method is summarized in Figure 2. First, the scene and scope is set by defining time
frame, scope and decision variables. Also major stakeholders need to be identified.
Second,keytrendsanduncertaintiesareidentified.Third,fourinternallyconsistentand
plausible scenarios are constructed based on the most important uncertain elements.
Finally,thestakeholderbehaviorintheresultingscenariosisassessed.Afterthescenario
process is completed the scenarios are typically used in planning strategic actions of a
givenmarketplayer.
1.Settingthesceneandscope
Definetimeframe,scope anddecisionvariables.Identifymajorstakeholders.
2.Identifyingkeytrendsanduncertainties
Keytrends=importantforceswhoseconsequenceshavenotyetunfolded.
Keyuncertainties =importantforceswhoseoutcomesarenotverypredictable.
3.Scenarioconstruction
Selecttwomostimportantkeyuncertainties scenariomatrix.
Addimpactofotherkeyuncertaintiesandtrends.
Assessinternalconsistency andplausibility,revise.
Assessstakeholderbehaviour.
4.Quantitativemodelling
Figure2:Scenarioplanningprocess(Schoemaker,1993;Schoemaker&Mavaddat,2000)
2.2 PESTanalysis
AcronymPESTstandsforPolitical,Economic,SocialandTechnological,anddescribesa
framework of macroenvironmental factors used in the environmental scanning
component of strategic management. The components of the acronym are sometimes
reorderedtoSTEP,andalsonewcomponentslikeLegalandEnvironmentalareaddedto
formPESTLE.
Tools and techniques for environmental scanning were firstly discussed by Aguilar
(1967)whodefineditasaprocessthatseeksinformationabouteventsandrelationships
inacompanysoutsideenvironment,theknowledgeofwhichwouldassisttopmanagement
initstaskofchartingthecompanysfuturecourseofaction(Fahey & King, 1977). This
externalenvironmentcanbedividedintotheoperatingenvironmentthesuppliersand
other interest groups, with which the firm deals, and the general environment the
national and global context of social, political, regulatory, economic and technological
conditions (Thomas, 1974). As can be understood, PEST analysis concentrates on the
generalenvironment.
In the academic literature environmental scanning and PEST analysis are not widely
covered.PESTismainlytakenasapracticaltoolforthecompaniestopmanagementto
be used for example in conjunction with SWOT analysis to help defining opportunities
and threats. Thomas (1974) describes the situation by saying that the idea of taking
PEST factors into account is somewhatinthenatureofconventionalwisdom, and that
PEST conditions are almostrituallyinvokedinplanningliterature. Thus PEST analysis
canbeusedtoperceivethesurroundingworldalsoinothersituationsthaninplanning
corporatestrategy.
TheInternetisnotanymoreapieceoftechnologyinvacuumbutitisalsoaffectedbythe
political,socialandeconomicinterests.UsageofPESTanalysisoffersbroadviewtothese
forcesthatmayaffectInternetevolutioninthefuture.Thusitisusedinidentifyingthe
keytrendsanduncertaintiesduringthebrainstormingsessionsandexpertinterviews.
2.3 Brainstorming
Rickards (1999) defines the brainstorming as an individual or group process of idea
generationfollowingstructuralguidelines forweakeningintrapersonalandinterpersonal
barrierstothegenerationofnewandusefulideas.Brainstormingisexploitedtoenhance
creativity and generate a large number of ideas. Practical evidence shows that
brainstorming leads to behavioral and ideational gains over outputs of conventional
individualandgroupwork(Rickards,1999).
The modern applications of brainstorming are associated with Alex Osborn who
popularizedthemethodinthelate1930s(Rickards,1999).Osborn(1963)hadtwokey
concepts, the principle of deferment of judgment and the principle of extended search,
basedonwhichhedefinedthefourgroundrulesforthebrainstorming:
1. Focusonquantity.
2. Withholdcriticism.
3. Welcomeunusualideas.
4. Combineandimproveideas.
10
Theserulesareusedinabrainstormingsessionlastingfromcoupleofminutestoseveral
hours.Thestructureforthesessionisgivenbythefacilitatorwhomotivatesandsteers
thebrainstorminggrouptotherightdirection.
The variety of brainstorming techniques can be divided into two broad groups
interactive and nominal. In interactive brainstorming participants interact during the
brainstorming while in nominal brainstorming the interaction is inhibited and
participants create ideas in isolation. Interactive brainstorming has various defects
including distraction, social loafing, evaluation apprehension, and production blocking,
whichleadtoproductivitylosses(Diehl&Stroebe,1991).Thusnominalgroupsaremore
effectiveintermsofproductivity,andtheproductivitycanbefurtherenhancedbyusing
electronic support systems (Rickards, 1999). However, productivity and efficiency are
not the only aspects (Rietzschel et al., 2006). Participants may favor social and
interactive,butlessefficientmethodswhichenablesharingofideas.Fromthepractical
point of view, the potential difficulty of choosing between interactive and nominal
methodscanbesidesteppedbyusingacocktailoftechniques(Rickards,1999).
2.4 Interviews
Interview is a very widely used research method which involves a researcher
(interviewer) asking questions and receiving answers from the people she is
interviewing(interviewee).Interviewsareconductednormallyfacetofaceandoneto
one,butalsotelephoneinterviewsaswellasgroupinterviewsarepossible.Interviews
are typically divided into three classes based on the degree of structure and
standardization (Robson, 2002). 1) Fully structured interview has predetermined
questionswithfixedwording,usuallyinpresetorder.2)Semistructuredinterviewalso
haspredeterminedquestionsbuttheorderandwordingofthemcanbechanged,some
questions can be omitted and new ones added, and the interviewer can give
explanations.3)Unstructuredinterviewsaremostinformal,theyhaveatheme,ageneral
areaofinterest,andtheconversationisfreeandcandeveloptoanydirection.
Here the concentration is on the latter two (semistructured and unstructured) which
King (2004) refers as qualitative research interviews. Kvale (1983) gives a formal
11
definition for qualitative research interview as aninterview,whosepurposeistogather

descriptions of the lifeworld of the interviewee with respect to interpretation of the
meaningofthedescribedphenomena. Qualitative research interviews are characterized
bythelowdegreeofstructure,preponderanceofopenquestions,andfocusonspecific
situations in the world of interviewee (King, 2004). Thus they concentrate more on
intervieweesopinionsthanabstractionsorgeneralopinions.
The interviewerinterviewee relationship plays a key role in success of an interview
(King, 2004). When structured interview tries to minimize the impact of interpersonal
factors, in qualitative interviews they are taken as an inseparable part of the research
process.Hencetheintervieweeisseenratherasaparticipant,notasaresearchsubject,
and she participates actively in shaping the course of the interview. Recruiting
intervieweesisanessentialstepwhichaffectsthequalityandvarietyofresults.Inorder
to cover the studied subject as wide as possible, the amount of interviewees and
diversityintheirbackgroundsisdesirable.Practicalreasons,mainlyamountoftimeand
resources,however,restrictthescopeoftheinterviewstudy.
Robson(2002)listssomeadvantagesanddisadvantagesofinterviewstudy.Onthepros
side is that interviews are flexible and adaptable to many problems, and they have
potentialofprovidingrichand highlyilluminatingmaterial.Itispossibletomodifythe
question, ask followup questions, and investigate motives. Facetoface contact also
gives an experienced interviewer a possibility to follow nonverbal cues. On the other
hand,thelowlevelofstructureandstandardizationraisequestionsaboutthereliability
and repeatability of the interviews. Interpersonal factors mean that biases cannot be
ruledoutandthattheskillsandpersonalityoftheinterviewerimpactonthequalityof
results.Furthermore,interviewing,includingalsopreparationandtranscribingofnotes,
isverytimeconsumingandthuslimitsthesamplesize.
12
3 HistoricalmilestonesoftheInternetevolution
TheInternetwasbornintheOctober29th,1969whenfirstpacketsweresentbyCharley
KlineatUniversityofCalifornia(UCLA)ashetriedtologintoStanfordResearchInstitute
(SRI) (Zakon, 2006). In the following four decades the Internet has evolved from a
researchprojecttoaworldwidecommunicationsnetworkhavinghugeeconomicvalue.
TheconductedmilestonestudytriesnottodepictthecompletehistoryoftheInternet 1
butithighlightsthemostimportantevolutionsteps.
3.1 Divisionintoresearchandcommercialeras
Internet evolution can be divided into the research era and the commercial era. The
foundationsoftheInternetweredesignedintheresearcheralastingfromthe60stothe
beginningofthe90s.Theinventionofpacketswitchingtheory(Kleinrock,1961;Baran,
1964) and global networking concept (Licklider & Clark, 1962) were major steps
towards computer networking and the Internet. In thenext years the plan for the first
version of the Internet called ARPANET (Roberts, 1967) was developed within the
computerresearchprogramatARPA 2 .Aftertheinitialnetworklaunchconsistingoffour
nodes,thedevelopmentoftheInternetcontinued.For20yearstheInternetwasmainlya
toolforcomputerscienceresearchers,anditwasmostlyusedindistributedcomputing
byloggingremotelytothehostsandrunningcommandsonthem(Leineretal.,2003).
PrivatizationoftheInternetbackboneenabledcommercializationoftheInternet.Atthe
same time Tim BernersLee invented Web which brought the Internet available to
everyman. These two parallel milestones work as a divide between research and
commercialeras.InthecommercialeratheInternethasexperiencedexponentialgrowth
in the amount of hosts, users, networks, traffic and services. The foundations have not
changedbuttheresearchconcentrationhasshiftedtosolvingscalabilitybottlenecksand
innovating new services and usages. The most new users have been nonprofessional,
andtheInternethasevolvedwaybeyonditsoriginalintention.
1For
example Hobbes Internet Timeline (Zakon, 2006) and ISOCs link collection (ISOC, 2009) offer
extensiveamountofinformationaboutthehistoryoftheInternet.
2The Advanced Research Project Agency of U.S. Department of Defense (DoD), renamed to DARPA (for
Defense)in1972.
13
3.2 Structureofthemilestonestudy
Studyingmilestonesismotivatedbytheneedtounderstandthehistoricaldevelopment
of the Internet. The term milestone covers important principles, seminal technologies
andprotocols,andrevolutionaryservices.DuetothestrategicquestionconcerningIETF,
thestudymapsthemilestonestoIETFstandardscalledRequestsforComments(RFCs).
Identified milestones can be divided into two broad categories, infrastructure and
servicemilestones.Infrastructuremilestones(Chapter3.3)areeitherfundamentsofthe
Internet or protocols that removed scalability bottlenecks, while service milestones
(Chapter3.4)areimportantapplicationsthatcreateddemand,temptednewusers,and
increasedtraffic.
TheselectionofmilestoneswasdonebasedonfewevaluationcriteriapresentedinTable
1 below. All the applicable criteria were graded in the threelevel scale largemedium
small,dependingonhowmuchamilestoneaffectedoneachevaluationcriteria.Someof
the effects happened before the others, for example DNS primarily enabled larger
numberofhostsbutlaterdomainnameshadsignificanteffectoncommercializationof
theInternet.Thustheprimaryeffectisunderlinedtoemphasizeitsimportance.
Table1:Evaluationcriteriaofthemilestones
Evaluationcriterion
Explanation
Numberofusers
Increasedtheamountofusers=createddemand.
Numberofhosts
Enabledincreasingofthenumberofhosts=scalability.
Amountoftraffic
Increasedtheamountoftraffic.
Numberofservices
Increasedthenumberofservicesandinnovations.
Economicimpact
Hadlargeeconomicimpactforsociety.
Timeconsumption
IncreasedusersInternetusage.
Changeinusagepatterns
ChangedpeoplesInternetusagepatterns.
The milestones are gathered in Table 2 and Table 3. A short description, year of the
introduction,mostimportantoriginalRFC,andthegradingbasedonevaluationcriteria
arepresentedforeverymilestone.
14
3.3 Infrastructuremilestones
When looked from the infrastructural (or supply) side the Internet has been a very
successfulscalabilitystory.Generallyspeaking,inframilestonescanbedividedintothe
foundations of the Internet dating back to the research era (the first four) and the
protocols that enabled increasing user base (the last five). All the infrastructure
milestonesarelistedinTable2belowanddescribedmoredeeplyinthenextsections.
Table2:Infrastructuremilestones.
Commercialera
Researchera
Milestones
ThemostinfluentialRFC
Evaluationcriteria
Nr.of
users
Nr.of
hosts
Amountof
traffic
Econ.
impact
Nr.of
services
Timecon
sumption
Changein
usage
patterns
Large
Large
Large
Large
Large
Description
Year
FirstRFC
Nr.(year)
RFCProcessRoughconsensus
andrunningcode
1969
1310(92)
TCP/IP
1974
675(74),
TCP,IP
791(81)
Endtoendargument
1981
1958(96)
TransitionfromHOSTS.txtto
DNS
1984
883(83),
DNS
884(83)
Medium
Large
Medium
Large
PrivatizationoftheInternet
backbone
1995
1105(89) BGP
Large
Large
CIDRenablesroutingtable
scalability
1993
1519(93) CIDR
Large
Medium
DHCPenablesdynamicIP
addressallocation
1993
1531(93) DHCP
Large
Medium
Small
NATalleviatesIPaddress
shortage&improvessecurity
1994
1597(94)
Medium
Large
Medium
Name
TheInternetStandards
Process
ArchitecturalPrinciples
oftheInternet
AddressAllocationfor
PrivateInternets
3.3.1 RFCprocessroughconsensusandrunningcode
TheInternetstandardizationprocessbasedoninformaldocumentscalledRFCsisasold
as the Internet. The first RFC titled Host Software was published on 7 April 1969 by
SteveCrocker.Itscontentsweremodestandforgettablebutitwastheinitiativethatwas
significant (RFC Editor et al., 1999). From that day on the idea of the RFC process has
been to be a fast distribution way for sharing ideas within the Internet community.
Hence all the RFCs are not official protocol specifications (although over time they are
more focused on them) but some of them are informational, and describe alternate
approachesorprovidebackgroundinformation.
OneofthekeystotherapidgrowthoftheInternethasbeenthefreeandopenaccessto
RFCs, which promotes the innovation because it allows the actual specifications to be
15
usedbyanyone,forexamplebyuniversitystudentsandentrepreneursdevelopingnew
systems (Leiner et al., 2003). Anyone can also submit a document to be an RFC but
typicallytheyaregeneratedbyIETF.OpennatureoftheRFCprocessisalsopresentin
IETF which is an open organization of individuals. Anyone can participate in the
meetingsandcontributetothework.Thestandardizationisbasedonroughconsensus
and running code meaning that a protocol needs to be widely accepted by the
community and its functioning needs to be proved by working applications before the
finalversionofanRFCispublished(Alvestrand,2004).
3.3.2 Endtoendargument
ThebasicdesignprincipleoftheInternetisknownastheendtoendargumentandwas
phrasedbySaltzer,ReedandClark(1984)asfollowed:
The function in question can completely and correctly be implemented only with
the knowledge and help of the application standing at the endpoints of the
communicationsystem.Therefore,providingthatquestionedfunctionasafeatureof
thecommunicationsystemitselfisnotpossible.
The key idea is that a network should do as little as possible, just transmit packets as
efficiently and flexibly as possible, and everything else, including error detection and
correction, reordering of packets, and encryption, should be left at the fringes
(Carpenter, 1996). This approach of a dumb network and smart end points was
revolutionary when introduced because of prevailing architecture in which it was the
networkssacrosanctresponsibilitytodoeverythingpossibletoensurethatitdoesnot
drop data (Huston, 2008a). The endtoend argument leads to the best effort traffic
patternandthenetworkneutralityprincipleofequalityofanykindoftraffic.Italsohas
hiddenassumptionofmutualtrustbetweenendpoints(Clark&Blumenthal,2001).
The endtoend argument has important impact on network architecture. It decreases
the complexity of networks, which reduces costs and facilitates future upgrades. From
the application perspective a dumb network allows new applications to be added
without changing the core, and these applications stay independent of implementation
and successful operation of applicationspecific services in the network (Clark &
16
Blumenthal, 2001). Thereby anyone can write a communications application, share it

withpeopleandstartusingit.Onthecontrary,smartnetworksliketelephonenetwork
inhibit this kind of behavior. Thus the most important impact of the endtoend
argumentisitsabilitytofosterinnovation(Isenberg,1998).
Although the endtoend argument is without a doubt one of thekeys to the Internets
success, it has also been criticized. For instance Moors (2002) as well as Clark and
Blumenthal(2001)statethattheendtoendargumentisnotappropriateineveryplace
anditshouldnotbetakenastheonlychoice.Somefunctionscanonlybeimplementedin
thecoreofthenetwork,andalsoperformancerelatedissuescandriveforcorelocated
features. Additionally, some of todays problems in the Internet stem from the original
design decisions and the endtoend argument which were made mutually trusting
researchcommunityinmind.
3.3.3 TCP/IP
TheInternetprotocolsuite,TCP/IP,formsthecoreoftheInternet.Thedefinitionofthe
InternetpresentedinSection1.2actuallyreliessolelyontheseprotocols.Nevertheless,
they havenot been present from the beginning. Theinitial hosttohost protocol called
NetworkControlProtocol(NCP)(Crocker,1970)wasuseduntilthetransitiontoTCP/IP
was executed on January 1, 1983 (Zakon, 2006). The development of TCP/IP started
already in the early 70s and the original paper presenting TCP was published in 1974
(Cerf & Kahn, 1974). At that time TCP and IP were bundled to one protocol which
allowedonlyvirtualcircuitstyleofcommunicationintheInternet.Someapplications,for
example early work on packet switched voice communication, revealed that in some
cases more unreliable datagram service would be needed. Thus in 1978 TCP was
reorganized into two protocols, TCP and IP, and User Datagram Protocol (UDP) was
introducedinordertoprovidedirectaccesstothebasicserviceofIP.
Since TCP and IP are fundamental pieces of the Internet, their design principles have
fundamentaleffectonthenatureoftheInternet.Whentheyweredesigned,thetoplevel
goal was to develop effective technique to utilize existing interconnected networks
17
(Clark, 1988). Clark also lists seven second level goals presented in the order of
importance:
1. Internetcommunicationmustcontinuedespitelossofnetworksorgateways.
2. TheInternetmustsupportmultipletypesofcommunicationsservice.
3. TheInternetarchitecturemustaccommodateavarietyofnetworks.
4. TheInternetarchitecturemustpermitdistributedmanagementofitsresources.
5. TheInternetarchitecturemustbecosteffective.
6. TheInternetarchitecturemustpermithostattachmentwithalowlevelofeffort.
7. TheresourcesusedintheInternetarchitecturemustbeaccountable.
ThehourglassstructureoftheInternetprotocolsuite(Figure3)summarizesthepoints2
and 3 nicely. TCP/IP does not restrict the network technology or the applications that
canbeusedintheInternet.Thisflexibilityhasallowedtheinnovationandenabledcost
savings because no dedicated network was needed to build since various existing
networks, most notably PSTN (Public Switched Telephone Network), could be used.
Later, when Internet usage has increased, also the economies of scale have had an
importanteffectinloweringthecosts.
Figure3:HourglassarchitectureoftheInternet(adaptedfromZittrain,2008).
18
3.3.4 TransitionfromHOSTS.TXTtoDNS
Domain name system (DNS) introduced in1983 andtakeninto use in 1987 solved the
scalability problems of HOSTS.TXT system (Levien, 2005). HOSTS.TXT was a single file
containing all host name to address mappings which was FTPed by all the hosts
(Mockapetris,1987).Thuseverychangeinthenetwork,forexampleaddinganewhost,
requiredupdatingofthefileandsendingtheentiretabletoallthehosts.Withtimewhen
increasing number of computers joined the Internet, the updating task became more
burdensomeandsuspecttofailure,theprocesswastooslowandnameconflictsstarted
to occur due to the flat name space (Levien, 2005). Most importantly, the HOSTS.TXT
system formed a scalability bottleneck since more hosts on the network meant more
updates, more hosts trying to download and more data to download. Distributed and
hierarchical structureof the DNS was and still is the key that enables the scalability of
hostnametoaddressmappings.
The importance of DNS is not restricted to enabling scalability. Hierarchical structure
allows local administration of names and addressing as well as local structure on the
name space. Domain names are mnemonic compared to IP addresses and so, after the
rapid growth of Web, they began to refer to products or services, rather than just
network resources (Levien, 2005). Additionally, many Internet companies, e.g.,
Amazon.com,usedomainnamesastheircorporatenames.
3.3.5 PrivatizationoftheInternetbackbone
In the early years of the Internet many networks existed, but not all of them were
compatiblewithandconnectedtoeachothers.Inthemiddleofthe80sNationalScience
Foundation (NSF) funded by the U.S. government built NSFNET that formed the initial
InternetbackbonebyconnectinguniversitiesandresearchorganizationsintheU.S.The
NSFNET backbone was restricted to research and educational purposes. The growing
number of users, however, was interested in using the Internet for commerce, which
encouraged commercial companiestoofferInternetconnectivitybybuildingtheirown
backbonenetworks(Kesan&Shah,2001).
19
The technical piece of the puzzle, Border Gateway Protocol (BGP), was created in the
beginning of 90s to replace the EGP (Exterior Gateway Protocol) used in NSFNET.
Although EGP had not faced serious scalability problems, decentralization of inter
domain routing allowed the increasing scale of the Internet. Foremost, BGP enabled
multiplebackbonesthroughfullydecentralizedroutingandthusmadeitpossibletoput
NSFNETbackboneouttopasture.PrivatizationandBGPtogethercombinedtheseparate
networksandpavedthewayfortheexplosivegrowth.
3.3.6 Enablinggrowth
In the early 90s it became obvious that the Internet had grown beyond anyone's
expectations and that the growth would continue and bring along serious scaling
problems related to routing and addressing (Clark et al., 1991). The original 32bit
addressspacewasseeninadequateinthelongtimespan,andtherealsowereplentyof
shorttermproblemsthatwereseenurgenttosolve.Additionally,theBGPtablegrowth
rate exceeded the growth in router hardware and software capabilities. Thus three
different solutions Classless InterDomain Routing (CIDR), Network Address
Translators(NATs)andDynamicHostConfigurationProtocol(DHCP)wereintroduced
toeasethesituation.
As RFC 1517 (Hinden, 1993) notes, IP address allocation based on classes (A, B, and C
with~16million,65536and256addressesrespectively)wasnotflexibleenough.TheC
classwastoosmallandtheBclasswaytoolargeformidsizedorganizations.CIDRsolved
this issue through more flexible (classless) address allocation, and at the same time
changed the BGP table growth rate from exponential to linear by introducing provider
addressaggregation(Huston,2001).
Network address translators were introduced to alleviate IPv4 address exhaustion by
enablingthesharingofoneglobalIPaddresswithmultiplehostsusingprivateaddress
space(Egevang&Francis,1994).AdditionalbenefitsofNATsincludeimprovedsecurity,
since all the inbounding connections are blocked to all the ports until the mapping is
completed (Hain, 2000). The use of NATs distorts the onetoone mapping between
Internet hosts and IP addresses and thus breaks the endtoend connectivity. Besides,
20
NATswerethoughttobeashorttermsolutionbuttheirextensiveusagehasdisrupted
implementationoflongertermsolutions,forinstancetheadoptionofIPv6.
DHCP, for one, was mainly developed to allow automatic and dynamic IP address
allocationandthussupportsystemandlocalareanetwork(LAN)managementandauto
configuration. Nevertheless, the timing was fortuitous because DHCP was also able to
helpwiththeconservationofIPaddresses(Levien,2005).
3.4 Servicemilestones
When looked from the service (or demand) side, the Internet has not limited services
andapplicationsthatcanusethenetwork.Thereisclearlyonemilestoneaboveallother
Web.Itsimpactisevaluatedlargeinallthecategoriesandaddedtothis,threeofthe
following four milestone services (search engines, TLS (Transport Layer Security) and
video streaming) are used mostly on top of the Web. The most influential service
milestonesarelistedinTable3belowanddescribedmoredeeplyinthenextsections.
Table3:Servicemilestones
Milestones
Year
FirstRFCNr.
(year)
FileTransferProtocol
1971
765(82)
FTP
OriginalKillerApp
Email
1973
822(82)
SMTP
WebtakestheInternet
bystorm
1866(95),
1993 1738(94), URL,HTML,HTTP
2068(97)
Searchengineschange
thewaypeoplebrowse
theWeb
1995
Researchera
Description
Commercialera
ThemostinfluentialRFC
2068(97)
Name
HTTP(Altavista,
laterGoogle)
TLSprovidesprivacyand
1996
enablesecommerce
2246(99) TLS(SSL)
P2PTrafficBoom
2000
Architectural
1958(96) Principlesofthe
Internet(Napster)
Videostreaming
proliferates
2005 2616,3550
HTTP,RTP
(Youtube)
21
Evaluationcriteria
Nr.of
users
Nr.of
hosts
Amount
oftraffic
Econ.
impact
Nr.of
services
Timecon
sumption
Changein
usage
patterns
Large
Medium Medium Medium
Large
Small
Large
Large
Large
Large
Large
Large
Large
Large
Large
Medium
Large
Medium
Small
Large
Large
Large
Medium
Medium
Small
Large
Small
Small
Small
Medium
Small
Small
Large
Medium
Small
Medium Medium
3.4.1 Filetransferprotocol
TransferringfilesovertheInternet was oneof the earliest applicationsintheInternet.
The first proposed mechanism was specified in RFC 114 (A File Transfer Protocol) in
1971.Afterthemanydevelopmentsteps,RFC765specifiedFTPforuseonTCP(Postel&
Reynolds,1985).BeforetheWebtrafficsurpassedFTPtrafficin1995,FTPproducedby
farthelargestamountoftrafficintheNSFNETbackbone(Merit,1995).
FTPwasandisusedformanypurposes.IntheearlydaysoftheInternetFTPwasutilized
indistributingtheHOSTS.TXTfile,predecessorofDNS.NowadaysFTPhasmuchsmaller
role but still many users connect to their web servers by using FTP, although more
secureoptionslikeSSH(SecureShell)areavailable.
3.4.2 Emailtheoriginalkillerapp
Already in the beginning, in the era of timesharing computers, it was found out that
extendinghumancommunicationwasanaturaluseofthenewtechnology.Inthe1971
Ray Tomlinson invented an email program to send messages across a distributed
network(Zakon,2006).RFC733definedformatoftheemailmessagesin1977,andthe
currentemailprotocolcalledSimpleMailTransferProtocolwasdescribedinRFC822in
1982.
Mostimportantly,emailprovidedanewwayforpeopletocommunicateandchangedthe
natureofcollaboration.Firsttheemailconnectedseparategroupsofcomputer science
researcher in building the Internet (Leiner, 2003), later it made corporate
communication more effective and drove (together with Web) consumers to buy
Internet access. And yet still it is the most popular purpose of use in the Internet
(StatisticsFinland,2008).
3.4.3 WebtakestheInternetbystorm
World Wide Web (Web, WWW, W3), developed at CERN by Tim BernersLee, was
releasedintheendofyear1990.Twobasicdesignprincipleswereprincipleofminimal
constraint (meaning as few specifications as possible) and principle of modularity and
information hiding (meaning that necessary specifications should be made
independently) (BernersLee, 1996). These principles resulted in three protocols: 1)
22
HTTP(HyperTextTransportProtocol),thenetworkprotocolusedbetweenWebservers
andclients;2)HMTL(HyperTextMarkupLanguage),themarkuplanguagetodescribe
the structure of web pages; and 3) URI (Universal Resource Identifier), the address
systemtoidentifyresourcesintheWeb(BernersLee,1994).
The Web enabled information dissemination over the Internet in easy and flexible
manner. The universality and accessibility of Web meant that people were able to surf
theWebindependentofoperatingsystemorcomputermodel.EspeciallyafterMozaic,an
early web browser, was released 22 April 1993, the Web proliferated at a 341,634%
annual growth rate (Zakon, 2006). Already in 1995 the Web surpassed ftpdata as the
servicewithgreatesttrafficintheInternet.ThusitcanbesaidthattheWebplayedakey
role in popularizing the Internet. During the last 15 years the Web has grown to be a
platform of tremendous (commercial) potential. It has changed the way we search
informationandconductbusiness.HencethetermsInternetandWebareoftenconflated
inpopularuse.
3.4.4 SearchengineschangethewaypeoplebrowsetheWeb
FindingthedesiredWebpagesandinformationisthestartingpointforusingtheWeb.In
the early days of the Web, navigating was based either on guessing the URL (Uniform
Resource Locator) or using directory services (e.g., Yahoo) (Levien, 2005). Search
engines, at the beginning for instance Excite, Lycos, and AltaVista, little later Google,
revolutionized the way people navigate in the Web. Finding a related RFC is, however,
not too easy, since the key components of search engines proprietary search
algorithms and databases are not standardized in RFCs. Thus the search engines are
linkedtotheWebthroughHTTP.
A study of Pew Internet Project reports that 84% of the U.S. Internet users have used
search engines, and on any given day, 56% of those online use them (Fallows, 2005).
AccordingtoAlexa.com(2009),thetoptwowebpagesintheInternetinApril2009were
google.com and yahoo.com, both search engines. The success of Google has created an
expressiontogooglewhichbyWebstersNewMillenniumDictionaryhasthemeaning
"tosearchforinformationontheInternet,especiallyusingtheGooglesearchengine.
23
Although the initial impact of search engines was how they changed the way people
browsetheInternet,economiceffectsshouldnotbeforgotten.Searchenginecompanies,
mostprominentlyGoogle,havebeenabletoconverttheirsearchenginesintoadvertising
business. This is doneboth by offering paidads related to websearchesandsellingad
spacetoalargebaseofWebpagesandusingsearchalgorithmstoshowrelevantadsin
eachpage.Theimportanceofsearchenginesinfindingnewinformationhasalsocreated
marketforsearchengineoptimizationthattriestohelpWebsitestoraisetheirposition
insearchresults(Levien,2005).
3.4.5 TLSprovidesprivacyandenablesecommerce
Need for securing privacy, authentication and data integrity in clientserver
communication was identified soon after the invention of the Web. The SSL (Secure
SocketLayer)protocolwasoriginallydevelopedbyNetscapeandafteracoupleofdraft
versions,thestable3.0versionwasreleasedin1996(Rescorla,2001).Shortlyafterthat
SSL development became responsibility of IETF which renamed the protocol to TLS
(Transport Layer Security) (Thomas, 2000). Most commonly TLS is used together with
HTTPtoformHTTPSusedinsecuringWebpages.Asaflexibleprotocollocatedbetween
applicationandtransportlayersitalsosupportsotherapplicationlayerprotocols.
TLS enabled reliable ecommerce and allowed the Web to be used as a commercial
service platform (Thomas, 2000). Ebanking, credit card payments, and using different
kindofonlineservicesrequiringauthenticationaresomeexamplesofapplicationswhich
useTLS.
3.4.6 P2PTrafficboom
PeertoPeer (P2P) systems came to notice of the wide public in 1999 through MP3
(MPEG1AudioLayer3)filesharingapplicationcalledNapster.ItmadesharingofMP3
fileseasyand pavedthe way forlater peertopeer filesharingapplications.Nowadays
peertopeerfilesharingisthebestknownapplicationofP2Pbutitisnottheonlyone.
Other possibilities cover VoIP (e.g., Skype 3 ), instant messaging (e.g., ICQ 4 ), remote
3http://www.skype.com
4http://www.icq.com/
24
collaboration (e.g., shared file editing), distributed computing (e.g., SETI@home 5 ) and
streaming media (P2PTV) (Beijar, 2008). Most of these P2P systems use proprietary
protocolsanddedicatedclientapplications.ThuslinkingtoRFCsispossibleonlythrough
basicdesignprinciplesoftheInternetsincetheriseofP2Papplicationsmeansactually
reverting from the clientserver architecture dominating in the Web to the original
Internetarchitectureconsistingofequalpeers(Oram,2001).
Peertopeer (file sharing) systems increased the Internet traffic substantially. Ipoque
(2007) found out that P2P is producing more traffic in the Internet than all the other
applicationscombined.InEasternEuropetheproportionofP2Ptrafficwasasmuchas
83%. The other implications relate to the economic aspects. Easiness to find and
download music, movies, TV series, and other content free of charge has affected the
purchasebehaviorofsomeusersbychangingthewillingnesstopayforcontentandby
allowing experimentation. Although the content industry anxiously claims that P2P
declines the sales, for example the study of OberholzerGee and Strumpf (2006)
regardingmusicsalessuggeststhatP2Pfilesharinghasnostatisticallysignificanteffect
onsales.Theincreasingtrafficamountandsymmetrictrafficpatternhavealsoeffecton
Internet service providers (ISPs) (transit) costs, which has made some companies to
restrictoroptimizepeertopeertraffic.
3.4.7 Videostreamingproliferates
Streamingmedia,especiallystreamingvideo,isresponsibleforlargeamountofInternet
traffic.EllacoyaResearch(2007)andCisco(2008a)reportthatInternetvideoaccounts
forabout20%oftheInternettraffic.Despitehistoricalimportanceofvideo,theimpact
and proportion of video traffic will most likely just increase. Cisco (2008a) namely
forecasts that already in 2012 Internet video will account for 50 % of total Internet
traffic. Therequirements ofInternet video streaming have been oneof the factors that
havecreatedmarketforcontentdeliverynetworks(CDNs).
5http://setiathome.ssl.berkeley.edu/
25
Theworldsthirdmostpopularwebsite(Alexa,2009),YouTube 6 ,createsabouthalfof
theInternetvideotraffic,atleastinNorthAmerica(EllacoyaResearch,2007).YouTube,
as most of the other video streaming services, uses Adobe Flash to display the video
(Ipoque, 2009). In Flash the video is delivered using HTTP/TCP, and the delivery
technique is called progressive downloading or pseudostreaming, since the file is
actuallydownloadedtotheuserbuttheplaybackcanbestartedbeforethewholefileis
delivered (Gill et al., 2007). There are also many protocols for nonHTTP video
streaming. IETF has standardized the RTP family (RTP, Realtime Transport Protocol;
RTCP, RTP Control Protocol; RTSP, Real Time Streaming Protocol) for this purpose.
Although nonHTTP streaming is used in live streaming, its amount of total Internet
traffic is petty (only 3% compared to 17% of HTTP video in North America (Ellacoya
research,2007)).
6http://www.youtube.com
26
4 Internetecosystem
ThischapterpresentsthecurrentstatusoftheInternetecosystem.Thepurposeistofirst
describe the industry structure, size and traffic characteristics, and then list some
identifiedproblemsaswellaseffortsthattrytotacklethem.
4.1 Internetinterconnectivity
TounderstandtheInternetecosystem,thedifferentplayersandtheirinterrelationsneed
to be explained. This can be handled both from technical, networkcentric perspective
and from economic point of view. This section takes the network level view while
Section4.2explainsthemostimportantstakeholders.
The Internet consists of heterogeneous networks called autonomous systems (ASes).
TheyareoperatedmostlybycommercialInternetServiceProviders(ISPs),butalsoby
corporations and other enterprise providers, universities, government agencies, and
content providers and other specialized service providers (Clark et. al, 2008). ISPs
connectendusersandbusinessestothepublicInternetbysellingInternetaccess.They
compete over customers on price, performance, reliability etc. but they must also co
operatetoofferuniversalendtoendconnectivity(Norton,2001).
TheinterconnectivitybetweenASesisarrangedbytwobasictypesofagreementspaid
transitandsettlementfreepeering.
Definition:A TransitRelationship is a business arrangement whereby an ISPprovides
(typicallysells)accesstotheglobalInternet(Norton,2002).
Definition: Peering is the business relationship whereby ISPs reciprocally provide
accesstoeachotherscustomers(Norton,2002).
Therecursivecombinationofthesestandardizedcontractsresultingfromcomplexand
dynamic bargaining game between pairs of ASes creates the web of interconnections
(Clarket.al,2008).
27
The simplified structure of the Internet consists of ISPs, content and enterprise
companies,andendusers.ISPscanbedividedintotwogroupsTier1ISPsandTier2
ISPs.GeneralizedInternetecosystemispresentedinFigure4.
Fullmesh
peering
Tier1ISPs
Transit
free
Partialmesh
peering
Tier2ISPs
Mustbuy
transit
Generally
nopeering
Content/Enterprise
Companies
Consumers
Mustbuy
transit/access
Figure4:GeneralizedInternetecosystem(adaptedfromNorton,2003,modified).
Definition:ATier1ISPisanISPthathasaccesstotheglobalInternetroutingtablebut
doesnotpurchasetransitfromanyone(Norton,2001). 7
SinceTier1ISPsdonotbuytransit,theyhavetogetaccesstotheentireInternetsolely
through peering relationships. This means that every Tier 1 ISP must peer with all the
otherTier1ISPs,andthustheamountofTier1ISPshasstayedquitelimited.According
toRenesysCorporation(2009),therewere13Tier1ISPsinJanuary2009.
Tier 2 ISPs are a heterogeneous group of ISPs that differ in geographical coverage,
amountofcustomersandproportionoftransitandpeeringtraffic.SomesmallISPsbuy
onlytransitandsomelargeISPshavevastamountofpeeringagreements.Thecommon
factoristhattheystillhavetobuytransit.
7Strict
definition of Tier 1 ISP requires that ISP is not only transitfree but also all of its peering
relationships need to be settlementfree. Technically there is no difference in settlementfree and paid
peering,sothelooserTier1definitionisusedhere.
28
ContentandenterprisecompaniesaretypicallycustomersofISPs.Theirconnectivityto
theInternetismostlybasedontransitagreements,andpeeringisrare.ISPsalsoconnect
endusers(consumers)totheInternetbysellingInternetaccess.
4.2 Stakeholders
While the Internet has extended its tentacles to the entire society, the amount of
stakeholdershasincreasedandtheirincentivestoinfluenceonInternetevolutionhave
become stronger and more diverse. Clark et al. (2002) call this process of adverse
interestsbetweenstakeholdersthetussle.Theyhaveidentifiedvariousstakeholdersof
the Internet landscape and some examples of tussles. Identified stakeholders include
users, commercial ISPs, private sector network providers, governments, intellectual
property rights holders, and providers of content and higher level services. Some key
playersare,however,missingfromthislist.Thereforeourownviewofthekeyplayers
andtheirinterrelationsisexplainedbrieflybelowanddrawninFigure5.
Networkinfravendors deliver network HW and SW (e.g., routers, fiber, radio

access components, network management tools) to both access and backbone
operators.
Backbone operators (ISPs) sell global Internet interconnectivity to access

operators.AnISPcanbeatthesametimebothbackboneandaccessoperator.
Access operators (ISPs) sell Internet access to the endusers and buy
interconnectivityfrombackboneoperators.
Device & Software vendors manufacture devices like computers, mobile

phones,andPDAsaswellassoftwarelikeoperatingsystems,browsers,andemail
clientstoendusersandcontent&applicationproviders.
AdvertisersenablemanyInternetservicesbypayingforadvertisingspace.
Content & Application providers produce Internet services that attract end
userstousetheInternet.
Endusers, covering both consumers and enterprise customers,access Internet

content, services and applications with their devices that run software. The
connectivityisofferedbytheaccessoperatorsandlegitimateactionsaredefined
bytheregulators.
29
Governments&Regulatorsenactlawsonthegroundsofthesocietysdemand.
Governments have also national motives for influencing actively the Internet
evolution.
Research institutes, for instance universities, develop Internet protocols and

technologies.
Standardizationbodies,alsoincludingInternetgovernancebodieslikeICANN,
standardizetechnologiesandallocateresources.
Advertisers
Enableservices
Device&
Software
vendors
DeliverHW&
SWforusing
Internet
services
Network
infravendor
Content&
Application
providers
DeliverHW&SW
Createattractive
services
Endusers
Provide
access
Access
operators
Provide
access
Backbone
operators
Developtechnologies
Regulatetheoperatingenvironment
Standardizetechnologies
Researchinstitutes
Governments&Regulators
Standardizationbodies
Figure5:StakeholdersoftheInternetecosystem.
4.3 SizeoftheInternet
HowbigtheInternetis?Theansweriswereallydonotknowbecauseitisunorganized,
uncataloguedandcontinuestogrowataphenomenalrate.However,somesourcesare
offeringguestimatesfromdifferentperspectivesrangingfromthepartsofthetechnical
infrastructure to the actual usage. Next sections give a brief overview to the question
frombothtechnicalandusageviewpoints.
30
4.3.1 Technicalviewpoint
Thefirstmeasureofscaleistheamountofautonomousnetworks(ASes)intheInternet.
Theamountofthemon14April2009was30872(ASN,2009).Thisfiguredoesnotgivea
verygoodestimateoftheInternetsizesincesomeoftheASes(typicallyISPnetworks)
are huge networks consisting of millions of hosts and users while others (for example
corporatenetworks)aremuchsmaller.
Thehostcountisanotherthingthatcanbeusedasameasureofscale.InternetSystems
Consortium(ISC)hascollectedthenumberofhostsadvertisedintheDNStwiceayear
since 1987. The latest survey conducted in January 2009 found 625,226,456 hosts
connected to the Internet (ISC, 2009). Figure 6 presents the survey results since 1994
andgivesaniceoverviewofthegrowthoftheInternet.
Figure6:Internetdomainsurveyhostcount(adaptedfromISC,2009).
4.3.2 Usageviewpoint
TheInternetsimpactonsocietycanbeevaluatedbydissectingthenumberofusersand
penetration rates around the world. Internet World Stats (2009) collects this
information from market research companies, international telecommunication agents,
andlocalregulators.Theirreportshowsthattherearealmost1.6billionInternetusers
31
among population of 6.7 billion, which means average penetration rate of 23.8%.
Althoughitcanbequestionedifthenumbersarealittlehigh,theygiveatleastarough
conceptionofthedimensions.
Statistics about number of users and penetration rates (Figure 7) reveal large regional
differences. The Internet plays the most significant rolein North America, Oceania and
(Western) Europe where the penetration rates are over 50%. However, the largest
growthpotentialisindevelopingcountriesinAsia,AfricaandLatinAmerica,whichhave
large population but modest penetration rates. Asia for instance has already now the
mostInternetusers(andChinaisthelargestcountry,298millionusers),butthisisnot
duetohighpenetrationratebutlargepopulation(56.3%oftheworldspopulation).
WorldInternetusersandpenetrationrates
Millions
InternetUsers
Penetration
700
80,0%
600
70,0%
60,0%
500
50,0%
400
40,0%
300
30,0%
200
20,0%
100
10,0%
0,0%
Figure7:Internetusersandpenetrationratesintheworldbygeographicregionson31March
2009(InternetWorldStats,2009).
32
TheusagecanalsobestudiedbylookingtheamountoftransferreddataintheInternet.
Ciscos (2008a) estimate based on the analyst projections suggests that the amount of
monthly Internet traffic in 2009 is 10666 PB. When this is divided among 1.6 billion
usersitmeansapproximately6.7GBtrafficperuserpermonthwhichis233MBperday.
That is quite a high number if an average consumer is considered but there are also
heavy users both in the private and in the commercial sector that certainly exceed the
number.
4.4 Trafficcharacteristics
WhatkindoftrafficiscarriedbytheInternet?Thisisaninterestingquestioninorderto
understandhowpeopletrulyusetheInternet.Furthermore,trafficcharacteristicsareof
highinterestfromtechnicalperspectivesincetheyrevealhowdependenttheInternetis
on a handful of protocols. Thus the traffic distribution in the network, transport and
applicationlayersispresentedinthefollowingthreesections.Duetothedecentralized
structureoftheInternet,itisdifficulttomeasurethetrafficcharacteristicsglobally.Thus
the numbers in this section are based on small subsets of Internet traffic which are
believedtopresenttheglobalsituationatleastcoarsely.
4.4.1 Networklayer
Network layer is the thin waist of the Internet protocol suite which offers transparent
connectivityforvariousapplicationsbetweendiverseunderlyingnetworktechnologies.
Hence IP is the only major protocol in this layer 8 . The original version of the protocol,
IPv4,dominatestheInternet.Thenewerversion,IPv6,whichofferslargeraddressspace,
has been available for over 10 years now but its adoption has been really slow. Mike
LebersglobalIPv6deploymentreporttellsthatcurrentlyonly4.4%ofallthenetworks
(ASes) and one of the 500 most popular websites (identified by Alexa.com) run IPv6
(Leber, 2009). On the traffic level the situation is even poorer. CAIDAs (2009)
measurementsshowthattheproportionofIPv6isaspettyas0.005%ofthetrafficina
backbonelinkbetweenSeattleandChicago(Table4).
8Internet
Control Message Protocol (ICMP) is important but its messages are encapsulated within IP
datagrams,andsecureversionofIP(IPsec)hasbunchofprotocols(ESP,GRE,etc.)thatcreatelittletraffic.
33
4.4.2 Transportlayer
OnthetransportlayerthemarketissharedamongTCPandUDP,twoprotocolsthathave
beenavailablefromtheearlydaysoftheInternet.Togethertheyaccountforover99%
of the traffic in a backbone link from Seattle to Chicago (CAIDA, 2009). Furthermore,
reliable TCP dominates the transport layer with its 92% share of the bits compared to
7% of unreliable UDP (Table 4). Similar proportion of TCP (94%) was identified in a
studyconcerningFinnishmobilepacketnetworkdatatrafficin2007(Kivi,2008).Onthe
packet level, the proportion of TCP is slightly smaller and proportion of UDPis higher,
whichindicatesthatUDPpacketsaremuchsmallerthanTCPpackets.Newerandmore
effective transport protocols for certain use cases (e.g., DCCP and SCTP) have been
developed to overcome some of TCPs shortcomings. However, these new transport
protocols are not typically recognized by transportaware middleware (e.g., NATs and
firewalls),whichrestrictstheirwidescaledeployment(Huston,2008b).
Table4:ProtocoltrafficdistributioninOC192backbonelinkfromSeattletoChicago(CAIDA,2009).
Protocol
TCP
UDP
ESP
GRE
1day*
bits/s packets/s
RSVP
other
91,86%
7,61%
0,31%
0,17%
0,0490
%
0,0054
%
0,0020
%
0,00%
*
**
***
****
1day
1week
4weeks
2years
ICMP
IPv6
81,01%
17,84%
0,62%
0,14%
0,3700%
0,0072%
0,0073%
0,00%
1week**
bits/s packets/s
92,39%
6,99%
0,38%
0,19%
0,0560
%
0,0047
%
0,0022
%
0,00%
81,71%
17,08%
0,64%
0,16%
0,4000%
0,0054%
0,0078%
0,00%
4weeks***
bits/s
packets/s
92,87%
6,42%
0,49%
0,21%
0,0580
%
0,0047
%
0,0025
%
0,00%
2years****
bits/s
packets/s
82,59% 92,49%
16,19% 7,79%
0,67% 0,35%
0,17% 0,33%
0,1200
0,4200%
%
0,0024
0,0055%
%
0,0029
0,0086%
%
0,00% 0,00%
83,69%
16,35%
0,48%
0,23%
0,4500%
0,0026%
0,0053%
0,00%
April162009April172009
March202009April172009
4.4.3 Applicationlayer
Variouscompanies(CacheLogic,2005and2006;EllacoyaNetworks,2007;Cisco,2008b;
Ipoque, 2007 & 2009; TeleGeography, 2009) have reported application traffic
distributions during the last years. The results show some variance so just the rough
34
estimatesarepresentedhere.About85%ofthetrafficisHTTPorP2P(e.g.,BitTorrent,
eDonkey, Gnutella) traffic. Some sources suggest that over 60% of the traffic is P2P
(CacheLogic, 2006; Ipoque, 2009) while others (Ellacoya Networks, 2007;
TeleGeography, 2009)reportthatHTTP has recently eclipsed P2P because of YouTube
style video streaming. About half of the remaining 15% of the traffic comes from non
HTTP streaming. VoIP (e.g., SIP), email (SMTP), and other application level protocols
cover the rest. The most significant discovery in application layer is the importance of
HTTP. It was developed for Web to support retrieval of Web pages but nowadays it is
usedbymultitudeofapplications.
4.5 Presentproblems
As the Internet has evolved beyond its original scale and scope the drawbacks and
deficiencies of the original design principles following from the endtoend argument
havebeenexposed.DavidClark,themostfamousadvocateoftheendtoendargument,
andBlumenthallistmultipletrendsandproblemsthatmayerodetheapplicabilityofthe
endtoend argument and lead to architectural change of the Internet (Clark &
Blumenthal,2001).
The strategic research agenda (SRA) of FI program (Nikander & Mntyl, 2007)
identifiessixpresentproblems:
unwantedtraffic,
chokingoftheroutingsystem,
mobilityandmultihoming,
consumptionandcompensation,
privacyandcompensation,and
trustandreputation.
TheseproblemsareusedhereasthebasisfordescribingsomeofthemotivesforFuture
Internetresearch.Nextsections(4.5.14.5.6)explaintheminmoredetail.
35
4.5.1 Unwantedtraffic
Unwantedtraffic,includingapplicationlevelflooding(e.g.,emailandinstantmessaging
(IM) spam), security attacks (e.g., worms, Trojan horses, or direct), and distributed
denial of service attacks, is a consequence of nave assumption that users act
benevolently. The Internet was built homogenous, mutually trusting research
community in mind, and it was assumed, that the sender does not send packets if the
receiver is not willing to receive them. Due to the economic reasons some parties are,
however, interested in forcing interaction on another (Clark & Blumenthal, 2001). The
unintentionalconsequenceoftheInternetsnetworkarchitectureisthatthemaincostof
communicationispaidbytherecipient.Andwhenthemarginalcostofsendingpackets
isveryclosetozero,therearefewincentivesnottosendunwantedtraffic.Thetypical
counterattackagainstunwantedtrafficisuseoffirewalllikemiddleboxesalthoughthey
have their own problems relating to reachability limitations. From wider perspective
unwantedtraffichastodowithfairnesssincethecostofunwantedtrafficismostlypaid
bythepartythatdoesnotevenwanttointeractatall.
4.5.2 Chokingoftheroutingsystem
RFC 4984 Report from the IAB workshop on routing and addressing (Meyer et al,
2007) names the routing scalability as the most important problem facing the Internet
today.TheproblemincludesthesizeoftheBGPtableandtheimplicationsofgrowthto
theroutingconvergencetimes.Consequentlythecorerouterhardwareneedstobemore
effective (faster memory and processing). Additionally, traffic engineering complicates
theroutingsystemsinceBGPdoesnotofferanygoodtoolsforit.Theroutingproblems
relate inherently to addressing. The nonallocated IPv4 address space is projected to
exhaust in 20112012 (IPv4, 2009), which brings urgency to solve the problem
somehow.Alltheproposedsolutions(transitiontoIPv6,reallocationofIPv4addresses
throughtransfermarkets(Mueller,2008),andmoreextensiveusageofNATs(Nishitani
et al., 2008)) increase stress on the routing system. Although the problems in routing
systemarenotacute(Huston,2009),theirdifficultycallsforimmediateattention(Meyer
etal.,2007).
36
4.5.3 Mobilityandmultihoming
MobilityisanotherdilemmawhichwasnotaddressedwhentheInternetwasdesigned.
Today,however,thereisaclearneedformobilityandmultihoming.Semanticoverload
of IP addresses is one of the reasons behind the mobility problem (Huston, 2006). IP
addressesareusedbothaslocators(answertoquestionWHERE)andidentifiers(WHO).
They work also as lookup keys in making local switching decisions (HOW). Mobile
deviceschangetheirlocationconstantlywhiletheystillwishtokeeptheiridentity.Thus
eitheranewnamespaceofferedforexamplebyHostIdentificationProtocol(Moskowitz
and Nikander, 2006) or level of indirection is needed to provide effective solution for
mobility and multihoming. Mobility is a problematic requirement not only from the
perspective of terminal mobility but also from the perspective of user, session and
processmobility.
4.5.4 Consumptionandcompensation
Deep in the Internet architecture lies an assumption of benevolent and cooperative
agents that together work for maximizing throughput in the network. But if agents
behaveselfishly,astheydomoreandmoretoday,somemechanismsforcompensating
the resources they (over)use are inevitable. The Internet, however, is missing
mechanisms and incentives for compensation. This can be seen as exiguity of resource
andcongestioncontrolmechanisms(Nikander&Mntyl,2007).Resourcecontroltries
to guarantee sufficient resources at all times while congestion control attempts to
allocateresourcesfairlyinthosesituationswherethereisscarcityofthem.Oneaspect
ofthisproblemishowtosatisfydiverseperformancerequirementsofdifferenttypesof
applications, and at the same time implement feasible and fair congestion control.
Anotherquestionis:DoestherevenuesintheInternetflowtothemwhoneedtodothe
investments?TheInternetlacksmeansofroutingmoney,whichisoneofthereasonsfor
thesuccessofadbasedbusinessmodels.Fromcustomerperspectivethekeyquestionis
usabilityofpayingmeaningthattheinconvenienceofpayingneedstobeminimized 9 .
9Onlyeasyenoughsolutions(usable,theamountofpaymentsisminimized)maysucceed,e.g.,payingonly
onceamonth(operators),payingnotatall=adbased(Google),payingisbundledwithdevice(Nokia).
37
4.5.5 Privacyandattribution
The relationship between users rights and responsibilities is not clear in the Internet.
The most critical tension is the one between anonymity and accountability (Clark &
Blumenthal, 2001). Privacy, if not absolute anonymity, is seen important in many
societies.Topreventbadthingshappening,thedesireforprivacyneedstobebalanced
againsttheneedforaccountability.Unfortunatelythereareverylittletoolsforfostering
accountability in the Internet architecture. This is related to the more common
characteristics of information technology recognized by Lawrence Lessig. His famous
quote Code is Law proposes that Internet design realized in TCP/IP has such
characteristicsthatmakeregulatingbehaviordifficult(Lessig,2000).DuetothisLessig
seesthatthetechnicalfoundationneedstobebuiltinawaythatprovidesbalancedlevel
ofprivacyandattribution.
4.5.6 Trustandreputation
AsimplemodeloftheearlyInternetamutuallytrustingcommunityisgoneforever.
However, users who do not trust each other still desire to communicate (Clark &
Blumenthal,2001).IntheirmorerecentpaperClarkandBlumenthal(2007)reassessthe
originalendtoendprincipleandreformulateitasatrusttotrustprinciple:
Thefunctioninquestioncancompletelyandcorrectlybeimplementedonlywiththe
knowledgeandhelpoftheapplicationstandingatapointwhereitcanbetrustedto
doitsjobinareliableandtrustworthyfashion.
Thismovesthefocusfromendpointstotrustworthypointsofexecution.AlthoughSrel
and Nikander (2008) see that the technical architecture needs to foster trust, the
problems of trust and reputation are largely nontechnical and relate for instance to
human factors like the ability to make decisions involving risks using existing and
possiblyinadequateinformation.
4.6 FutureInternetresearch
AlthoughtheInternettechnologieshavebeenstudiedextensivelyfromthebeginning,a
newwaveofresearchactivitieshasemergedinrecentyearsduetotheincreasedpublic
awareness of the shortcomings in the Internet architecture. Future Internet is a
38
summarizingtermforalltheseresearchactivitiesthatstrivefordevelopingtheoriginal
Internet. The diversity of Internet technologies means that the related research topics
arewidespread.Someeffortsconcentratesolelyonincrementalevolutionwhileothers
aimforcompletearchitecturalredesign(alsocalledasacleanslateapproach).Inmany
researcheffortstheseapproacheslivesidebysidesothatincrementaldevelopmentsare
used to tackle shortterm problems while clean slate is seen as a longterm solution.
Despiteofthemanycleanslateresearchefforts,notalltheacademicsareconvincedof
their rationality or value. Milton Mueller, scientific committee member of Internet
GovernanceProject(IGP),claimsthatpromisingnewInternetmaybeagreatstrategyfor
governmentfundingbutitisnothonestsincetheinertiaoftheInternetaffectssothat
thereisnoreplacementoftheoldInternetwithanewone(Mueller,2009).
Table 5 lists several Future Internet research efforts in Europe and elsewhere. Future
Internet Assembly (FIA) is a European Union initiative that acts like an umbrella over
diverse research projects funded by Seventh Framework Programme (FP7). National
research efforts, like Finnish Future Internet program, supplement Europeanlevel
projects.OutsideofEurope,U.S.NationalScienceFoundationhaslaunchedtwoprojects
(GENIandFIND),andJapaneseandSouthKoreanshavetheirowneffortstoo.Although
all these national or regional projects aim for rising to the challenges the Internet is
facing, and they speak for and understand the need of international cooperation, a
significant motivation for them is the (foreseen) strategic importance of the Internet.
Thuseverynationtriestoshoreuptheirfuturepositioninthenetworkedworldthrough
beingakeyplayerinthedevelopmentofthenextgenerationInternet.
39
Table5:FutureInternetresearchefforts
Europe
FIA:FutureInternetAssembly
FutureInternetprogramme
InternetdelFuturo
GLab
GRIF:GroupedeReflexion
InternetduFutur
Country
EU
Finland
Spain
Germany
AmbientSweden
Sweden
France
www.vinnova.se/upload/
EPiStorePDF/AmbientSweden.pdf
IBBT:InterdisciplinaryInstitute
Belgium
forBroadbandTechnology
LuxembourgIPv6Council
Luxembourg
Italy
TheNetherlands
Ireland
UK
Other
GENI:GlobalEnvironmentfor
NetworkInnovations.
FIND:FutureInternetDesign
AKARI:ArchitectureDesign
ProjectforNewGeneration
Network
FIF:FutureInternetForum
Link
www.futureinternet.eu
www.futureinternet.fi
www.internetdelfuturo.es
www.germanlab.de
www.ibbt.be
www.ipv6council.lu
http://cit.fbk.eu/future_internet
www.futureinternet.ez.nl
www.futureinternet.ie
www.internetcentre.imperial.ac.uk
Country
Link
USA
www.geni.net
USA
www.netsfind.net
Japan
http://akariproject.nict.go.jp/eng/
index2.htm
SouthKorea
www.fif.kr/
40
5 Scenarioconstructionprocess
Inthischaptertheprocessusedinconstructingscenariosispresented.Thefindingsare
expressed in form of key trends and uncertainties which summarize the acquired
understandingoftheFutureInternet.
5.1 Brainstorming
Keytrendsanduncertaintieswereidentifiedinthreebrainstormingsessionsorganized
in the autumn of 2008. Each session had 68 academics/industry experts representing
different stakeholders. PEST framework was used to cover broadly all the important
macroenvironmental factors affecting the Future Internet. Different domains of the
framework were brainstormed separately in two phases (1. nominal, 2. interactive).
First, every participant had 10 minutes to write statements to Postits about forces
havingeffectontheFutureInternet.Discussingwasforbiddenbutcoupleofkeywords
were shown to guide and help the thinking (Figure 8). Second, the participants were
asked to present their thoughts in the order of importance. In this phase (30 min)
discussionwasopenandsimilarPostitsweregrouped.Afterrapidexchangeofthoughts
agroupedideawasplacedononeofthefourboxesofflipchartmatrix(Figure9)based
onitsimportanceanduncertainty.Alltheidentifiedforcesfromthreesessionsarelisted
inAppendixA.
1.Political/regulatoryforces
2.Economic/industryforces
Legislationindifferentlevels
Governmentpolicies
Spectrumpolicy
Competitionpolicy
Ecological/environmentalissues
4.Technologicalforces
3.Socialforces
Economictrends
Industryevolution
Structureoftheindustry
Goalsofthecompanies
Localvs.globaleconomy
Users,values,attitudes
Demographics
Culture,lifestyle
Consumerbehaviour
Demandforservices
Keydevelopmentsintechnology
Technologybottlenecks
Rateoftechnologicalchange
Technologycompetition
Standardization,interoperability
Technologyaccess,patents
Figure8:Listofkeywordsusedtofeedthebrainstorming
41
5.2 Keytrends
Key trends are important factors that are certain or very likely to realize and have
significantimpactontheFutureInternet.Theyareunderlyingallfourscenariosandare
assumedtobevalidwithareasonableprobabilityforthechosentenyearperiod.Final
key trends presented in Table 6 are combined from several trends identified in the
brainstorming sessions. The presentation is divided into four categories based on the
PESTframework.
Table6:Keytrends
Political/Regulatorytrends
PT1:ThesocietywillbeincreasinglydependentontheInternet.
PT2:Theworld(andtheInternet)ismovingfromunipolartomultipolar.
PT3:Theusageandallocationofspectrumwillbemoremarketbased.
PT4:Environmentandenergywillbemoreimportant.
Economic/Businesstrends
ET1:Theworldismovingfromproductstoservices.
ET2:UsingICTbecomeslowcostcomparedtomanualalternatives.
ET3:PowerconsumptionbecomesacostdriverinICT.
ET4:Globalizationcontinues.
SocialTrends
ST1:TheInternetisintegratingdeeperintoeverydaylife.
ST2:Desireforallaroundavailabilityincreases.
ST3:Socialnetworkingwillbefasterandstronger.
ST4:Contentcreationwillbemoreuserdriven.
ST5:InternetgenerationcontinuestodriveInternetusage.
Technologicaltrends
TT1:MobilealwaysonInternetconnectivityincreases.
TT2:Performancecontinuestoimprove.
TT3:Complexityofsoftware,servicesandarchitecturesincreases.
TT4:Diversityofnetworksanddevicesincreases.
TT5:Remotemanagementofnetworkandhomedevicesincreases.
42
Uncertainty
Keyuncertainties
Keytrends
Importance
Figure9:Flipchartmatrixusedinevaluatingtheimportanceanduncertaintyofidentifiedforces.
5.2.1 Politic/Regulatorytrends
` PT1: The society will be increasingly dependent on the Internet. Economy,
industry, administration, and education move their operations increasingly to the
Internet and manual fallbacks in problem situations are disappearing. This raises
governmentalinterestinregulativecontrolthroughreregulation.
` PT2:The world(andtheInternet)ismovingfrom unipolartomultipolar. The
U.S.centeredwesternworldlosesitsdominantrolesincetheriseofChina,India,and
otherdevelopingnationsscattersthepoweraroundtheglobe.Additionally,thenext
2billionInternetuserscomemostlyfromthe3rdworldanddevelopingnations.
` PT3: The usage and allocation of spectrum will be more marketbased.
Increasing mobile Internet usage channels more spectrum for Internet access.
Spectrum usage will be more effective and spectrum auctions are used in most
countries.
` PT4:Environmentandenergywillbemoreimportant.Environmentalawareness
increasesandenergyconsumptioniscontrolledandregulatedstricter.
43
5.2.2 Economic/Businesstrends
` ET1: The world is moving from products to services. The money is on the
services, because producing goods is highly competed on the global space. The
Internetspeedsupthisdevelopment.
` ET2: Using ICT becomes lowcost compared to manual alternatives. Cost
reductionsandpossibilityforrationalizationofbusinessprocessesdriveadoptionof
ICTineveryfieldofeconomy.
` ET3:PowerconsumptionbecomesacostdriverinICT.AwarenessofICTspower
consumption and environmental effects increases at the same pace with improving
performance of devices. Therefore, energy efficiency becomes an important design
criterion.GreenICTisalsoseenashavingmarketingvalue.
` ET4:Globalizationcontinues.Thisoldtrendcontinuestoholdtruesincecountries
dependmoreandmoreoneachotherandbordersdisappear.However,inthefuture
globalization will be stronger in service and knowledge industries than in
manufacturingindustries.
5.2.3 Socialtrends
` ST1:TheInternetisintegratingdeeperintoeverydaylife.Mappingbetweenthe
real and virtual worlds tightensand peopleareincreasingly able and willing to use
Internet services. Tighter integration creates need for improvements in security,
trust,andprivacy.
` ST2: Desire for all around availability increases. People are used to being
reachable all the time with their mobile phones and now the same level of
accessibilitytoemail,socialnetworkingsites,andinstantmessagingisgeneratinga
demandformobiledataservices.Thisissupportedbytheincreasinguseoflocation
andcontextinformation.
` ST3: Social networking will be faster and stronger. Social networking services
gainimportanceandaffecthowpeoplecommunicateandconsume.Forexample,the
44
increasing usage of ratings and suggestions from other consumers changes buying
behavior.
` ST4: Content creation will be more userdriven. The easiness of creating and
sharingcontentintheInternetdrivestoYouTubeandWikipedia 10 styleofservices
whereusersareactiveparticipantsandnotjustpassiveconsumers.
` ST5: Internet generation continues to drive Internet usage. Young people are
eagertoadoptnewserviceswhileoldpeoplearenotabletodothat.Thispreserves
thegenerationgapbetweentheInternetgenerationandolderpeople.
5.2.4 Technologicaltrends
` TT1:MobilealwaysonInternetconnectivityincreases.TheInternetwillbeused
more and more with small, portable devices like mobile phones, PDAs, and ultra
portable PCs. Additionally, for many new users mobile connectivity will be the first
andonlyaccessmethod.
` TT2: Performance continues to improve. Processing power improves, optical
transmission boosts transfer rates and storage capacity increases. These
improvementscanalsobeseeninbetterpriceperformanceratios.
` TT3: Complexity of software, services and architectures increases. Patchon
patch tradition and new requirements increase the complexity of networks. At the
sametimeusageofnewapplicationsisstilltoocomplexformostusers.Thisraises
usabilityandreliabilityquestionstoanewlevel.
` TT4:Diversityofnetworksanddevicesincreases.TheInternetofthingsspreads
ubiquitous computing quietly and increases the amount of hosts significantly. The
diversedevicebaseisconnectedtotheInternetwithavarietyofaccesstechnologies.
Alsomachinetomachinecommunicationbringsnewrequirementsfornetworking.
10http://www.wikipedia.com
45
` TT5: Remote management of network and home devices increases. Managing

prolificandmorecomplexdevicebasewillbecarriedoutmoreandmoreremotely.
Thiswillhappenbothinhouseholdsandinthecorenetwork.
5.3 Initialkeyuncertainties
Key uncertainties are important factors with uncertain direction and impact on the
FutureInternet.Whilekeytrendsformastablegroundforscenarios,keyuncertainties
make them distinct from each other. Identifying and formulating the key uncertainties
was not as straightforward as finding key trends. The process was iterative and
consistedofthreesteps.First,theinitialkeyuncertaintiespresentedinthischapterwere
formed based on the brainstorming sessions. Second, some experts were interviewed.
During this step many importantthings were discovered. Third, the gathered feedback
wasusedtoformthefinalkeyuncertainties.Eachstepusedinthebuildupofthefinal
keyuncertaintiesispresentedonebyone.
ThemainconcerninthebrainstormingsessionsrelatedtothescalabilityoftheInternet.
Depleting address space combined to painful IPv6 migration, choking routing system,
increasingenergyconsumptionandproblematicpurposesofuselikemulticastingraised
a concern, will the Internet scale up. Collision between Internets builtin freedom and
increasingpressuretocontroltheusageformedtotheuncertaintywhethercontrolwill
increaseintheInternet.IncreasingcomplexityoftheInternetcombinedtoitscriticality
broughtupconcern,whatwouldhappeniftheInternetweretofacealargercollapse.To
summarize,theinitialmostimportantkeyuncertaintiesarelistedbelow.
Initialkeyuncertainties
1)WilltheInternetscaleup?
2)WillcontrolclearlyincreaseintheInternet?
3)WilltheInternetfacealargercollapse?
First two of these were chosen to form the initial scenario matrix (Figure 10), while a
largercollapsewasseenratherasacatalystthatcouldspeedupthechangesandthusis
ofhighinterest.
46
WilltheInternetscaleup?
Willcontrolclearly
increaseintheInternet?
Yes
No
No
Yes
Figure10:Initialscenarioaxes
5.4 Expertinterviews
Due to the rapid nature of brainstorming it was not possible to get complete
understandingofforces,especiallyofuncertainties,duringthesessions.Thusaltogether
11 experts from different fields were interviewed to deepen the knowledge. The
interviewswereconductedbetweenDecember5,2008andJanuary21,2009inHelsinki
andEspoo.Table7liststheinterviewedpersons.
Inthebeginningofeachinterviewthetopicwasintroducedbyexplainingthekeytrends.
The main focus was to confirm the selected most important key uncertainties. The
interviewswereunstructuredandtheemphasisvarieddependingonintervieweesarea
ofexpertise.Forexamplewithregulativeauthorityrepresentativesconcentrationwason
thecontroluncertainty.Althoughtheinterviewssucceededinimprovingunderstanding,
the initial purpose of confirming chosen scenario axes was not accomplished. Topics
behindtheinitialmostimportantkeyuncertaintieswereseenimportantandhadgreat
deal of uncertain elements but the formulation of them was problematic. Following
sectionsexplainwhytheinitialmostimportantkeyuncertaintiesneededtoberefined.
47
Table7:Listofinterviewees
Name
PekkaNikander
Workingtitle
Chiefscientist
Professor,
JukkaManner
Networking
technology
KlausNieminen SeniorAdvisor
Communications
KariOjala
Counselor
OssiPllnen
SeniorExpert
Manager,Future
HannuFlinck
Internet
TimoAliVehmas VPCIC
HeadofOperations,
ReijoJuvonen
Researchand
Technology
Professor,Principal
MarttiMntyl
Scientist
MattiPeltola
PhDStudent
Heikki
Professor,Network
Hmminen
economics
Organization
EricssonResearchNomadicLab
TKK/Dep.ofCommunications&Networking
FICORA
MinistryofTransportandCommunications
NokiaSiemensNetworks
Nokia
TKK/HIIT
5.4.1 Scalability
Discussionsaboutscalabilityleadtothedilemmathatalthoughmanyscalabilityrelated
issueswereseenuncertaininbrainstorming,thepossibilitythattheInternetwouldnot
scale,atleastduetotechnicalreasons,wasseennotuncertain but impossible by most
interviewees. When thought again, also the identified key trend about the societys
increasing dependency on the Internet (PT1, see Section 5.2.1) creates economic
pressurethatensuresnecessaryeffortsandinvestmentstosolvescalabilitybottlenecks.
The key discovery was that scalability may be realized in many ways, not only
implementing improvements to the entire Internet. Parallelism could be one solution,
which, however, threatens the integrity of the Internet. Altogether the most severe
scalabilityproblemsrelatetonewuseswherethebesteffortInternetisnotadequate.
5.4.2 Control
Word control caused different interpretations among the interviewees. Some of them
associatedittoregulativecontrol,othersrathertooperatorcontrol.Thusamoregeneric
48
definitionofrestrictedversusfreeInternetwasbroughtup,whichmatchedquitewellto
thespiritofthebrainstorming.Coupleofintervieweesproposedaseparateexamination
from the viewpoint of all the possible control authorities including regulator, ISP,
application service provider and consumer. This thought was processed and discarded
since the resulting worlds would have been highly diverse depending on the chosen
authority.
Indiscussionswithregulativeauthorityrepresentativesitbecameclearthatcontroldoes
not automatically mean restrictions but may as well protect freedom. This applies
especiallytoregulativecontrolwhichmaybeseenasanadjustmentleverrangingfrom
anarchy to full control. The Internet has been selfregulated, which has enabled
anarchistic behavior. Exponential growth and increasing role in todays society has
brought sideeffects that need to be controlled. In the best case regulation enables
freedomthroughreasonablesetoflaws,notanarchisticfreedom,andthussomelevelof
controlisfavorable.
Based on the interviews evolution of the control in the Internet is comparable to the
evolutionofautomotivelegislation.Drivingacarwasnearlycompletelyunregulatedin
the beginning but in the course of time when car penetration has increased also the
control has increased. This analogy is supported by the interviewees opinion that
controlwillincreaseintheInternetduringthenext10years.Henceitisjustifiedtosay
thatincreasingcontrolisatrend,anditisactuallytheresultofthisdevelopmentwhich
remainsnebulous.
5.4.3 Collapse
AlargercollapsewasdefinedasaneventthatwouldblackoutpartsoftheInternetfora
short period of time. It has significant analogies with the current financial crisis. If a
larger collapse would happen, it would probably cause severe economical losses and
reduce peoples trust on the Internet. Regardless of possible consequences many
interviewees were skeptical if a larger collapse really would have a permanent and
significantimpactontheInternetsevolution.Theexpertsarguedthatdecisionmakers
donothaveenoughknowledgeon thepossiblesolutionsandthatpeopledo noteasily
49
changetheirbehavior.Thustheeffectswouldbelimitedtofasterimplementationofnew
technical or regulatory means, increase in regulative control, and higher level of back
ups.Themostimportantnotionfromthisuncertaintyisthatthefearforcollapsedrives
forpreemptiveactionslikeimprovingresiliencyofthenetworksandservices.
5.4.4 Otherthoughts
After fruitful discussion about initial key uncertainties the experts were asked to give
some thoughts about possible scenarios sketched in Figure 10. The most problematic
scenario was the one with no scalability and no increasing control. Firstly, the
intervieweeswereseeingscalabilityandcontrolincreaseastrends,andsecondly,ifthe
Internetdidnotscaleup,thecontrolwouldcertainlyincrease.Thisopinionsignaledthat
chosenscenariovariableswerenotindependentenough.
Although the interviews did not give too much direct support to chosen scenario
variables, they succeeded in refining them. Understanding the real issues behind the
control uncertainty was one important thing but especially significant was the
repeatedly mentioned concern about fragmentation of the Internet. Most often
fragmentationwasseenasaresultof,andsolutionto,scalabilityproblemsbutexperts
also saw it as a possible outcome of increasing control and a larger collapse. Based on
thesetwokeyfindingsitwaspossibletoformulatethefinalkeyuncertainties.
5.5 Finalkeyuncertainties
The final key uncertainties are divided into two groups: the most important span the
scenario matrix while the less important add flavor to the scenarios. Additionally,
uncertaintyrelatedtothepossiblecollapseoftheInternetisdiscussedseparatelyfrom
thescenarios.AllthekeyuncertaintiesarelistedinTable8andintroducedindepthin
thenextsections.
Table8:Finalkeyuncertainties
Themostimportantkeyuncertainties
U1:Whatwillbethenetworkstructure?
U2:Whatisthelevelofopennessofcontent,applications,andhosts?
50
Otherkeyuncertainties
U3:Wherewilltheintelligencebelocated?
U4:WhatwillbethedominatingbusinessmodelintheInterneteconomy?
U5:Howwillsolutionsfortrust,securityandauthenticationbeimplemented?
U6:Willthetrafficbetreatedneutral?
U7:Theamountofstandardization:standardsvs.proprietarysolutions?
U8:Wherewillthestandardizationhappen?
Separatekeyuncertainty
U9:WillInternetfacealargercollapse?
5.5.1 Themostimportantkeyuncertainties
The most important key uncertainties were derived in an iterative process in which
interviews played a key role. The initial key uncertainty of scalability of the Internet
formed to the question of the network structure, to the level of fragmentation of the
Internet, to be exact. The control uncertainty, for one, transformed to deal with the
opennessofcontent,applicationsandhosts.
U1:Networkstructure
TheFutureInternetmayeitherremainawholenetworkoritmayfragmentintomany
networks. The characteristics of these two extremes one network vs. fragmented
networkarepresentedbyrelatingquestionslistedbelow.
Relatingquestions
WilltherebefreeconnectivityintheInternet?
WilltheInternetbeabletoscaleup?
WilltheInternetbesuitedtoallpurposesofuse?
Although the Internet consists of many different networks they still form one Internet
where, at least theoretically, every host is able to connect to every other host only by
knowingtheirIPaddresses.TheflexibilityoftheInternetprotocolsuitehasallowedthe
allIP trend meaning that the IP technology is used for various networking needs
including telephony and video services. This development underlines the possible cost
savingsthattheeconomiesofscaleenablewhenonlyonenetworkinfrastructureisused.
51
Regardlessofthesamenetworktechnology,telephonetrafficisstillseparatedfromdata
traffictoitsownnetwork.Inatrulysinglenetworkthereisnotthatkindofseparation.
Allthetrafficflowsinthesamewiresanddiverserequirementsofdifferenttraffictypes
canbetakenintoaccountatthenetworklevel.Fundamentalprerequisiteofonenetwork
tobepossiblecanbeexpressedasasloganonesizefitsall.
Fragmentation would mean that free endtoend connectivity would be questioned.
Extensive usage of NATs, firewalls and other middleboxes alike disturb already
nowadaysendtoendconnectivity.Duetotheimportanceoftheconnectivitycomplete
separationofthenetworksdoesnotseemfeasiblebuttheconnectivitymaybeheavily
restricted so that all traffic between networks travels through gateways. The
fragmentationdoesnotneedtohappeninthephysicallevelbutitcanaswelloreven
moreprobably happen intheservicelevelthroughoverlaynetworks.Theseoverlays
borrowonlytheconnectivityfromtheInternetandusetheirown,possiblyproprietary
protocols to fulfill requirements that the core Internet architecture is not capable to
satisfy.Thesesolutions,however,breaktheInternetarchitectureintentionallyandthus
increasecomplexityoftheInternetecosystem.
Scalability (from a technical viewpoint meaning a large enough address space, fast
enoughroutingprotocolsandalgorithms,andsmallenoughenergyconsumption)isone
issue that can be solved either in the level of the Internet architecture or by building
separate networks. The applicability of the Internet to every imaginable and non
imaginable purpose of use is another type of scalability issue that affects substantially
theleveloffragmentation.Forinstance,endtoendmulticastingandendtoendquality
ofservice(QoS)arenotwellsupportedbythebestefforttypeofservice.Increasingreal
time(video)trafficisoneofthoseapplicationsthathavecreateddemandforspecialized
network fragments called content delivery networks (CDNs). They are able to offer
guaranteedqualityofserviceforthosewhoarewillingtopay.Privacyrequirementsof
companies create business case for virtual private networks (VPNs) and national
security concerns may make some countries to build separate secure networks with
strongauthentication,oreventoclosetheirnetworksfromtheworldoutside.
52
U2:Opennessofapplications,servicesandhosts
Whiletheotherscenarioaxishasbasisinthenetworklayerthisonerelatestotheupper
layersoftheprotocolstack.Applications,servicesandhostsmayeitherbeopenlikePCs
andtheiropensourcesoftwareorclosedlikeApplesiPhonewithproprietarysoftware.
Afterrecognizingtheimportanceofthisuncertainty,thesamethoughtwasfoundfrom
JonathanZittrainsbook(2008).Zittrainusesthewordgenerativitytodescribethelevel
of openness. The both possible worlds open and closed are explained by relating
questionslistedbelow.
Relatingquestions
Arethehostsfreelyprogrammable?
Areuserswillingtobedependentonasingleactor?
Douserspreferbundlingorbuyingseparately?
Theworldofopenapplications,servicesandhostsistheworldofPClikemultipurpose
devices.Asingledeviceisusedtoaccessvariouskindsofapplicationsandservices,and
isabletosufficemostpurposesofuse.Successfulandopenstandardization,particularly
intheapplicationlevel,andhighavailabilityofopensourcesoftwaremeanthateveryone
hasinprinciplethepossibilitytoprogramownapplications.Closedapplications,services
and hosts, for one, are optimized for some usages (or even for a single use).
Specialization may enable better usability and fewer bugs since all the use cases are
predictable, but it restricts versatility. Security is another issue that is much easier to
takeintoaccountinclosedsystems.Actually,Zittrain(2008)seessecuritynuisancesof
opensystemsoneofthemostimportantdriversforclosedworld.
Allthecausalfactorsrelatetothequestion:whohasthecontroloverusersactions?In
theopenworlduseristhekingofthehill.TheInternetofferswideselectionofservices
anduserhasthefreedomofchoice.Hecaninstallwhicheverapplicationshewantsandis
not locked in to one service for a long period of time. Thus open world is naturally
competitive and business model wise mostly advertisementbased. Anonymity is still
possible in the Internet, which makes it more difficult to enforce copyright and IPR
(Intellectual Property Rights) regulation than in closed systems. Respectively, in the
53
closedworld,userhashandedthecontroltoanotheractor.Companiescanmosteasily
acquirethiskindofcontrolpositionthroughenduserdevicesthatcannotbechangedto
newonesasoftenasapplicationsandservices.Therearealreadyplentyofexamplesof
this kind of closed devices. For instance, Microsofts xBox 11 , like all the other game
consoles, is actually a PC performancewise but Microsoft decides, which games are
allowedtoberuninthem.Thestrongcontroloveruserscreatesopportunityforother
businessmodelsthanadbased,e.g.,subscriptionbased,tosucceed.
Users strong position in the open world has, however, a flip side responsibility.
Openness requires more purchase decisions, and users knowledge on purchase
situations needs to be higher. Also finding and installing services rely on users
competence and activity, and the same applies to security. Bundling devices,
applications,servicesandevennetworkstogetherisonemethodbywhichastakeholder
maytrytogetcustomerlockedinanddependentonsingleactor.Fromuserperspective
bundlingreducestheamountofpurchasedecisionsandmaytherebybeaneasierchoice,
especiallyfortechnologynonenthusiasts.Anotheradvantagefromcustomerperspective
is that the providers of closed systems can more easily be held accountable and
responsibleinfrontofmalfunctioning.
5.5.2 Otherkeyuncertainties
The other key uncertainties with their extreme outcomes are presented here briefly.
Theyarescaledinafivepointscalebetweenthetwoextremes(Figure11Figure16).
Later, when scenarios are presented in Chapter 6, the values for each scenario are
showed.
Originally the Internet was a dumb network connecting smart hosts. The hosts were
equal in their capabilities and roles. Clientserver model used widely in the Web
differentiated the roles of the hosts. High level of intelligence in clients indicates more
important role of peertopeer model, whereas significant amount of intelligence in
serversspeaksforclientservermodel.Itisalsoconstantlyquestionedifanyintelligence
11http://www.xbox.com/
54
shouldbeinsertedtothenetwork.Thusthequestionhereisdividedintotwoquestions:
1)clientvs.serverand2)endpointsvs.network.
Wherewilltheintelligencebelocated?
Clientvs.Server
Inclients
Inservers
Endpointsvs.Network
Inendpoints/edges
Inthenetwork
Figure11:U3Wherewilltheintelligencebelocated?
Simpleandfreeadbasedbusinessmodelhasbeenbyfarthemostsuccessfulrevenue
model when Internet services are considered. Transactionbased business model, like
payingwithPayPalorcreditcards,hasbeenmostlyusedwhenphysicalgoodsaresold
throughtheInternet.Additionally,subscriptionbasedmodelwouldbehighlyinteresting
to companies and simple enough for users. Thus the big question here is, will the
Internetbusinessbemostlyadbasedordoothermodelsbreakthrough?
WhatwillbethedominatingbusinessmodelintheInterneteconomy?
Adbasedmodel
Othermodels
Figure12:U4WhatwillbethedominatingbusinessmodelintheInterneteconomy?
U5:Howwillsolutionsfortrust,securityandauthenticationbeimplemented?
Lack of trust, security and authentication is a recognized challenge that needs to be
tackled somehow, at least in the case of missioncritical applications. Universal, open
solutionsbuiltinthearchitectureareareasonableoption,butclosedsolutionsrelating
for example to separate network or to providercontrolled solutions in closed
architectureareotherchoices.
55
Howwillsolutionsfortrust,securityandauthenticationbeimplemented?
Openlyinthearchitecture
Somewhereelseinclosedform
Figure13:U5Howwillsolutionsfortrust,securityandauthenticationbeimplemented?
Theprincipleofnetneutralityrequiresthatallcontent,sites,andplatformsaretreated
equally(Wu,2009).Inaneutralnetworktrafficflowsrelatedtoforinstanceebanking,
videostreaming,peertopeerfilesharingoremailingarenottreateddifferentlybutthey
all have same priority level from the network perspective. Blocking content and
communicationisoneofthethingsthatviolatenetneutrality.
Willthetrafficbetreatedneutral?
Yes
No
Figure14:U6Willthetrafficbetreatedneutral?
U7:Amountofstandardization:standardsvs.proprietarysolutions?
The Internet architecture relies heavily on open standards (RFCs). On the application
and service level, proprietary solutions have, however, an important role. For example
some important network overlays, VoIP network Skype and peertopeer file sharing
networkBitTorrent 12 ,arebasedonproprietarysolutions.Standardsallowcompetition,
while proprietary solutions enable emergence of monopolistic pockets. The question
hereis:willtheInternetbebasedmostlyonstandardsoronproprietarysolutions?
Amountofstandardization:standardsvs.proprietarysolutions?
Standards
Proprietarysolutions
Figure15:U7Amountofstandardization:standardsvs.proprietarysolutions?
12http://www.bittorrent.org/
56
Internetrelated issues have been traditionally standardized in the IETF. After the
commercialization, other forums have emerged including W3C (The World Wide Web
Consortium) concentrating on Web standards and 3GPP (The Third Generation
Partnership Project) working for the third generation mobile phone system.
Standardizationcould alsobedoneinindustrydrivenforumsthat would be open only
forpartoftheInternetindustry.
Wherewillthestandardizationhappen?
IETF
Industrydrivenforum
Figure16:U8Wherewillthestandardizationhappen?
5.5.3 SeparatekeyuncertaintyU9:Collapse
CollapseuncertaintywasfirstlyintroducedinSection5.3asaninitialuncertaintyafter
whichitsroleinthescenarioswasevaluatedagain.Alargercollapseblackingoutparts
oftheInternetforsomeperiodoftimecouldwakeuptheInternetcommunity,especially
decisionmakers, and thus disrupt Internets evolutional development and speed up
some changes. Depending on its nature, a larger collapse could create pressure to
whichever direction following both scenario axes (although pressure towards network
fragmentationandclosedapplications,servicesandhostsseemsstronger)meaningthat
a larger collapse will rather lead to a scenario, not vice versa. Hence the collapse
uncertainty is studied separately from the scenarios by identifying several possible
causesandtheirconsequences(Table9).
57
Table9:U9Possiblecausesforcollapse
Cause
Explanation
Averysignificantnationinthe
Terrorismor
Internet(e.g.U.S.)closesitsnetwork
cyberwar
infrontofterrorismorcyberwar.
Awormusingseveresecurityholein
"Ciscoworm" routerscausesaMorrisworm
(Reynolds,1989)likephenomenon.
Consequences
Functionalhaltforsomeperiod
oftime.Driverformultipolarity.
Internetconnectivitybreaksall
aroundtheworld.
Information
breachor
seepage
Breachorseepage(andmisuse)of
massiveamountofsensitivepersonal
informationstoredbye.g.Google,
Facebookorcreditcardcompanies.
TrustintheInternetdecreases
anditishardtogetback.
Opennessvanishes.Handling
moneyintheInternetdecreases.
Signaling
fault
Internetroutingandsignalingsystem
Levelofprotectionincreases.
ismoreandmorecomplex,whichmay
Operatorcontrolincreases.May
leadtomisconfigurationscausing
speeduptakeupofIPv6.
signalingfault.
Newtypeof
unwanted
traffic
Hugeamountsofsomenewtypeof
unwantedtraffic.
Consequencesvarydepending
onthetypeofunwantedtraffic.
Publickey
Systemsusingpublickey
Somefundamentaldeficiencyinpublic
cryptography
cryptographyneedtobeshut
keycryptographyisfoundandit
becomes
downandanewsecurity
cannotbeeasilyfixed.
unusable
solutionneedstobeadopted.
Sudden
extinctionof
IPv4
addresses
FreepoolofIPv4addressesisaboutto
beexhaustedincoupleofyears.
Hoardingofaddressesmayleada
suddenextinctionofthem.
ADoSattackonafragileandcritical
Denialof
componentoftheInternetmayblack
Serviceattack
outpartsoftheInternet.
GrowthoftheInternetis
questioned.Newwaystoallocate
IPv4addressesoradoptionof
IPv6needed.
OpennessofInternetdecreases.
Pressureforlegalharmonization
andforcatchingcriminals
increases.
TrustintheInternetdecreases.
Virusdeleting Rapidlyspreadingvirusdeleteslotsof
Peoplewakeuptodemand
lotsofdata
importantdata.
meanstoprovidebettersecurity.
Spam
Increasingamountofspammakes
emailunusable.Itisimpossibleto
separateusefulmailsfromspam.
Emailsmaybesenttofalseaddresses
ortheymaynotbedeliveredatall.
58
Betterwaytoavoidspam
needed(e.g.,authentication,
emailstamp).Otherwaysto
communicate(e.g.,IM,SMS,
socialnetworkingsites)catchon.
6 Scenarios
The final scenarios were formed iteratively based on the interviews. The two most
importantkeyuncertaintieswerechosentoformascenariomatrix.Finallythescenarios
were named descriptively to illustrate their idea immediately. The scenario matrix
includingsomedescriptivecharacteristicsofeachscenarioispresentedbelowinFigure
17.
Open
Onenetwork
Fragmentednetwork
Wild&Free
ContentdrivenOverlays
Freeconnectivity/programmability
Extremecompetition/innovation
Accessnetworksopenforall
Ad&creditcardrevenues
Consumerrules
DeviceContentBundles
Closed
Opennessofcontent,
applicationsandhosts
Networkstructure
Dedicatedpackageddevices
Devicedrivenbundling
Verticalseparation
Subscriptionrevenues
Devicevendorrules
Manyseparateoverlaynetworks
Separationinvisibletousers
Accessoperatorsasgatekeepers
Adrevenues
Contentproviderrules
IsolatedWalledGardens
Completeverticalbundle
Onestopshopping
IMStakesoff
Usagebasedrevenues
Mobileoperatorrules
Figure17:Scenariomatrix
As for the Internet evolution, it cannot be described only by two variables. Other key
uncertainties and their impact on the scenarios are presented in Figure 18. The
uncertaintiesarevaluedinthefivepointscalebetweenthetwoextremes,andtogether
withtrendstheyaddmoreflavortothescenarios.Althoughvaluesoftheuncertainties
givesomeideaaboutthescenarios,theyaredefinedinmoredetailinthenextsections.
First, the descriptive narratives about the Internet in 2018 are presented for each
scenario. This is followed by explaining the differences between the scenarios in
architecturalstructureandinvaluedistributioninthenextchapter.
59
1=Wild&Free
2=ContentdrivenOverlays
3=DeviceContentBundles
4=IsolatedWalledGardens
Inclients
Clientvs.Server
Inservers
Endpointsvs.Network
Inendpoints/edges
Inthenetwork
o
1
3
2
4
Adbasedmodel
Othermodels
U5:Howwillsolutionsfortrust,security,andauthenticationbeimplemented?
Openlyinthearchitecture
Somewhereelseinclosedform
Yes
2,3
No
U7:Amountofstandardization:standardsvs.proprietarysolutions?
Standards
Proprietarysolutions
IETF
1,2
Industrydrivenforum
Figure18:Uncertaintiesvaluedinfivepointscale
6.1 Wild&Free
In the wild and free Internet a multitude of services and applications are offered in a
single network. The Internet architecture ishourglassshaped like the original Internet
architectureandinsomesenseWild&FreemeansreturningtotherootsoftheInternet,
where peertopeer communications is really important. Users have versatile, freely
programmabledevicesthatcanconnectfreelytoeachotherandtoanyavailablecontent,
service and application. Free connectivity has reached its ultimate level since access
networkshavebeenopenedandalluserscanconnecttotheInternetthroughwhichever
60
networktheywant.13 Consumersarekingswhoenjoytheeverincreasingsupplyofnew
servicesandarenotwillingtoconstrainontheserviceofferingofasinglecompany.Due
toabsenceoflockinsconsumerscaneasilyswitchfromaservicetoanother,ifabetter
servicebecomesavailable.Thusitisdifficultforotherbusinessmodelstocompetewith
advertisementbased model offering the service for free to the customers. In some
cases,however,creditcardbasedpaymentsareused.
ThecurrentandforthcomingscalabilityproblemshavebeensolvedsothatInternetcan
handlebothincreasingamountofusers,hostsandtrafficaswellasoldandnovelusage
scenarioswithdiverseservicelevelrequirements.Someexamplesofsuccessstoriesare
eventual adoption of IPv6 and implementation of endtoend multicast and endtoend
QoSintheprotocollevel.Successinsolvingthecoreproblemsintechnologicaldomain
hasallowedregulatortoconcentrateonenablingcompetitionandinnovationinthefree
markets.ThelastmonopolisticpocketintheInternet,accessnetworks,isbroughtdown,
andthustheindustryiscompletelyhorizontalandextremelycompetitive.
The IETF has come to its own and standardization and open source software are seen
important.Trust,securityandauthenticationareimplementedopenlyinthearchitecture
butonly concerningthoseapplicationsthatrequirethem.Dueto free programmability
and imbalance between regional regulations, unwanted traffic remains as a serious
problem and the race between malware makers and security companies continues.
Additionally, anonymity in the Internet is still possible, although tracking of (hostile)
usersiseasierthannowadays.
6.2 IsolatedWalledGardens
Isolatedwalledgardensareaccessoperatorcentricnetworkfragments 14 whichbundle
all the components of the value chain devices, network and content together. The
complete vertical bundle means that accessoperators have control overend usersand
content.Devices,softwareandInternetconnectionaresoldatthesametimebyaccess
operatorswhoalsotakethe responsibilityformanagingthecompletepackageremotely.
13Thiscouldberealizedforexamplebysocializingtheaccessnetworks,introducingglobalauthentication
orextendingadbasedbusinessmodeltoaccessnetworks.
14Alsosomecountriesmaybuildisolatedwalledgardenswhichcoverthewholecountry.
61
This decreases the amount of (purchase) decisions needed and makes life easy for
consumers. Customer experience and security are optimized by letting consumers to
install only those applications and use only those services that access operators have
allowed to be offered in their networks. Restrictions in installing applications leads to
thesituationinwhichmostoftheservicesarerunningontheserversofthe providers
whoneedcloudcomputingstyleofsolutionstoscaletotheincreasingusage.
TechnologywiseIsolatedWalledGardensistheIMS(IPMultimediaSubsystem)world.
Refinedusagebasedmonthlybillingispossibleandisusedextensivelylikeinthetelco
world.Thedominantroleofaccessoperatorsmeansthatthelargestamountofprofitsis
flowingintheirpockets.Japanesemobiletelephonymarket(e.g.NTTDOCOMO 15 )isan
existingexampleofstrongoperatorcontrolledvaluechainwhichresemblestheIsolated
WalledGardenscenario.
Some of the network fragments are interoperable, while others are isolated by design.
However, interworking between competing access operators is always separately
negotiated,whichmeansmoreproprietarysolutionsandstandardizinginterfacesonly.A
strong industry forum driven by access operators has replaced IETFstyle of
standardization. This change combined with restricted programmability results in the
situationwhereopeninnovationandenteringtheInternetbusinessbecomesmuchmore
difficult leading to oligopolistic markets and higher prices. Internet pioneers and
academics are longing for thegoodoldInternet, while many users are satisfied with
bettersecurity,qualityofserviceandtrustbetweentheusers.
6.3 ContentdrivenOverlays
BecausethecommonInternetarchitecturehasnotbeenabletosupportallthevarying
service requirements of different applications, the network is fragmented into overlay
networksbasedoncontenttypeandapplication.Companies,mostlycontentproviders,
offerbetterqualityofserviceindedicatedservicenetworkslikevideostreamingCDNs.
At the same time governments and companies have built secure networks for critical
functions of the society. Although all these overlays are built on the top of IP, they are
15http://www.nttdocomo.com/
62
optimizedforcertainuseandhavemuchtightercontroloverthetrafficgeneratedbyend
users. To achieve better performance and conformance to standards, the common
Internet architecture is intentionally broken when needed, which increases the
complexityofthenetworks.
Building of dedicated networks is paid by competing content providers which try to
attractasmanycustomersaspossibletomaximizeadvertiservalue.Thisputscompanies
inunequalpositionsincethoseproviderswhichareabletopayforbetterquality,getit,
andthosewhichcannotafford,mustcontentthemselvesonpoorerquality.Thisfavors
large players which can attract more customers and thus more advertisers by offering
betterservices.Fromconsumerpointofviewthesituationisgoodsincefragmentationof
thenetworksisinvisibletothem,andtheyjustenjoybroadvarietyoffree,goodquality
services.Furthermore,regulatorsfightforopencompetitioninthemarkets,whichkeeps
prices low, decreases companies possibilities to get customer lockin, and favors ad
based business model. Although content providers skim the cream off the cake, access
operatorshaveimportantroleinforwardingthetraffictotherightoverlaynetwork.
Ifoptimizednetworkfragmentsfullytakeover,thedevelopmentofthebasic,besteffort
Internet may stop since the concentration is on contentdriven overlays. Nevertheless,
the network can also defragment, if an overlay (e.g., information networking or social
networking)solvesthelargestproblemsthatleadtothefragmentation.
6.4 DeviceContentBundles
WhileInternetusagemovesincreasinglyfromPCstomobilephonesandotherportable
devices, users interest ininstalling applications and updating their devices themselves
decreases.Atthesametimedevicevendorsoffertemptingdevicecontentbundleswhich
combine devices and services in a seamless manner. Due to their unbeatable user
experience, includingenhanced reliability and security,and hyperusability, consumers
arewillingtoacceptlockin.TheInternetisstillwholebutconsumerschoosewhichpart
oftheserviceofferingisavailabletothemwhentheypurchasetheirdevices.
63
Updatingandinstallingnewservices ispossible onlythroughdevicevendorcontrolled

serviceportals(likeApplesAppStore 16 andNokiasOvi 17 )meaningthatdevicevendors
take their share of all the purchases. Restricted ability to install applications raises the
roleofwebapplicationsandclientservermodel,whichchangesuserdevicesfromactive
participantstomerelypassiveterminalsandincreasestheimportanceofserviceclouds.
This development leads to closed and more dedicated devices, resembling todays
Xboxes and iPhones 18 . Device manufacturers control enables subscriptionbased
revenuemodelsinwhichdevicepriceincludesaccesstocontentandservicesforsome
periodoftime.NokiasComeswithMusichandsets 19 areanearlyexampleofthiskindof
innovativebundle.
16http://www.apple.com/iphone/appstore/
17http://ovi.nokia.com/
18http://www.apple.com/iphone/
19http://comeswithmusic.com/
64
7 Comparisonofscenarios
In the previous chapter the scenarios were described one by one. In the following
sections they are compared in terms of technical and business architecture and value
distributionbetweendifferentstakeholders.
7.1 Architecture
The architectural differences of the scenarios are illustrated in Figure 19. The
presentation is simplified and takes into account only content, network, and end user
devices. Content covers not only textual, audio and video data but also services and
applications. The architecture can be understood both as business and as technical
architecturewhichareuniforminthescenarios.
Wild&Free
ContentdrivenOverlays
DeviceContentBundles
IsolatedWalledGardens
Figure19:Simplifiedbusinessandtechnicalarchitecturesinscenarios
Difference in network structure separates the lefthand side scenarios from the right
hand side scenarios. Openness of applications, services and hosts determines how
devicescanaccessthecontent.ForexampletheWild&FreeandDeviceContentBundles
scenariossharethesamearchitecturewiththeexceptionthatinDeviceContentBundles
65
thedeviceinuse(andultimatelythedevicevendor)defineswhichpartofthecontentis
accessible.
In the Contentdriven Overlays the content and network are bundled together, but
devices can connect to every network, whereas in the Isolated Walled Gardens all the
elements device, network and content are bundled together. Although the walled
gardens are isolated, communication between users belonging to different networks is
restrictedlypossiblethroughaccessoperatorcontrolledgateways.
7.2 Valuedistribution
The power positions of the most important stakeholders are illustrated by presenting
value distribution in the scenarios. Figure 20, inspired by Christensen et al. (2001),
depicts in a simplified manner to whose pockets the profits are flowing in. Device
vendors include enduser device manufacturers like Nokia, Apple and Dell. Content
providers cover software companies (Microsoft, SAP), Internet service giants (Google,
Yahoo), media houses (NBC, BBC) and entertainment companies (Disney, Universal).
Access operators include traditional ISPs like Comcast and Verizon as well as mobile
operatorslikeVodafoneandOrange.Backboneprovidersandnetworkinfravendorsare
neglected in the figure since their profits are not analyzed deeper, although they may
varybetweenscenarios.
The presentation is qualitative, although the money is presented as amount of coins.
Thus only the ratio of coins inside a scenario and between the scenarios should be
examined. For example, the DeviceContent Bundles scenario is more profitable for
devicemanufacturersthanforaccessoperatorswhiletheContentdrivenOverlaysoffers
bestprospectsforcontentproviders.Oneimportantissuetoclarifyis the difference in
the amount of coins between theWild & Free and other scenarios. Due to the extreme
competition covering every business sector the profits are smaller than in other
scenariossincealargerpartofthemoneyremainsinconsumerspockets.
66
Figure20:Simplifiedvaluedistributioninscenarios
Which scenario is best for the society? Regulators goal is to maximize social welfare
whichiscompoundedofcustomersurplusandproducersurplus(Courcoubetis&Weber,
2003). Although it is difficult to say, which of the scenarios would be optimal from
regulators perspective, some speculation is still possible to carry out. The shortterm
societal optimum is close to users momentary optimum which may be against of
industryplayerssuccess.However,regulatorstypicallytrytoseefarinsteadofstriving
for shortterm user value maximization. Thus the functioning of the whole Internet
industryisofkeyimportanceandregulatorsneedtotakeintoaccountthewelfareofall
thestakeholders.
At least from shortterm economic perspective Wild & Free seems to be favorable
scenario because, due to extreme competition, customers costs are smallest. Anyway,
Wild&Freecouldbethebestscenarioalsointhelongterm,sinceatleastinthepastthe
open and horizontal market (and network) structure has been highly successful in the
Internet. Furthermore, characteristics of Wild & Free are typically linked to the fierce
67
pace of innovation which is seen as a key ingredient in chasing the societal optimum.
Innovation is naturally possible also in the other scenarios but higher market entry
barriersanddominantplayerscontroloverinnovatorshinderperfectcompetitionand
full use of innovation possibilities. Completely vertical industry structure of Isolated
WalledGardensseemstobeespeciallyundesirablefromthisperspective.
Ontheotherhand,theotherscenariosthanWild&Freemaybeabletooffermorevalue
to users through better customer experience. Thus the societal welfare can also be
reached in these scenarios but the regulators task is more difficult since more
restrictionstopreventemergenceofmonopolisticpocketsinthemarketareneeded.If
thetwoscenarioaxesarecomparedtoeachother,theuncertaintyconcerningopenness
of applications, services and hosts has higher impact on social welfare than level of
fragmentation has. Because of this, Contentdriven Overlays can be seen as a slightly
betterscenariofromregulatorsperspectivethanDeviceContentBundles.
68
8 Fromscenariostoresearchstrategy
Converting scenarios into strategy is a natural step after the scenario construction.
Normally one stakeholder, either a company or an industry branch, is chosen and
strategic suggestions to cope with each scenario are given. Based on the strategic
analysisthestakeholdercaneitherprepareforeachscenarioortrytouseits(market)
powertopushthroughthemostfavorablescenario.
The key stakeholder in this exercise is the Finnish Future Internet (FI) program. It
consistsofmultiplestakeholdersthatprobablydonothaveacommonopiniononeach
topic. Due to this the task is somewhat different compared to planning strategy for a
singleindustryplayer.Hencethestrategicsuggestionsarelimitedtoresearchstrategy.
Thegoalistodescribethekeyresearchtopicsineachscenario.Theideaistojumptothe
year2018,imagineonescenarioatatime,anddescribewhichdevelopmentstepshave
beenparamount.Byanalyzingwhichchallengeshaveneededtobeaddressedinorderto
ascenariotocometrue,itispossibletosuggestthekeyresearchtopicsforeachscenario.
Next two sections first describe the current research plan of the FI program, and then
present the key research topics for each scenario and their appearance in this plan.
Furthermore, some suggestions, how FI program should provide for each scenario, are
given.Expertswerenotinterviewedinthispartoftheworkandsothefollowinganalysis
isbasedonlyontheselfobtainedunderstandingofthescenarios.
8.1 ResearchplanoftheFutureInternetprogram
Theresearchthemesarestructuredintosixworkpackages(WP0WP5)andtosixcross
issues like illustrated in Figure 21 (Juvonen, 2008). Thefocus is on three main themes
studied in the work packages 13, namely the health of the Internet routing system
(WP1),explorationofwaystoimprovethequalityofendtoendconnectivity(WP2),and
investigation of new ways of information storage and delivery (WP3). Additionally,
several other important research topics with broader scope are covered by the cross
issues.
69
The program looks at solutions with different time scales (Nikander & Mntyl, 2008).
Shorttermresearchtopics(12years)haveaforeseeabledeploymentplaninnextfew
years, whereas the main focus of FI program is in medium term research giving
applicableresultsin35years.Longtermresearch,forone,providestheoverallvision,
and answers to the question where do we want to be in 10 years by also suggesting
morefundamentalchangestotheInternetarchitecture.
Figure21:WorkpackagesandcrossissuesofFIProgram(Juvonen,2008).
8.2 Identifiedresearchtopics
Table10listssomeresearchtopicswhichwereidentifiedtobeimportantatleastinone
depicted scenario. The concise list highlights some topics brought up during the
research.Theimportanceofeachresearchtopicineveryscenarioisevaluatedinthree
level scale highmediumlow and the possible place in structure of the FI program is
definedinthelevelofworkpackage(WP)orcrossissue(Cr.I.).
As can be seen, some of the research topics are equally important in every scenario.
These topics stem from trends, like the first two topics in the list that tie up with the
trendsPT4andTT1(seeSection5.2).Trendbasedresearchissecureandeasilyarguable
although achieving unique breakthrough results may be more difficult since the whole
70
research community is wrestling with the same problems. The importance of other
research topics varies more between the scenarios. Some of them have background in
trendsbutmostofthemrelaterathertouncertainties.Forexamplesomeresearchtopics
aremoreimportantinfragmentednetworksandsomeinonenetwork.Itisalsoclearly
noticeable that majority of the research topics is important in the Wild and Free
scenario,whereasfarfewertopicshavehighimportanceintheIsolatedWalledGardens
scenario.Thiscanbeinterpretedsothatmanyresearchtopicsrequiremoreattentionin
thewildandfreeworld,wherethesolutionsneedtobeimplementedmainlyinthecore
architecture,thanintheworldofisolatedwalledgardens,wherethesolutionscanalso
be implemented either in individual network fragments or by closing the applications,
servicesandhosts.
Table10:Importanceofthechosenresearchtopicsinthescenarios
Researchtopic
Energyconsumption
Mobilityandmultihoming
Privacyvs.Accountability
Cognitiveradio
Universal(3rdparty)authentication
Traffic&congestioncontrol,load
balancing
Lawfulandfairuseofresources
Configurationagility
Informationnetworking
CleanslateInternet
Routingscalability
Securityandtrustopenlyinarchitecture
PeertoPeer
DHCPstylenetworkselection
Interconnectivitybetweennetworks
BuildingseparatesecureInternet
Cloudcomputing
IPMultimediaSubsystem(IMS)
Importanceinscenarios
PlaceinFI Wild&
program
Free
Content Device Isolated

driven Content Walled
Overlays Bundles Gardens
WP2,Cr.I.
WP2,Cr.I.
Cr.I.
Cr.I.
High
High
High
High
High
High
High
High
High
High
WP2
High
Medium
High
Medium
WP2
WP3
WP3
WP1
Cr.I.
High
High
High
High
High
High
High
Low
Low
Low
Low
Medium
Medium
Medium
Medium
Medium
Medium
Medium
High
High
High
High
Medium
Medium
Medium
High
High
Medium
Low
Low
Low
Low
High
Low
Low
Low
Low
Low
Low
Low
Medium
Medium
Medium
Medium
Low
Medium
Low
High
71
High
High
High
High
High Medium
High
Low
Medium
Low
8.3 FutureInternetprogramandscenarios
ThissectionsuggestshowFIprogramcouldpreparetoeachscenario.Scenariosaregone
throughonebyoneandtheanalysisrestsontheidentifiedresearchtopics(Table10).
8.3.1 Wild&Free
The Wild & Free scenario is close to the traditional IETF ideology. All the critical
shortcomingsaresolvedinthecoreInternetarchitecturewhichisuniformandrobust.
Since FI program has heavy emphasis on IETF, this scenario is closest to the
program.Table10suggestsalonglistofresearchtopicsimportantinthisscenarioand
mostofthemarealreadyaddressedbytheprogram.TheproblematictransitiontoIPv6
in mind it is arguable that making changes to the core architecture is difficult, even
thoughthetechnicalsolutionswouldbereadyandofferundisputablebenefits.Thusthe
deployment of the derived results needs to be planned already from the start, if the
intentoftheprogramistobesomethingmorethanjustanacademicexercise.Backwards
compatibilityplaysakeyrolesincetheovernighttransitiontoanewarchitectureisnot
possible. Clean slate approach of information networking may be inevitable for Wild &
Freetocometruebutitsdeploymentneedstobeevenmorecarefullyplannedthanthe
deploymentofincrementalchanges.
8.3.2 IsolatedWalledGardens
The FI program has not prepared particularly well to this scenario. Finnish access
operatorsaresmallcomparedtothegiantslikeVodafoneandtheydonothavepowerto
pushforthisscenarioeventhoughtheycertainlywouldbeinterested.However,network
device vendors, especially those who have strong position in mobile access networks,
maywelcomethiskindofworldwhichfavorsplayerswithtelcobackground.
In Isolated Walled Gardens majority of the problems is solved by creating closed and
more controlled networks. In this kind of environment most of the current and
foreseeableproblemsrelatingforexampletoscalability,securityandtrustareeasierto
solve and this can be partially done by existing means. Hence the amount of research
needed is smaller and large part of it is carried out by access operators. The research
topics relate mostly to interconnectivity and interfaces between diverse network
72
entities.Thecomplexityofanisolatedwalledgardenmaybesmallerthanthecomplexity
of a single network but the overall complexity increases with the amount of isolated
walledgardens.
8.3.3 ContentdrivenOverlays
FIprogramisnotbestpositionedtodeploy contentdrivenoverlaysbecausethelistof
program participants (and Finnish industry in general) is missing large content
providers which would benefit from research results that enable overlays. FI program
couldeitherstartcooperationwithlargecontenthousesorpositionitselfasadeveloper
ofanIETFcentricandproviderindependentplatformwhichallowsbuildingofvariable
specified overlays through flexible tuning. WP3s work for information (or content
centric)networkingsuitswelltothisscenariowhichactuallyoffersacleardeployment
path for such clean slate approach. Another interesting research problem is how the
hostscanchoosearightoverlayforrightpurpose.ForinstanceaDHCPlikemechanism
fornetworkselectioncouldsolvetheproblem.
8.3.4 DeviceContentBundles
This scenario is close to the heart of Nokia, a key stakeholder of the FI program.
Interestingly,however,nospecialattentionispaidforthisscenario.Fortunatelythisis
notalargeproblemsinceduetothearchitecturalsimilaritytheresearchtopicsresemble
thoseoftheWild&Freescenario.Altogetherscenarioaxisbasedonnetworkstructure
separates the (mostly architecturecentric) research needs more than the more
applicationoriented axis of openness. However, to make out in this scenario the FI
program should concentrate more on developing lucrative business models and
unbeatableuserexperienceintheworldofclosedandmorelimitedhosts.Becauseofthe
highimportanceofserversideoperationorcloudcomputingcooperationwithrecently
announced Supermatrix project is an important and natural possibility since many
stakeholdersparticipateonbothprojects(Finnet,2009).TheSupermatrixprojectaims
at bringing 100megabits connections to homes and providing the whole desktop as a
servicebymovingexecutionofapplicationsandstorageofdatatolocalsupercomputers.
73
9 Conclusion
This chapter concludes the key findings, discusses the limitations of the study and
exploitationoftheresults,andfinallysuggestssometopicsforfutureresearch.
9.1 Keyfindings
Oneofthekeyfindings,eventhoughanobviousone,isthatinthefuturetheInternetwill
beacriticalinfrastructuralpartofthesociety.Theincreasingimportancediversifiesthe
stakeholdercorpsandchangesthecooperativeplayfieldintoatussleinwhichpolitical,
economic and social motives are at least as important as technical enablers. The
conductedstudyonhistoricalmilestonesproposesthatthesuccessstoryoftheInternet
is based on the flexibility of Internet architecture and the capability of Internet
community to introduce technical, bottleneckremoving solutions as an answer to the
increasingdemandanddiversifiedapplicationrequirements.FortheInternetdependent
societyoftomorrowsolvingofemergingchallengesisevenmorecrucialbuttheworkis
moredifficultandthepossiblesolutionsmoreunsureanddiversethanbefore.
Scenario planning process bounds the uncertainty through identifying the key trends
and, foremost, the key uncertainties. The uncertainty revolving around the Future
Internetisdefinedbytwoindependentkeyuncertainties:networkstructurereferringto
the level of network fragmentation, and openness of applications, services, and hosts
relating to who has the control over usage possibilities. The different outcome
combinationsfromtheseuncertaintiescreatefourscenariosthatpresentandsummarize
thealternativefutures.
The first scenario, Wild&Free, means transition back to the original Internet in which
freeconnectivityandfreeprogrammabilityprevailandtheInternetindustryisentirely
horizontalandhighlycompetitive.TheIsolatedWalledGardensscenariodescribesafully
opposite world. The Internet is fragmentedintoaccessoperator controlledislandsand
allthecomponentsdevices,network,andcontentarebundledtogetherthroughthe
vertical industry structure. The two latter scenarios present different shades of gray
betweentheseblackandwhiteworlds.InContentdrivenOverlaysnetworkisfragmented
intomultipleoverlaynetworksbasedonthecharacteristicsofcontenttypes.Contraryto
74
IsolatedWalledGardenstheseparationisinvisibletousersthatcanaccessanynetwork
content bundle available. In DeviceContent Bundles, for one, network is technically
similar to the network in Wild & Free but the strong device vendor control drives for
simple,closeddevicesthatrestrictthechoiceofapplicationsandservices.
The most important finding derived during the scenario process is that the challenges
the Internet is encountering can be solved at many levels. As the scenarios propose,
there are three possible ways to satisfy emerging requirements and solve forthcoming
bottlenecks.First,IETFstrivesforopennessandnetworkintegritypreservingsolutions
implemented in the core Internet architecture. Second, business interests of various
stakeholders and failures in implementing some features in the core architecture pave
the way for solutions that rely on building separate network fragments or closing the
applications,services,andhosts.Third,acompletelynewcleanslateapproachmightbe
able to solve a bunch of problems, although the deployment conditions of a complete
redesignarenotstudiedcloserinthisthesis.
The strength and relevance of uncertaintybased scenarios depend on the match of
technicalandbusinessarchitecturesineachdistinctivescenario.Eachproducedscenario
corresponds to a characteristic technical and business architecture, which may trigger
valuable forwardlooking debates among experts. Differences between the constructed
scenarios concerning the power relationships and value distribution between
stakeholdersrevealtheunderlyingtensionsanddifferinginterestsofstakeholders.This
suggests that the actual outcome ten years from now will likely be a hybrid of several
scenarios.
Scenario planning should stimulate decision makers to consider changes they would
otherwiseignore.AshortanalysisconcerningtheresearchstrategyoftheFinnishFuture
Internetprogramrevealsthattheresearchfocusisclearlyinthecorearchitectureandin
the Wild & Free scenario. The other scenarios are seen resulting from failures of the
IETFcentricresearch.Thekeyfindingfromthestrategicanalysisisthatdeploymentof
newsolutionsneedstobeplannedcarefullyalreadyfromthebeginning.
75
9.2 Discussion
The main purpose of the conducted scenario analysis is to clarify possible future
advances in the Internet evolution. It limits the amount of uncertainties, generates
consistent pictures, and identifies issues that might otherwise remain ignored. The
constructedscenariosdepictfourdistinctfuturesandoftenwhenpresentingtheresults
to audience, I was asked to evaluate odds for their realization. This thesis, however,
deliberatelyrefrainsfromassigningprobabilities.Theonlyanswer that can be given is
thatiftheanalysishassucceeded,theInternetin2018canbedefinedbytheconstructed
scenariospaceeitherasoneofthescenariosbutmoreprobablycontainingelements
frommultiplescenarios.
Due to the chosen scope, the scenarios mainly present the final states leaving the
description of paths and turning points leading to the scenarios to smaller attention.
Becauseofveryholisticframingoftheresearchquestionsthe scenariostaketopdown
approachwhichlimitstheirtechnicalprecision.Acrosstheboard,nospecifiedtechnical
solutions are presented and finding of them remains a task of later research.
Furthermore,althoughthescopeoftheresearchwasglobal,theresultingperspectiveis
inevitablyFinlandcentricandmaybebiasedtowardsissuesimportantforparticipated
experts.
An important contribution of this thesis is that it guides the discussion about Future
Internet. During the construction process the scenarios were not looked from a
perspectiveofagivenstakeholder,whichmadetheprocessmoredifficult.Nevertheless,
the neutrality of the scenarios means that the results can be exploited in planning the
strategic actions of any stakeholder. Presented analysis of the research strategy of FI
program(seeChapter8)isjustanexampleofthepossibilitiestheconstructedscenario
analysisoffersofhelpformanagerialdecisions.
9.3 Futureresearch
Thisthesishasincreasedunderstandingofthebigpictureofhistoricalandforthcoming
evolutionoftheInternet.Thenexttaskistodivedeeperandconcentrateontopicswith
narrowerscope.Aspresentedinthediscussionpart,analysisoftheturningpointsthat
76
changetheevolutionpathtowardsacertainscenariocouldbeenhancedbyanalyzingthe
scenarios from multistakeholder perspective. One interesting topic could be further
analysis on the conflicting interests and motives between the stakeholders from
economic perspective. On the whole, the future research could keep technical view at
armslengthandconcentratemoreoneconomic,socialandpoliticalviewpoints.Analysis
onpoliticalandregulatoryprerequisitesforthescenariosoriginallyapartofthisthesis
but omitted in order to narrow the scope could reveal the required change in the
politicalclimatetoenablethescenariostocometrue.
Quantitative research is an essential and natural followup that can increase
concreteness of the scenario analysis. Because it is impossible to get quantitative data
fromthefuture,othermethodsneedtobeused.Modelingofhistoricaldevelopmentand
extrapolatingderivedresultstothefutureisonepossibleoption.Fromtechnoeconomic
perspective, price ratio changes between different technologies can have fundamental
impactonarchitectureevolutionandallocationofrolesinthevaluenetworks.Analyzing
forexamplethepriceratioevolutionbetweentransportandstoragecouldgiveacue,if
thefutureoftheInternetisbasedmoreontransportorstorageofdata.Systemdynamic
modelingisanotherstepthatcanbetakenbasedonthecreatedscenarios.
All in all, the Internet has changed the society more and faster than any other
developmentduringthelastcenturybutitspotentialisnotyetcompletelyunleashedor
even understood. Produced scenarios shed light on the multifaceted future research
topics by illustrating interesting visions of the future. Scenarios, however well
constructed,canembodyonlyafinitesetofphenomena.Therebytheactualevolutionof
theInternetcanbestudiedonlybyfollowingreallifeevents.
77
References
Aguilar,F.(1967),Scanningthebusinessenvironment.NewYork:Macmillan.
Alexa.com(2009),Thetop500sitesontheWebonApril2nd,2009.
http://alexa.com/topsites(accessed2April2009).
Alvestrand,H.(2004),TheRoleoftheStandardsProcessinShapingtheInternet,
ProceedingsoftheIEEE,vol.92,pp.13711374.
ASN(2009),The32bitASNumberReport.
http://www.potaroo.net/tools/asn32/(accessed14April2009).
Baran,P.(1964),OnDistributedCommunicationNetworks,IEEETransactionson
CommunicationsSystems,vol.12,pp.19.
Beijar,N.(2008),PeertoPeerArchitectureandSignaling,LectureslidesoftheSignaling
ProtocolscourseonApril4,2008.
http://www.netlab.tkk.fi/opetus/s383115/2008/kalvot/P2P_lecture.pdf
(accessed8December2008).
BernersLee,T.,Cailliau,R.,Luotonen,A.,Nielsen,H.F.,Secret,A.(1994),TheWorldWide
Web,CommunicationsoftheACM,vol.37,pp.7682.
BernersLee,T.(1996),WWW:Past,Present,andFuture,Computer,vol.29,pp.6977.
Bradfield,R.,Wright,G.,Burt,G.,Cairns,G.,VanDerHeijden,K.(2005),Theoriginsand
evolutionofscenariotechniquesinlongrangebusinessplanning,Futures,vol.37,
pp.795812.
CacheLogic(2005),Internetprotocoltrends19932004,inP2Pin2005.
http://www.readwriteweb.com/archives/p2p_growth_trend_watch.php
(accessed19April2009).
CacheLogic(2006),Internetprotocoltrends19932006,inP2PFileSharingThe
EvolvingDistributionChainpresentationofDavidFerguson.
http://www.dcia.info/activities/p2pmswdc2006/ferguson.pdf(accessed19April
2009).
CAIDA(2009),ChicagopassivenetworkmonitorA.
http://www.caida.org/data/realtime/passive/?monitor=equinixchicagodirA
(accessed18April2009)
Carpenter,B.(1996),RFC1958ArchitecturalPrinciplesoftheInternet.
http://www.ietf.org/rfc/rfc1958.txt(accessed20March2009).
78
Cerf,V.,Kahn,R.(1974),AProtocolforPacketNetworkIntercommunication,IEEE
TransactionsonCommunications,vol.22,pp.637648.
Christensen,C.M.,Raynor,M.,Verlinden,M.(2001),Skatetowherethemoneywillbe,
HarvardBusinessReview,vol.79,pp.7281.
Cisco(2008a),VisualNetworkingIndexForecastsandMethodology,20072012.
http://newsroom.cisco.com/dlls/2008/ekits/Cisco_Visual_Networking_Index_06
1608.pdf(accessed5April2009).
Cisco(2008b),VisualNetworkingIndexApproachingtheZettabyteEra.
http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/
ns827/white_paper_c11481374.pdf(accessed19April2009).
Clark,D.(1988),ThedesignphilosophyoftheDARPAInternetprotocols,ACMSIGCOMM
Computercommunicationreview,vol.18,pp.106114.
Clark,D.,Chapin,L.,Cerf,V.,Braden,R.,Hobby,R.(1991),RFC1287TowardstheFuture
InternetArchitecture.
http://www.ietf.org/rfc/rfc1287.txt(accessed1April2009).
Clark,D.,Blumenthal,M.(2001),RethinkingthedesignoftheInternet:theendtoend
argumentsvs.thebravenewworld,ACMTransactionsonInternetTechnology
(TOIT),vol1,pp.70109.
Clark,D.,Sollins,K.,Wroclawski,J.,Braden,R.(2002),TussleinCyberspace:Defining
TomorrowsInternet,ProceedingsofACMSIGCOMM2002,pp.347356.
ClarkD.,Blumenthal,M.(2007),Theendtoendargumentandapplicationdesign:therole
oftrust,ProceedingsoftheConferenceonCommunication,Informationand
InternetPolicy(TPRC),Sept.2007
Clark,D.,Faratin,P.,Gilmore,P.,Bauer,S.,BergerA.,andLehrW.(2008),TheGrowing
ComplexityofInternetInterconnection,Communications&Strategies,No.72,4th
quarter2008,pp.5171.
Courcoubetis,C.,Weber,R.(2003),PricingCommunicationNetworks,JohnWiley&Sons.
Crocker,S.(1970),RFC36ProtocolNotes.
Diehl,M.,Stroebe,W.(1991),Productivitylossinideageneratinggroups:Trackingdown
theblockingeffect,JournalofPersonalityandSocialPsychology,vol.61,pp.392
403.
79
Egevang,K.,Francis,P.(1994),RFC1631TheIPNetworkAddressTranslator(NAT).
EllacoyaNetworks(2007),TheYouTubeeffect:HTTPtrafficnoweclipsesP2P.
http://arstechnica.com/news.ars/post/20070619theyoutubeeffecthttp
trafficnoweclipsesp2p.html(accessed5April2009).
EuropeanCommission(2008),BledDeclaration.
http://www.futureinternet.eu/publications/bleddeclaration.html(accessed27
March2009).
Fahey,L.,King,W.(1977),EnvironmentalScanningforCorporatePlanning,Business
Horizons,vol.20(4),pp.6171.
Fallows,D.(2005),SearchEngineUsers,reportofPewInternet&AmericanLifeproject.
http://www.pewinternet.org/~/media//Files/Reports/2005/PIP_Searchengine_
users.pdf.pdf(accessed2April2009).
Finnet(2009),Finnetstartsamajorparadigmshiftincomputinginvolvinglargestlocal
andglobalsystemsuppliers,bulletinon17April2009.
http://www.finnet.fi/index.asp?id=116&menupath=116&language=1&klik=1
FNC(1995),DefinitionofInternet,U.S.FederalNetworkingCouncil.
http://www.nitrd.gov/fnc/Internet_res.html(accessed19March2009).
Gill,P.,Arlitt,M.,Li,Z.,Mahanti,A.(2007),Youtubetrafficcharacterization:aviewfrom
theedge,Proceedingsofthe7thACMSIGCOMMconferenceonInternet
measurement,pp.1528.
Hain,T.(2000),RFC2993ArchitecturalImplicationsofNAT.
http://www.ietf.org/rfc/rfc2993.txt(accessed11May2009).
HeikkinenM.,Hmminen,H.(2008a),ScenarioPlanningofMobilePeertoPeerService
Usage,Proceedingsof7thInternationalConferenceonMobileBusiness(ICMB
2008),pp.145152.
Heikkinen,M.,Matuszewski,M.,Hmminen,H.(2008b),Scenarioplanningforemerging
mobileservicesdecisionmaking:mobilePeertoPeerSessionInitiationProtocol
casestudy,Int.J.InformationandDecisionSciences,vol.1,pp.2643.
Hinden,R.(1993),RFC1517ApplicabilityStatementfortheImplementationofClassless
InterDomainRouting(CIDR).
80
Huston,G.(2001),AnalyzingtheInternetsBGProutingtable,TheInternetProtocol
Journal,vol.4,nr.1.
Huston,G.(2006),IdentityandLocationinIP,presentationinCAIASeminar,Swinburne
UniversityofTechnology.
http://www.potaroo.net/presentations/20061130whoareyou.pdf(accessed16
April2009).
Huston,G.(2008a),TheEndofEndtoEnd?
http://www.potaroo.net/ispcol/200805/eoe2e.html(accessed20March2009).
Huston,G.(2008b),10YearsLater.
http://www.potaroo.net/ispcol/200806/10years.html(accessed19April2009).
Huston,G.(2009),BGPin2008.
http://www.potaroo.net/ispcol/200903/bgp2008.html(accessed16April
2009).
InternetWorldStats(2009),WorldInternetUsageandPopulationStatistics,Miniwatts
MarketingGroup.
http://internetworldstats.com/stats.htm(accessed9April2009).
Ipoque(2007),InternetStudy2007.
http://www.ipoque.com/resources/internetstudies/internetstudy2007
(accessedDecember8,2008).
Ipoque(2009),InternetStudy2008/2009.
http://www.ipoque.com/resources/internetstudies/internetstudy2008_2009
(accessed3March2009).
IPv4(2009),IPv4AddressReport.
http://www.potaroo.net/tools/ipv4/(accessed16April2009).
Isenberg,D.(1998),Thedawnofthestupidnetwork,netWorker,vol.2,pp.2431.
ISC(2009),TheISCDomainSurvey.
https://www.isc.org/solutions/survey(accessed20April2009).
ISOC(2009),InternetSocietyHistoriesoftheInternet.
http://www.isoc.org/internet/history/(accessed18March2009).
Juvonen,R.(2008),ICTSHOKFutureInternetProgramme,presentationonTivitForesight
SeminaronDecember4th,2008.
81
http://www.tieke.fi/mp/db/file_library/x/IMG/36751/file/TIVITForesightSemin
ar2008Juvonen.pdf(accessed28April2009).
Karlson,B.,Bria,A.,Lnnqvist,P.,Norlin,C.,Lind,J.(2003),WirelessForesight:Scenarios
ofthemobileworldin2015,Chichester,UK:Wiley.
Kesan,J.,Shah,R.(2001),FoolUsOnceShameOnYouFoolUsTwiceShameOnUs:What
WeCanLearnfromthePrivatizationsoftheInternetBackboneNetworkandthe
DomainNameSystem,WashingtonUniversityLawQuarterly,vol.79,pp.89220.
King,N.(2004),Usinginterviewsinqualitativeresearch,inCassel,C.,Symon,G.(eds.),
Essentialguidetoqualitativemethodsinorganizationalresearch,London,UK:
Sage,pp.1122.
Kivi,A.(2008),Mobiledataserviceusagemeasurementsresults20052007.
http://www.netlab.tkk.fi/~jakivi/publications/Kivi_Mobile_Data_Service_Usage_
2005_2007.pdf(accessed19April2009).
Kleinrock,L.(1961),InformationFlowinLargeCommunicationNets,RLEQuarterly
ProgressReport,July1961.
Kvale,S.(1983),TheQualitativeResearchInterview:APhenomenologicalanda
HermeneuticalModeofUnderstanding,JournalofPhenomenologicalPsychology,
vol.14,pp.171196.
Licklider,J.,ClarkW.(1962),OnLineManComputerCommunication,AFIPSJoint
ComputerConferences,ProceedingsoftheMay13,1962,pp.113128.
Leber,M.(2009),GlobalIPv6DeploymentProgressReport.
http://bgp.he.net/ipv6progressreport.cgi(accessed18April2009).
Leiner,B.,Cerf,V.,Clark,D.,Kahn,R.,Kleinrock,L.,Lynch,D.,Postel,J.,Roberts,L.,Wolff,
S.(2003),ABriefHistoryoftheInternet,version3.32.
http://www.isoc.org/internet/history/brief.shtml(accessed19March2009).
Lessig,L.(2000),CodeisLawOnLibertyinCyberspace,HarvardMagazine,vol.103(2).
http://harvardmagazine.com/2000/01/codeislaw.html(accessed16April
2009).
Levien,R.,Austein,R.,Borgman,C.,Casey,T.,Dubberly,H.,Falstrom,P.,Halvorsen,P.,
JenkinsM.,Klensin,J.,Mueller,M.,Nelson,S.,Partridge,C.,Raduchel,W.,Varian,H.
(2005),SignpostsinCyberspace:TheDomainNameSystemandInternet
Navigation,Washington,D.C.:NationalAcademiesPress.
82
Merit(1995),Relationbetweenpercentageofbytes&packetstransferredpermonthby
service.
http://www.cc.gatech.edu/gvu/stats/NSF/BothP.GIF(accessed6April2009).
Meyer,D.,Zhang,L,Fall,K.(2007),RFC4984ReportfromtheIABWorkshoponRouting
andAddressing.
Mockapetris,P.(1987)RFC1034DomainNamesConceptandFacilities.
Moors,T.(2002),Acriticalreviewof"Endtoendargumentsinsystemdesign",IEEE
InternationalConferenceonCommunications,vol.2,pp.12141219.
Moskowitz,R.,Nikander,P.(2006),RFC4423HostIdentityProtocol(HIP)Architecture.
http://tools.ietf.org/html/rfc4423(accessed16April2009).
Mueller,M.(2008),ScarcityinIPaddresses:IPv4AddressTransferMarketsandthe
RegionalInternetAddressRegistries.
http://www.internetgovernance.org/pdf/IPAddress_TransferMarkets.pdf
Mueller,M.(2009),MemotoJohnMarkoff:Thereareno"doovers"inhistory,Blogpostin
InternetGovernanceProjectwebpage.
http://blog.internetgovernance.org/blog/_archives/2009/2/28/4107895.html
Nikander,P.,Mntyl,M.(eds.)(2007),ICTSHOKFutureInternetResearchAgenda.
http://www.futureinternet.fi/publications/ICT_SHOK_FI_SRA_Research_Agenda.p
df(accessed15April2009).
Nishitani,T,Miyakawa,S.,Nakagawa,A.,Ashida,H.(2008),CommonFunctionsofLarge
ScaleNAT(LSN),IETFInternetDraft.
http://tools.ietf.org/html/draftnishitanicgn01(accessed16April2009).
Nordlund,T.,Tallberg,M.,Hmminen,H.(2007),Scenariosformanagementofdigital
homes,Proceedingsof6thConferenceonTelecommunicationTechnoEconomics
(CTTE2007),pp.18.
Norton,W.B.(2001),InternetServiceProvidersandPeering.
http://www.cs.cmu.edu/~srini/15744/readings/Nor01.pdf(accessed25
February2009).
83
Norton,W.B.(2002),ABusinessCaseforISPPeering.
http://pages.cs.wisc.edu/~akella/CS640/F06/readings/norton.pdf(accessed25
February2009).
OberholzerGee,F.,Strumpf,K.(2007),TheEffectofFileSharingonRecordSales:An
EmpiricalAnalysis,JournalofPoliticalEconomy,vol.115,pp.142.
Oram,A.,(2001),PeertoPeer:HarnessingthePowerofDisruptiveTechnologies.
Sebastopol,CA:O'Reilly,pp.818.
Osborn,A.(1963),Appliedimagination:Principlesandproceduresofcreativeproblem
solving,ThirdRevisedEdition,NewYork,NY:CharlesScribnersSons.
Postel,J.,Reynolds,J.(1985),RFC959FileTransferProtocol(FTP).
Reding,V.(2009),Internetofthefuture:Europemustbeakeyplayer,speechtotheLisbon
Councilon2February2009.
http://ec.europa.eu/commission_barroso/reding/docs/speeches/2009/brussels
20090202.pdf(accessed27March2009).
RenesysCorp.(2009),InternetCaptivityandtheDepeeringMenace,presentationat
NANOG45conference.
http://www.renesys.com/tech/presentations/pdf/nanog45Internet
Peering.pdf(accessed25February2009).
Rescorla,E.(2001),SSLandTLS:DesigningandBuildingSecureSystems,UnitedStates:
AddisonWesley.
Reynolds,J.(1989),RFC1135TheHelminthiasisoftheInternet.
http://www.ietf.org/rfc/rfc1135.txt(accessed5May2009).
RFCEditoret.al(1999),RFC255530YearsofRFCs.
Rickards,T.(1999),Brainstorming,inM.Runco,andS.Pritzker(eds.):Encyclopediaof
Creativity,SanDiego,CA:AcademicPress,vol.1,pp.219227.
RietzschelE.,Nijstada,B.Stroebe,W.(2006),Productivityisnotenough:Acomparisonof
interactiveandnominalbrainstorminggroupsonideagenerationandselection,
JournalofExperimentalSocialPsychology,vol.42,pp.244251.
84
Roberts,L.(1967),MultipleComputerNetworksandIntercomputerCommunication,
ProceedingsofthefirstACMsymposiumonOperatingSystemPrinciples,pp.3.1
3.6.
Robson,C.(2002),RealWorldResearch,2ndEd.,Cornwall,UK:Blackwell,pp.269291.
Saltzer,J.,Reed,D.,ClarkD.(1984),Endtoendargumentsinsystemdesign,ACM
TransactionsonComputerSystems(TOCS),vol.2,pp.277288.
Schoemaker,P.(1991),Whenandhowtousescenarioplanning:Aheuristicapproach,
JournalofForecasting,vol.10,pp.549564.
Schoemaker,P.(1993),Multiplescenariodevelopment:Itsconceptualandbehavioral
foundation,StrategicManagementJournal,vol.14,pp.193213.
Schoemaker,P.(1995),Scenarioplanning:Atoolforstrategicthinking,Sloan
ManagementReview,vol.36,pp.2540.
Schoemaker,P.,MavaddatV.(2000),ScenarioPlanningforDisruptiveTechnologies,inG.S.
DayandP.J.H.Schoemaker(eds.):WhartononManagingEmergingTechnologies,
NewYork:JohnWiley&Sons,pp.206241.
Schwartz,P.(1998),TheArtoftheLongViewPlanningforthefutureinanUncertain
World,Chichester,UK:JohnWiley&Sons.
Silva,J.,Campolargo,M.(2009),MessagebyJoaodaSilvaandMarioCampolargointhe
FIAwebpage.
http://www.futureinternet.eu/home/futureinternetassembly/praguemay
2009.html#c135(accessed24April2009).
Smura,T.,Sorri,A.(2009),FutureScenariosforLocalAreaAccess:IndustryStructureand
AccessFragmentation,inProceedingsoftheEighthInternationalConferenceon
MobileBusiness(ICMB2009),Dalian,China,June2728,2009.
StatisticsFinland(2008),PurposesoftheuseoftheInternetinspring2008,percentof
Internetusersbyagegroup.
http://stat.fi/til/sutivi/2008/sutivi_2008_20080825_tau_001_en.html
Srel,M.,Nikander,P.(2008),SocialAspectsofTrustintheInternet:Issuesand
Incentives.
https://www.tml.tkk.fi/~pnr/publications/CTTE_sarela_2008.pdf(accessed16
April2009).
85
TeleGeography(2009),TeleGeographyInternationalTelecomTrendsSeminar,PTC2009.
http://www.ptc.org/ptc09/images/papers/PTC%2709_TeleGeography_Slides.pd
f(accessed25February2009).
Thomas,P.(1974),EnvironmentalAnalysisforCorporatePlanning,BusinessHorizons,
vol.17(5),pp.2738.
Thomas,S.(2000),SSLandTLSessentialssecuringtheWeb,NewYork,U.S.:Wiley.
Zakon,R.(2006),HobbesInternetTimeline.
http://www.zakon.org/robert/internet/timeline/(accessed18March2009).
Zittrain,J.(2008),TheFutureoftheInternetAndHowtoStopIt,NewHaven&London:
YaleUniversityPress.
VanderHeijden,K.(1996),ScenariosTheArtofStrategicConversation,Chichester,UK:
JohnWiley&Sons,pp.305.
Wack,P.(1985),Scenarios:UnchartedWatersAhead,HarvardBusinessReview,vol.63,
pp.7389.
Wu,T.(2009),NetworkNeutralityFAQ.
http://timwu.org/network_neutrality.html(accessed27March2009).
86
Appendices
A)Summaryofbrainstormingresults
The forces identified in the three brainstorming sessions are grouped into four groups
based on the evaluation of importance and uncertainty. The groups are explained
inFigure 22 below. The most important groups key uncertainties (4) and key trends
(3) are also colored. Additionally, the number of Postits grouped together is given.
Someforcesremainedungroupedduetotimeconstraints,andtheyaremarkedwith.
Uncertainty
Key uncertainties
Key trends
Importance
Figure22:Explanationofthegroupsofidentifiedforces
29.8.2008Firstbrainstormingsession
Political/RegulatoryForces
Group PostIts
Wherestandardizationhappensandbywhichauthority?
OperatorControlincreases(netneutrality,regulatorhassomethingtosay).
GovernmentalControl(censorship,regulatory)increases.
Energyconsumptionincreasesandgreenvaluesbecomemoreimportant.
Economic/BusinessForces
Group PostIts
TheInternetwillbesplittomanyInternets(safe&restrictedvs.anarchy;
geographically).
ThebusinessmodeloftheInterneteconomywillchange(adsvs.
micropaymentsvs....).
Industryconsolidationincreases(fewerplayers).
Technologicalleadership:Westernvs.Asian.
87
SocialForces
Group PostIts
Securityandprivacywillbecomemoreimportant.
Needformobilityandallaroundavailabilityincreases.
Contentcreatingwillbeeasierandmoreuserdriven.
Socialnetworkingincreasesandpeoplespendmoretimeonline.
Needforsimplicityincreases.
Servicedevelopmentwillbemoreuserdriven.
Remoteworkwillincrease.
TechnologicalForces
Intelligencemovesfromedgestothenetwork(cloudcomputing,data
storage).
Routingsystemischokingundertheincreasingtraffic.
Group PostIts
4
P2Preplacesclientservermodelasthemainarchitecture.
Unwantedtrafficwillstayasaproblem.
Theamountofusersandhostswillbeincreasing(IoT,3rdworld).
Opensourcegainsmorepopularity.
Complexity(ofprograms,services,architectures)increases.
ShortageoftheIPv4addresseslimitstheamountofhoststhatcanbe
connectedtotheInternet.
MoreandmoremiddleboxesintheInternet.
Trafficoptimizationincreases(P2Poverlayoptimization,YouTubeservers
insidetheISPs).
Deviceswithlimitedprocessingetc.capabilitieswillbecomemorecommonin
theInternet(InternetofThings).
IMSFuture:Yes/No.
Closedapplianceswillbecomemorecommon.
VideoincreasesInternettrafficremarkably.
Mobiledevicebatterylifestaysasabottleneck.
Datastoragecapacityisnotabottleneck.
88
23.9.2009Secondbrainstormingsession
Political/RegulatoryForces
Group PostIts
WorldchangesfromU.S.drivenunipolartomultipolar,samehappenstothe
Internet?(Openness?ChinaoutoftheInternet?)
IPRregulationchanges,buttowhichdirection?Increasinglystrictandbroad?
Alternative(opensource,creativecommons,filesharing)?NoIPRatall?
EUcannotagreeabouttheFutureInternet>failureinitsFutureInternet
objectives(nosupportingindustryforresearch,shouldbecolocated)
Horizontalizationofecosystemsdevelopsgradually(Politiciansdrivetothis
direction,companiesoppose).
Lawmakersetshighpenaltiesformisuse(spam,DoSattacks).
AmountofspectrumgivenfortheInternet?
Spectrumwillbeauctionedinmostcountriesandliberalizationofspectrum
continues.
Privacyandtrustissuesbecomemoreimportantandrequirenewkindof
legislation.
GovernmentroleisincreasingandpoliciesboostInternetuse&traffic>re
regulation.
Legislationwon'tchangedramatically(disruptively)in10years(allthe
changesarealreadyatleastunderdiscussion).
StrictenergylimitsforICT.
Powerandenergyconsumptionspeeduphorizontalization(ownershipof
things).
Globalregulator?
EUdefinesminimumInternetaccess(freeofcharge?).
Environmentalcriseswillcausemorerestrictivelegislation/Environmental
regulationincreasestheuseofICTandenv.informationinservices>driver
fornewservices/environmentalconcernsboostbroadbandbuildup.
Flatrate,flatarchitecturewilldominateintheInternet.
EUwillsteadilyincreaseitsroleandharmonization.
MonopolypositionofhugeInternetplayersrecognized(Googleetc.).
Fixed/mobiledifferentiationwilldisappearandfluxwillchange.
ChinaandIndiawillconfusethedevelopmentoftheFutureInternet.
Competitionwillbeoligopolic.
Globalhubmodel(Google)winsoverlocal.
89
Group PostIts
Industrydivergenceorconvergence;Contentcreationbyindustryorbyend
users;Porterthinkinginindustry,altruismsupportedbythepeople.
Business>global,Individual>local;Content&Apps>local,Network&
Access>global.
Mostcompaniesfocustosinglelayerplayerinecosystem.
Internetmicropaymenthappens?
Anewplayer(likeGoogle)disturbsthecurrentecosystem.
Emergingmarkets(China,India)takethelead,manufacturinggoestoChina&
India>Servicesincreasinglyimportant.
Environmentalcriseschangeeconomicalrulesandbringbusiness
opportunities.
EconomicvalueoftheInternetdrivesthefuture;Technoeconomicalcriteria
willdominatetheinvestmentsforFutureInternet.
Increaseofadbasedbusinessmodels.
Internetmatures(impactontechnologylifecycle).
ICTbecomescheapcomparedtononICT>bottleneckselsewhere.
Biggeroperators>Consolidation.
Mobilenetworkvendors(NSN,Ericsson,etc.)donotknowwhattodo.
Globalizationincreases(informationexchangeincreases)buttransporting
goodsdecreases.
Routercompanies(Cisco,etc.)willloseleadership.
WirelessInternethelpslocaleconomics.
2billionnewuserswillbemostlywireless.
Chinahastwoanglesofincidence,EuropeandUSAjustone.
Servicesarenotequal,someofthemmoreprominent>servicesector
divided.
Thewholeeconomiccyclein10years>firstdepression,thenhighseason.
Horizontalmarketsrule.
Flatraterulesinnetworksandcontent.
Newglobalhubswillemerge.
Somekeyplayerstrytobuildverticalbundles.
90
SocialForces
Group PostIts
Consumerpreference:bundlesvs.buyingseparately?(especiallyrelationship
betweenpossibilities)
Privacyconcernswillslowdowntakeupofnewservices.
Threatsbetweenvirtualworldandrealworld(e.g.Chinesewantoutside).
Mappingbetweenrealandvirtualworldsimproves>acceptabletopeople,
bringsnewservices.
Flashcrowdsrules>grassrootactivism>largeimpacts.
Userdrivenservicedesign&engineeringincreases.
Socialnetworkinggainsimportance.
SegmentationinallInternetserviceswilldevelop.
Next2billionuserswillcomefrom3rdworldandNICs(willusemostly
wirelessaccess).
Agingpopulation>morefreetimeandserviceconsumption.
Languagebarrierslessimportant,butalsolotoflocallanguagecontent(IDN).
Businessusewillcontinuetogrowfast.
Fragmentationofcontentmarketswillincrease.
GenerationY(borninthe80s)becomeseconomicallyimportant.
Communalityovercomesindividuality.
PeopleareincreasinglyableandwillingtouseInternetservices.
Consumerasthekingwilldominatenewservices>personalization,mass
customization.
Outsourcingofethicalandmoralbehavior(Microsoftvs.Linux,Googlevs.??);
Ecologicalbehaviorbecomessociallyimperative.
Consumerswillbereadytopayforevidentaddedvalue.
Greenvalueswilldriveforgreenservices.
Communitiesaretribesforlikeminded.
Valueleadershipimportanttrendsetter.
Neutralgooddonotneedtocaresomuchaboutvalues.Noharm=important.
91
TechnologicalForces
Group PostIts
InternetwithoutIPwillemerge;Ethernetrules;1Tlinks>newprotocol
stack;scaling100x>packettransmission.
Newsolutiontoprivacyproblemswillbeinvented;Severalwaystosupport
secureidentity;Universalcitizenauthenticationhappens.
Opensourcesoftwarebecomestotallydominant;Opensource,open
interfaces:howmuchandwhere?
Silicontechnologydevelopsstill10years(Moore'slaw).
Webservices/XMLwins(possiblestopper:inefficiency>energy
consumption).
Whichaccess?Optical,copper,cableorwireless?
Selforganizingnetworkssimplifycomplicationsofmanagementand
configuration.
Cognitiveradiowillchangespectrumuse.
Internetofthings>Ubicompspreadsquietly;Stronglocalconnectivity;Need
fornewtechnologycompetences.
Powerconsumptionbecomesthemainbottleneck>biggerdriver.
Softwaretechnologiesbecomebottleneck;Applicationdevelopmentstilltoo
complexformostusers.
Fragmentationofradiotechnologieslimitsmobility.
Servervs.client:bothgetfat.
Cellulartechnologydevelopmentceases(FDM,TDM,WDMallused).
Costofintensiveuseofwirelesstechnologygoesdown.
Technologyoptimization=fiberasfaraspossible,onlylast100mwireless
(optimizingpowerconsumption).
IP/Internettakesitall(broadcastetc.).
BroadbandascommonasGSMtoday(bothfixedandwireless).
Highresolutiondisplaysinmobiledevicesenablebetterservices/UIs.
Locationawaretechnologiesspread.
Internetstructurewillremain.Access,core,servicesandcontent
horizontalize.
92
7.10.2009Thirdbrainstormingsession
Politic/RegulatoryForces
Group Postits
Legalinterceptionwillbecomemandatory;Contentfilteringwillincrease.
UN+ITUwilltakestrongerroleinfairsharingofInternetresources(e.g.IP
addresses/names)<>Commercialapproach.
Developmentsin"greenICT"?Willthisaffectthecommoncitizen?
AnonymityintheInternetdisappears(Userregistration>willInternetusers
beforcedtoregister/acquirealicensetousethenet?).
Globalandregionalharmonizationathighlevelselectedtopics(stillplentyof
localdifferences,thedevelopmentstillslow).
Needfornetneutralitydiscussion?Actionsrequired?
Political&regulatoryinteresthasrisen>nomoreafreeindustry.
Moralentrepreneursbecomemoreimportantinshapingtechnology
regardlessofwhat'sactuallyhappening.
Regulationisapoliticalissueacrossinpoliticalspectrum>getsoutofhands
ofexperts.
Group Postits
Communicationwillnotbeafastgrowingindustry>backtonormal
economy(ICTisstillgrowingthough);Increaseincostcontrol>investments
needtohaveclearcost/revenueimpact.
Productivityvs.innovationinoperatorstrategiesiscrucial(howmuchR&D?)
L2+L3split/combinationdisruptstheInternetinfravendormarket(network
devicemanufacturers).
Proprietarysolutionsvs.industrystandards>whatwillprevail?
Movementfrombitpipeproviderstoservices(orjustpacketizingservices).
Theroleoftheservicesincreases.
Servicemarketisglobal,accessmarketandcontentlocal.
Contentwillbeincreasinglylocalizedanddependentonlocationinformation.
Toomuch/heavyemphasisontheInternetonly>whathappenswhenlinks
break,cyberattacksetc.?
Remotemaintenanceofdevicesincreases.
Newbusinessopportunities:identity/trustprovider(banks),remotenetwork
managementprovider.
ISP/Telcoconvergencecontinues.
Providersincreasinglysubcontractservicedevelopment.
93
SocialForces
Group Postits
Lackoftrustincreases,becausetheInternetisincreasinglyhostile.
Digitaldividechangesitsnature.NetintegratestoeverydaylifeinOECDbut
notsomuchinthe3rdworld.Mobilephonesfunctionlikeanequalizingforce.
Netismorenaturalforyoungusers(courage/interesttotrynewthings);Net
ismoreessential/integratedpartofeverydaylife;Ordinarylifeincreasingly
dependentonnotjustcommunicationsbutalsoservices.
PeopletooInternetcentric=informationflows,communicationflows>
peoplestopphysicalactivities.
Opennessvs.privacy:privacyneededinspecificapplicationsonly>high
commercialvalue,opennessenough/neededinmostapplications.
Natureofnetandeaseofusemakepeoplecareless(givetoomuch
informationtotheInternet).
Communications,sharingandusercreatedcontentdrivenewapplications.
MorelateadopterstouseICT>betterusabilityneeded
Consumermarketbecomesincreasinglyavote>recommendations,price
comparisonsmorepopular>affectsonbuyingbehavior.
Consumersacceptcomplex(&judged)informationinvirtualworld.
Generationgap:youthiseagertoadoptnewservices,elderscan'tstaywith.
TechnologicalForces
Group Postits
IncreaseduseofIPv6togetherwithIPv4>canroutershandle?
Numberoftechnologiesin4Gwireless?Singleormanymorethanin3G?
SecureInternetemerges,providingtrust(butalsoanonymity).Mayleadto
isolatedislands.
IETFlosesitsroleasaprimaryInternetstandardizationbody>new
industrydrivenforum.
P2Pmodelwillbedevelopedinlargescale(betterreputation);P2P
technologiesdeployedasmaininfodeliverytechnology.
ShortageofIPv4addresses.
Technologychangestoofast>ordinarypeopleneedtoomuchtechno
awareness.
Theroleofmobilitywillincrease.
Increasedprocessingpowerandbandwidthsleadtohigherpower
consumption>highcosts,globalwarming(canthisbesolvedsomehow?).
Existingequipmentandtechnologieswillpreventanymajorchangesin
mediumterm>slowchanges(legacyburden).
Batterycapacityremainsasbottleneck(Increasedwirelessbandwidth
requireshugebatterycapacities).
94

TD 07 96

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

TD 07 96

Diunggah oleh

Hak Cipta:

Format Tersedia

HELSINKIUNIVERSITYOFTECHNOLOGY

Internet on kasvanut ulos alkuperisist raameistaan, samalla kun sen

Ennen tulevaisuuden tutkimista perehdytn Internetin historiallisiin

Skenaariot esittvt, ett Internetin kohtaamat haasteet voidaan ratkaista monella

Identify key trends and uncertainties of the Future Internet evolution by

Analyze the applicability of the strategic research agenda of Finnish Future

definition for qualitative research interview as aninterview,whosepurposeistogather

Blumenthal, 2001). Thereby anyone can write a communications application, share it

Medium Medium Medium

Networkinfravendors deliver network HW and SW (e.g., routers, fiber, radio

Backbone operators (ISPs) sell global Internet interconnectivity to access

Device & Software vendors manufacture devices like computers, mobile

Endusers, covering both consumers and enterprise customers,access Internet

Research institutes, for instance universities, develop Internet protocols and

` TT5: Remote management of network and home devices increases. Managing

Updatingandinstallingnewservices ispossible onlythroughdevicevendorcontrolled

Content Device Isolated

Anda mungkin juga menyukai