Anda di halaman 1dari 32

q Intelligence

Quarterly
Journal of Advanced Analytics
3

Q2
Transformational IT
Mikael Hagström
Top 5 reasons why CIOs want 5

1O
business analytics
Patrick Van Deven
Keys to the BI kingdom 7
Stan Gibson
Managing the data asset 9
Tony Fisher
The challenge of change 12
Kelly LeVoyer
Toward an integrated approach 15
to economy statistics
Four methods for 17
high-performance computing
Keith Collins
Why business analysts are 21
so important for IT and CIOs
Thomas Wailgum
Ready for takeoff 23
Postmodern storage 26
Allan Russell
Why you need a customer state vector 28
Jim Goodnight
q Intelligence
Quarterly
Journal of Advanced Analytics
Second Quarter 2010

Editorial Director Intelligence Quarterly is published SAS is the leader in business analytics
Mikael Hagström quarterly by SAS Institute Inc. Copyright software and services, and the largest
mikael.hagstroem@sas.com © 2010 SAS Institute Inc., Cary, NC, independent vendor in the business in-
Editor-in-Chief USA. All rights reserved. Limited copies telligence market. Through innovative
Alison Bolen may be made for internal staff use only. solutions delivered within an integrat-
alison.bolen@sas.com Credit must be given to the publisher. ed framework, SAS helps customers
Otherwise, no part of this publication at more than 45,000 sites improve per-
Managing Editor may be reproduced without prior formance and deliver value by making
Anne-Lindsay Beall written permission of the publisher. better decisions faster. Since 1976 SAS
anne-lindsay.beall@sas.com
has been giving customers around the
SAS and all other SAS Institute Inc. world THE POWER TO KNOW. ®

Copy Editors product or service names are reg-


Amy Dyson • Trey Whittenton istered trademarks or trademarks
Editorial Contributors of SAS Institute Inc. in the USA and
Anne-Lindsay Beall other countries. ® indicates USA
registration. Other brand and product
Keith Collins
names are trademarks of their respec-
Wayne Eckerson
tive companies.
Katie Fabiszak
Tony Fisher
Stan Gibson
Kelly LeVoyer
Jennifer McDonald
Marcie Montague
Anita Poellaer
Craig Rubendall
Allan Russell
Peter Schütte
Cathy Traugot
Waynette Tubbs
Patrick Van Deven
Thomas Wailgum
Art Direction
Brian Lloyd
Photography
John Fernez
Steve Muir
SAS® Business Analytics Software
Data Management | Analytics | Reporting | Targeted Business and Industry Solutions

What if you could increase revenue by 66%


using your data to make confident, fact-based decisions?

You can. SAS gives you The Power to Know.®

SAS Business Analytics software helps organizations across every industry


discover innovative ways to increase profits, reduce risk, predict trends,
and turn information assets into true competitive advantage.

www.sas.com/decisions
for a free research paper

SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. 53502US.0510
P3|IQ|1OQ2

Transformational IT
by Mikael Hagström, Executive Vice President of SAS Europe, Middle East, Africa and Asia Pacific

It is now more than 1.5 years since the Consider when the role of the CIO While the CIO 1.0 started to optimize the
Lehman Brothers collapse shook the first gained acceptance. I refer to this business, the CIO 2.0 looked at how to
financial world and marked the begin- period as the era of CIO 1.0. Around realize its full potential. Now, in the new
ning of the worst economic recession the time that ERP systems were put normal, the CIO 3.0 will also help trans-
in recent history. As I mentioned in in place, the CIO 1.0 became relevant form the business to meet the needs of
the last issue of Intelligence Quarterly, due to the size and the importance the future.
we appear to be past the worst and of the ERP investment. Take Nokia,
on our way to a recovery. While eco- for example, who at its peak invested I see three key trends that are further
nomic forecasts still vary, we do know US$100 million annually into the affecting the CIO 3.0:
for certain that the way we operated in deployment of its ERP systems. 1. The focus on the customer is in-
the past will not return. CIO magazine, creasing. More and more knowledge
McKinsey & Company, and others are The next wave was the CIO 2.0, who
increased focus on gaining competi- about the customer is being captured
referring to this new way of operating and used to improve the business.
business as the “new normal.” tive advantage through incremental im-
The concept of customer analytics is
provement of core business processes
evolving into true customer insight into
Equally relevant in this new economy is and moved away from the earlier model
the customers and their needs, with a
the changing role of the CIO and of IT where “one size fits all.” The CIO 2.0
strong focus on customer profitability.
departments in general. As the strategic began to add value to core segments
importance of the data asset and its im- of the business, even taking part in 2. The IT “sandbox” becomes more
pact on the entire organization becomes business planning and shifting IT’s sta- and more powerful. With that, the
known, IT – as the enabler of that asset tus from that of a cost center to a value need to run models on nearly all
– also increases in importance. creator and a core business enabler. operations across the organization
The CIO 2.0 began closing the gap increases. Using information to man-
Within many organizations, business de- between business and IT. age the entire supply chain has almost
mands are increasingly outpacing IT’s
exclusively replaced the earlier focus
ability to deliver. Technology benefits get Today, in the new normal, the CIO 3.0 has
on automation and standardization
recorded in the business units while IT become a business leader with a seat in
that the ERP wave brought forward.
becomes a cost center under increased the boardroom. The CIO 3.0 is playing a
scrutiny. Instead of joining forces, the key role in the planning, development and 3. Real-time information is being used
two areas often find themselves at odds. execution of companywide strategy be- to make operational decisions.
In the new normal, however, the role and cause – within smart organizations – intel- Running complex models to evaluate
impact of IT is changing. ligence fuels every business decision. risk-weighted performance by client
IQ|1OQ2|P4

or pricing optimizations on every prod-


uct, for example, will become more
and more data-intensive and require 4 building blocks of an Integrated Business Analytics Platform
data from more sources across the
organization. Real-time information is 1. Data integration focused on data governance to reduce the movement of data, and data
quality to get it right at the source. New strategies are developed to decide what source
increasingly being used in areas rang-
data to keep, what to aggregate, in what format and how to recreate if necessary.
ing from risk and trading to sales calls
or delivery. This trend is accelerating 2. A flexible framework with a focus on elasticity and best-of-breed databases. Other op-
with the use of social media analysis. timized storage features include in-memory processing as a clear performance advan-
tage and flat tables as a clear cost advantage. To properly manage IT spending, you need
These trends have one thing in common.
to be flexible and work independently along the whole chain, including cloud computing.
To be successful in the new normal, IT
Otherwise, you will risk finding yourself locked in, with relational database management
needs to better capitalize on what could
systems (RDBMS) as a scarce resource.
become your most valuable asset: data.
3. Access to a wide range of analytical techniques and methodologies has never
The continued explosion in data vol-
been more important. Following the automation-based ERP trend, reaching the next level
umes results in higher costs for data
of business process optimization will require a comprehensive range of analytical capabili-
storage and a stronger emphasis on
ties – ranging from basic models to self-learning networks and social intelligence. It’s just as
data movement and quality. From what important to enable users with specialized, industry-specific solutions that solve the problems
I’ve seen and heard as I talk with senior they face every day. Enablers such as cloud computing will further accelerate this trend.
executives around the globe, there is no
indication that this trend will subside. 4. Empowering business users is another clear imperative. Self-service is the
In fact, SAS has long anticipated these mantra of the day, but one size does not fit all. Usability and data visualization capabilities
increases, and has been focusing its re- deploy business analytics across the enterprise and empower decision makers with the
search and development resources to ability to leverage transactional systems and BI in real time. In-memory BI and mobile BI
engineer a single, integrated platform allow for transformational convergence of core business processes when fueled by real-
for business analytics that builds on time BI (refer to the first-quarter issue of Intelligence Quarterly for more on real-time BI).
four building blocks of: data integration;
data storage; advanced analytics;
and customized business intelligence,
In nearly every industry, business ana- We have indeed entered the digital age
including mobile BI and in-memory BI
lytics is quickly assuming a key role as where time to knowledge has become a
for user empowerment.
the primary competitive differentiator clear competitive advantage. q
These four building blocks support for top-line business strategies. You’ll
the area of analytical intelligence see that in this issue of Intelligence
where real-time BI can bring intensely Quarterly, where today’s leading
competitive solutions. For example, organizations – including Briggs &
powerful grid environments can be run Stratton, Fraport AG and the Belgium
at a fraction of the old server costs and Federal Public Service Economy – are
enable banks to run hundreds of thou- leveraging analytics to create value,
sands of price-risk algorithms in real mitigate risk and optimize operations.
time – or even allow a trader to secure
the spread in real time on the trading
floor. Likewise, retailers who can no As the Executive Vice President of SAS Europe, Middle East,
longer afford to waste 15 percent or Africa and Asia Pacific, Mikael Hagström is passionate
about providing a culture where innovation can flourish, re-
more in markdowns alone don’t have
sulting in market leadership both for the organization and
to. In the new normal, new ways to its customers. He is responsible for optimizing business
better manage markdowns or reduce performance, delivering revenue and managing operations
waste levels become critical. Business in more than 50 countries with individual P&L (profit and
analytics makes that possible. loss) centers and nearly 4,000 employees.
P5|IQ|1OQ2

Top 5 reasons why CIOs


want business analytics
SAS executive explains how to bring efficiency to IT
by Patrick Van Deven, SAS Belgium Country Office Manager

We work hard at SAS to understand our professional services division, and – make a wholescale change across his
customers’ business problems, but we most recently – I’ve led the country organization. But who did he choose
also make every effort to understand office for SAS Belgium. Based on my last year when looking for a complete
what IT leaders are confronted with on experiences with CIOs in nearly every platform for a new site in a foreign
a day-to-day basis. Let me be clear: It’s industry, I’ve come to believe there are subsidiary? He chose SAS, partly
not that IT isn’t interested in the business five main reasons CIOs enjoy working because he wanted a test case for the
value, or that IT leaders don’t understand with SAS Business Analytics: benefits of an integrated system.
it. That would be insulting.
System integration Portability
The CIO of a Belgian Federal Ministry To deal with budget cuts and demands The SAS System is portable, and we
explains it like this: “IT invests in new for efficiency, IT leaders are always look- engineer our releases in a way that
technologies that deliver massive ing for ways to reduce disparate skill minimizes migration issues. The CIO of
efficiency gains to the business, but sets so that they don’t have to have a large manufacturer recently told me
then we [IT] are left with the recur- different employees supporting a they migrated their 15-year-old SAS
ring charges from these investments complex, patchwork system. We have programs from a VAX VMS system to
– including maintenance, licenses, customers managing SAS with just one Windows in just a few hours, and they
upgrades – while the business or two full-time employees for a popu- still run fine. In contrast, he has decid-
enjoys the benefits. The following year, lation of hundreds of users for critical ed to drop Business Objects, because
the benefits are forgotten and the IT business applications. One international his team had to rewrite everything to
budget gets cut again.” Over time, this bank I work with has a long history of migrate it to XI, and that was the third
pattern creates such frustration that business analytics developments, and time he confronted that same issue with
IT leaders become leery of vendors they are a best-of-breed shop because them. That rewriting is something he
who pitch business benefits. they’ve implemented a series of dispa- has never experienced with SAS.
rate systems over a long period of time.
In my 11 years at SAS, I’ve worked The CIO knows that situation very well, Diversity
with dozens of CIOs, and I’ve helped and he understands fully the value of SAS Analytics can do almost anything.
with hundreds of SAS implementa- an integrated platform and the cost That means we can replace a lot of other
tions. In that time, I worked two years savings that could come with it. How- systems. If standardization is the agenda,
in sales, five years as Director of our ever, it’s not a strategic time for him to SAS can help reduce vendors in order
IQ|1OQ2|P6

The SAS Benelux office, Robiano Castle, houses 110 employees.

to reduce overall costs. We’ve helped to replace QMF and Focus on the usage fees on the mainframe have larger
customers successfully replace CA mainframe. In both cases, SAS collateral costs than the actual SAS li-
MICS, Focus, IBM QMF and more. One mainframe renewal fees are lower cense fees. For every $1 spent on the SAS
retail bank we work with decided to than the competitors for the same renewal fee, they’re spending $2 for CPU
buy the entire SAS Business Analytics or better functionality. time and $1 for storage. We’re now help-
Framework for data integration, busi- ing that customer move 80 percent of
ness intelligence and analytics. Then  odernization
M their SAS usage to cheaper platforms,
they started looking at SAS IT Resource The CIO of a Belgian payroll man- where CPU cost will be 50 cents per
Management, and realized if they bought agement company said that SAS is dollar of the SAS renewal.
it to replace CA MICS, the savings they probably the only software that will
saw by not paying the MICS license fees survive the phaseout of their mainframe. As you can see from these examples,
paid for the entire SAS investment, includ- He says that negotiating with us is to- SAS makes every effort to support IT
ing the business analytics framework. tally different than with the other legacy and the business. It doesn’t have to
Plus, we used a SAS Macro to complete vendors who know that they have no be either or. When we take the time to
the conversion project in 1/20 the time future anymore and therefore play it understand both, everyone benefits. q
it would have taken for the bank to do extremely hard. At SAS, we help these
it themselves manually. In the end, customers move away from a lock-
the bank invested 400 man days to in situation where they get charged ONLINE
rip-off prices by platform providers. At SAS Belgium and Luxembourg:
complete the conversion, but increased www.sas.com/belux
revenue by US $1 million per year. one company, we learned that SAS

Flexibility Patrick Van Deven is the Country Office Manager


SAS can do what others vendors do, for SAS in Belgium and Luxembourg. Van Deven
but those vendors can’t do everything started at SAS in 1999 as Account Manager and
was subsequently appointed Professional Services
that we do. In other words, customers
Director in September 2001. Prior to joining SAS,
can use their existing SAS products Van Deven was Managing Director of BJL I.T.,
for more than they currently do to- a Brussels-based services company.
day without paying anything more.
We’re working with two different CIOs
P7|IQ|1OQ2

Keys to the BI kingdom


How to unlock business intelligence with data integration
by Stan Gibson, independent technology writer

Without business intelligence, a compa- duplication and inaccuracy and applying environment to Unix-based servers, and
ny’s leaders are likely to find themselves metadata – data about the data – so SAS added the insight and business
stumbling in the dark, deprived of the that it can be accessed by business reporting we needed on top of SAP
information they need to make key intelligence applications. R/3 to make the system complete.” Now,
strategic decisions. Business intelligence Felsing says his team can concentrate
(BI) applications are needed to shed With the advent of operational ERP on providing executives with more
light on critical information and reveal systems, it has become necessary to strategic-level information.
the path to greater competitiveness. pull in data for BI from those systems as
BI applications that have been success- well, in what has become known as “op- Taming a data tsunami
fully deployed across enterprises enable erational BI.” Operational BI can speed Briggs & Stratton is not alone. Accord-
managers to notice patterns, focus on up the business intelligence process, or ing to a recent report by research firm
their meaning and act on their findings the “time to intelligence,” because the IDC, the need for access to different
to improve business performance. data is obtained and made actionable types of data – including data from
in real time (or near real time) from operational sources such as SAP, real-
But before an effective business intel- the systems on which a company’s time data and unstructured data – is a
ligence application can be deployed, operations are running. major factor driving industry growth.
a firm foundation of reliable data must Meanwhile, the market growth asso-
be built. This is the purpose of data Briggs & Stratton is a case in point. ciated with traditional batch-oriented
integration. Data integration is the In 1998, the maker of small gasoline extract-transform-load (ETL) data
bringing together of data from a engines deployed SAP R/3, only to find integration will “slow to a crawl.”
variety of sources in a variety of formats, that the new ERP system did not work
cleansing it of duplication and inaccu- with the reporting system it had in place. The reason for tapping integrated data
racy, and assuring it’s available to BI The company needed to quickly de- from new sources, including data from
tools when it’s needed. With an effective ploy a new BI system drawing on the ERP systems, is that it can expose
data integration infrastructure in place, operational data from the SAP R/3 opportunities for improving productivity,
correct and timely data can be delivered applications. Briggs & Stratton had says Dave Loshin, President of consult-
to BI applications, which in turn can previously deployed a SAS Business ing firm Knowledge Integrity Inc. This
serve up strategically valuable informa- Intelligence solution and once again the is possible because when data from
tion to business managers. company turned to SAS. a variety of sources comes together,
patterns emerge.
Effective data integration tools bring “SAS, as a company, really stepped up
together data not only from structured to the plate for us to resolve the situa- In the case of Briggs & Stratton, the
sources such as database management tion,” says Grant Felsing, Decision company developed a system for track-
systems, but also from unstructured Support Manager at Briggs & Stratton. ing historical information and monitoring
sources such as text files, eliminating “We moved from a legacy mainframe quality trends, facilities operations and
IQ|1OQ2|P8

failure rates for each of its engine series. Warehousing Institute, has researched “SAS provides the entire infrastructure.
This quality improvement application and written about collaborative data We have data integration and data qual-
automatically flags potential areas of integration in a recent report calling ity, analytics and business intelligence
concern and sends e-mails to notify for the formation of teams made up of software,” says Hausman. In the critical
managers and executives when prob- workers from different groups with a area of operational data, SAS products
lems may be emerging. Briggs & stake in the data integration process. are certified by SAP to be compliant
Stratton was able to identify and address When two companies merge, for ex- with that vendor’s products.
a multimillion-dollar quality issue as ample, a data integration team made up
SAS’ data cleansing products are a key
soon as the SAS application was imple- of technical and business people from
ingredient. It’s important to note that
mented. “Identifying that one problem at both companies should be formed
cleansed data can be fed back into
such an early stage in production may to develop and execute a data integra-
the operational system from whence it
have saved us more than $3 million,” tion strategy, so that data assets of the
came. SAS CEO Jim Goodnight puts
says Felsing. combined company can be leveraged
it this way: “It’s a shame to wait until
in a coherent manner.
Companies that don’t take advantage people want to use data in a warehouse
of all the information available to them, Getting the toolset right before cleaning up the data. It should be
regardless of where the data actually At least as important as any organiza- cleaned up in the operational system.”
sits or what application created it, will tional measures for data integration are
Since corporate executives and man-
take longer to make critical business the right tools. SAS Enterprise Data
agers never know for sure exactly
decisions, notes Ken Hausman, Global Integration Server–featuring technol-
what data they might need to access
Product Marketing Manager for Data ogy from SAS subsidiary DataFlux– can
for their business intelligence appli-
integration products at SAS. access virtually all data sources and
cations, it makes sense to integrate
can extract, cleanse, transform, conform,
data in a structured manner across
Hausman adds that having all the aggregate, load and manage data.
the enterprise. Doing so assures the
information you need, when you need It supports real-time data integration
correct data will be available for the
it, enables you to make better, as well services and service-oriented archi-
business intelligence applications that
as faster, decisions. For example, sales tecture. In addition, it supports data
need it, when they need it. That will give
tracking data may be in one sys- warehousing, migration, synchroniza-
business leaders the knowledge they
tem, while inventory data may be in a tion and federation.
need to make the strategic decisions
mainframe, and customer data may their companies require. q
Although it is possible to select point
be purchased from external sources.
products from many different vendors, it
Still other data could be in flat files or
stands to reason that a collection of tools
Excel spreadsheets. Without the abil-
that are engineered to work together is
ity to access and assess all the relevant
likely to result in far fewer headaches.
data, you could miss critical information
that could affect your decision.

Companies that address the need for Stan Gibson is a technology writer in the Boston area and
data integration often face management the former executive editor of eWEEK.
challenges as well. As a result, manag-
ing the integration process is becoming
a best practice in its own right.

Philip Russom, Senior Manager of


Research and Services at The Data
P9|IQ|1OQ2

Managing the data asset


How to treat your data like the high-value business asset that it is
by Tony Fisher, President and CEO of DataFlux

Data is now collected and saved from collection, organization, enhancement,


every conceivable source – Internet monitoring and retirement of your data
applications, front-office and back-office assets throughout the process. While
systems, trading networks, social media you don’t need all the answers at the be-
– and such complexity requires a sophis- ginning, you need a solid plan on how to
ticated, deliberate process for managing proceed – and what the ultimate success
this vital information. After all, data holds indicators will be. Also, these success
the key to sales, marketing, customer indicators have to map to the business
support, production and other initiatives. problem identified earlier; the reason for
Without an accurate view of customers, the project (cut costs, mitigate risks, en-
products, materials, locations and hance revenue, etc.) provides a crucial
assets, how can a company compete in guide for the Define phase.
today’s marketplace?
During the definition phase, an
Organizations must approach data organization should first answer a
management in the same fashion that series of questions about:
they manage any process – with a
well-defined, repeatable methodology. • People. Who’s involved? And for what
To accomplish this, you need a data purpose? This step outlines everyone
management lifecycle methodology involved in the data management
to manage, monitor and maintain process, including executive sup-
data to benefit every phase of the port, manager/director sponsorship
business. DataFlux recommends this and business and IT involvement.
six-phase process: Organizations also set up steer-
ing committees and/or stewardship
1. Define. 4. Execute.
teams to facilitate collaboration on
2. Discover. 5. Evaluate.
cross-functional issues.
3. Design. 6. Control.
• Road map. Where are we now?
Let’s look at each step in detail. Where do we want to go? What ob-
Define stacles are in our way? Often, the first
The Define phase of the data manage- task after selecting the right people is
ment methodology is just as important to determine a path to a successful
as mapping out a journey. The deci- outcome – including the definition of
sions made at this phase will guide the a successful outcome.
IQ|1OQ2|P10

• Source systems. What data will we •D


 ata exploration. This diagnostic an environment that accommodates the
need? Where is that data coming phase is concerned with document- needs of your business.
from? The road map tells the story of ing the data in your organization and
where the project is intended to go. the characteristics of that data. Data The Design phase requires consolidation
This part of the Define phase will in- discovery arms you with information and coordination, all the while concen-
about the accuracy, consistency and trating on three major imperatives:
form the team on the source systems
and which data will play a role in the reliability of your data. 1. Consistency of rules. Ultimately, an
data management project. • Data profiling and auditing. Data pro- organization needs one set of busi-
• Business processes. Which business filing alerts you to data that does not ness rules that can be stored centrally
processes will be affected? How will match the characteristics defined in but deployed across all data sources,
better data enhance the way the or- the metadata compiled during data applications and lines of business.
ganization operates? This part of the exploration. But, more importantly, 2. Consistency of the data model. The
Define process maps the data man- data profiling can also tell you if the data model is the single, definitive
agement strategy to existing business data meets the business rules and source for how your data maps to
processes. Better data can ultimately definitions established in the Define your business. Through the process
streamline business processes, as less phase. In addition, data profiling can of creating a well-structured data
time is spent reconciling confusing help you determine the relationships model, you identify the appropriate
views or managing poor-quality data. across your data sources – where source systems and begin to reconcile
you have similar data, where data is in multiple views, if required.
• Business rules and data definitions.
conflict, where data is duplicated and
How do we define “customer?” How 3. Consistency of business process-
where data may be dormant.
do we want to optimize procurement es. During the Define and Discover
and spending? This phase seems • Data cataloging and business phases, you will identify processes
simple enough, but it can be decep- vocabulary. You need a development that are potentially affected. Now,
tively difficult. Billing might define environment where data sources the task is to provide consistency
“customer” as anyone that receives can be combined and rationalized. across these processes. When cre-
an invoice, while customer support A place where you can group data ating business rules, you have to
may only want to know who the user sources into projects to allow you to know how to reconcile questions like
is. These decisions will form the basis work across your data sources and “Is this a new customer or an exist-
of business rules and data definitions develop a consistent environment for ing customer?” or “Is this a customer
that will guide later phases. managing your data. Data catalog- in good standing?” By understand-
ing lays the groundwork for all data ing the processes that are affected,
Discover management tasks to follow. Data you can design more effective rules
Data can only be useful if you understand catalogs must be augmented with to automate business processes.
where it is, what it means to your organi- business definitions and vocabular-
zation and how it relates to other data in ies, allowing the business user to Execute
your organization. The Discover phase is comfortably navigate the landscape. Now that the business users have es-
designed to do just that. tablished how the data and rules should
Design be defined, it is up to the IT staff to en-
Every new application implementation, After completing the first two steps of sure that databases and applications
data warehouse development, data mi- the data management methodology, you adhere to the definitions. There are
gration or consolidation initiative should will be able map your strategy, identify many types of architectures involved
start with data discovery. Additionally, sources, understand the underlying for- in this phase: enabling ERP and CRM
any time that new data sources enter mats and structures, as well as assess applications via proprietary interfaces,
your organization, start with data discov- the relationships and uses of data. Now enabling data marts and data ware-
ery. Data discovery has several compo- you face another challenge – taking all of houses via extraction, transformation
nents to it, and each prepares you for these different structures, formats, data and loading (ETL) flows, enabling MDM
your data initiatives: sources and data feeds, and creating systems via service-oriented architecture
P11|IQ|1OQ2

(SOA)/ETL or other technologies. The important? After all, you just spent lots of Control
method and management of enabling time, energy and resources to get your One thing is certain in today’s information
the data in any of these environments is systems to a point where the business age: A wide variety of data will continue
a decision that IT has to make in order to users have a consistent and validated to quickly pour into your organization.
ensure the integrity and integration into view of your organization. Isn’t it time to It is easy to see why data is a key asset.
the various systems. just enjoy the success of all this effort? However, it is also important to recog-
nize when data needs to be retired. The
One potential pitfall in the Execute phase Actually, the opposite is true. Very few
Control phase is about reassessing data.
is to duplicate the rules and standards organizations are static – they are forever
If data is no longer useful to your organi-
from the Design phase for each applica- growing and evolving. For example, you
zation, you must be able to retire the data
tion or data source. When duplicating the add new partners that bring new data to
appropriately. This allows you to free
rules and definitions across siloed, un- the table. Your business changes, sales
up resources that are being expended
related systems, multiple, point-to-point regions are created or modified, you take
maintaining the data environment.
interfaces are inadvertently created. on new initiatives and you develop new
These rules definitions must then be products. All of these changes must be For example, let’s look at a common data
updated, separately, each time a rule or reflected in your data, which makes the problem facing financial services firms.
business initiative changes. Evaluate phase so important. When mergers, acquisitions and divesti-
tures occur, you need the ability to purge
Naturally, this approach is highly imprac- Your mantra for success at this point
or re-categorize data. You don’t want to
needs to be: 1. Monitor; 2. Review; and
tical for the IT team to manage. A better spend resources managing the data of a
3. Optimize. Data should be monitored
solution is to build the definitions once company that no longer exists.
and validated as it enters your organi-
and ensure that you have the ability to
zation to verify it is meeting your rules. Lastly, it is important to promote your
collectively apply those definitions across
Those rules need to be constantly moni- successes across your organization.
your organization. As one IT director put
tored to ensure they are still meeting the When you began your life cycle, you
it: “We want to build our standards and
needs of your business. Efforts in dis- were solving a business problem. By the
rules once and then have the ability to
covery, design and execution will allow time you have reached this phase in the
use them repeatedly and propagate to
you to consolidate the rules and require-
the entire organization seamlessly.” life cycle, you should have improved your
ments into a single environment. With
business. Communicate and evangelize
For each data source, each business the ability to centralize the required data
these messages to help everyone from
process and each application that is management rules, the changes can be
senior management on down recognize
modified to the new data definitions, immediately propagated across the or-
that the efforts were successful and the
you need to: ganization, without duplication of effort.
business is improved. This demonstrates
Monitoring is a joint activity between the business benefits of a sound data
• Understand the requirements.
IT and business users. IT monitors and management methodology across the
• Validate that the new integration meets validates that systems are running within organization, and it paves the way for
the requirements. their required service-level needs. Busi- support of future initiatives. q
• Deploy the interface into production. ness users also benefit from the moni-
toring reports – constantly reviewing the
By repeating this process during reports and validating that business
the execution phase, you can create needs are being met while making chang-
the data management rules to guide es when the business needs change.
the collection and organization of data,
test its integrity, and move to the next
phase of the process. Tony Fisher is President and CEO of DataFlux, a
wholly owned subsidiary of SAS, which enables
Evaluate companies to analyze, improve and control their
data through an integrated technology platform. He
A healthy data life cycle requires a robust
has guided DataFlux through tremendous growth as
monitoring and reporting system. The it became a market-leading provider of data quality
data needs to be consistently monitored and data integration solutions.
so it remains fit-for-purpose for your
organization. Why is this so critically
IQ|1OQ2|P12

The challenge of change


IT Director Renee Nocker chronicles Kimberly-Clark’s shift to an analytical company

Renee Nocker, the director who leads Nocker, Director of IT services for Enter-
the business analytics effort at CPG prise Business Intelligence, is passionate
giant Kimberly-Clark Corp., loves about the topic – and the potential – of
a challenge. And it’s a good thing, business analytics, and is working to in-
especially when her CFO asks her still this more data-driven approach into
questions like, “How are we going to Kimberly-Clark’s business operations. “I
change the way people think?” see it as a journey,” she says. “Because
I realize it might take years to get to the
This is the type of question that point where the entire company and the
analytics evangelists like Nocker are entire business process is managed this
faced with as their executives – at least, way. But it’s the right thing to do, and I
the innovative ones – commit to the fully believe in the value it will bring to
vision of becoming an analytical the company. An analytical approach
company. To enable faster, more con- will help us make better decisions us-
fident, data-driven decision making at ing more finely tuned, information-driven
every level of the company is a daunting facts in a more efficient way. It will drive
objective, and one that Kimberly-Clark consistency and speed in how we
has embraced with vigor. manage the levers that impact all of
our business process outcomes.”
Contrary to popular belief, the tech-
nology is only one element of this Nocker is quick to clarify that Kimberly-
transformation. Surrounding the soft- Clark is not entirely new to business
ware and hardware are personal and analytics (the company currently
professional issues such as chang- applies analytic models to its sup-
es in role definitions, skill sets and ply chain and customer relationship
processes. Ironically, there are leaps management efforts, among others)
of faith required, even in the very effort but early on, she became convinced
to infuse more fact-based decision that the company could do it much
making into a company’s operations. more powerfully. It was this conviction
that set her on the course for change.
P13|IQ|1OQ2

Making the business case who live and breathe the processes,
Nocker needed more than the power strategies, language and needs of their
of her conviction to make the business assigned business-unit clients. “They
case for piloting what could become can then work more effectively with my
significant changes to Kimberly-Clark’s development and engineering team to
business processes. She needed quan- determine what solutions the business
tifiable business impact to get the units need,” she says.
attention of her stakeholders. But how
do you anticipate the return on invest- In fact, if the connection with the busi-
ment of process improvement? “We ness units is so critical, Nocker wrestles
asked the business leaders to quantify whether the IT support for analytics,
what the new capabilities would mean query and reporting applications should
for them,” says Nocker. “For example, just live there. “Many companies have a
it will generate x million in new sales, competency center sitting in the busi-
reduce inventory by x amount. Based ness, where they don’t do as much of
on that, I put a stake in the ground and the back-end development but have a
said I would deliver $5 million in busi- seat at the table to plan strategy,” says
ness benefit in this endeavor. At the end Nocker. “Unfortunately, I don’t see that
of year one, we actually delivered about being widely adopted – but I do believe
$23 million worth of business benefit. that will ultimately change, and business
analytics will be seen as more of a busi-
“Of course, I fully realize that the question ness function than an IT function.”
is whether these are hard benefits. Quan-
In the meantime, Nocker says, “We’re
“I put a stake in the ground tifying the expected benefits of business
definitely changing the paradigm here
analytics is always very difficult. The con-
and said I would deliver clusion I have come to over the years is at Kimberly-Clark in positioning some
$5 million in business that the business case for analytics can of our IT staff as internal business
consultants.”
benefit in this endeavor. At be intuitive in nature; we know we need
to do this and we know we will operate
the end of year one, better if we have this information and use
While implementing this liaison-type
role within her own department,
we actually delivered about it wisely. It’s hard to put a price tag on the
Nocker herself has been on a mission
$23 million worth abilities to make faster, better decisions,
to share the analytics vision with peers
predict where our business is going and
of business benefit.” optimize outcomes,” she concludes.
outside of her department. “I person-
ally have spent a lot of time talking
Renee Nocker to and educating business leaders in
A change in mindset
Director of IT services for areas of the company that I believe
Once the business case was adopted,
Enterprise Business Intelligence, could leverage more analytic capa-
Nocker moved ahead to the education
Kimberly-Clark Corp. bilities – global functions like financial
stage of her journey, during which she
encountered two distinct but related analysis and pricing – and saying, ‘Here
challenges. “For one thing, business are the types of capabilities we can
analytics isn’t widely understood on bring, how can we work together?’”
the business side and business ana-
Heavy workloads can be another very
lytics is much more powerful when the
practical obstacle to overcome. Nock-
technology is fully engaged with the
er says, “Many of the business leaders
business needs.” However, she says,
that my consultants work with are fac-
leading to the second challenge, “that
ing huge operational challenges and
isn’t necessarily how people expect to
objectives that they consider higher
work with an IT organization.”
priorities — and that they are relying
To bridge what has traditionally been a on my consultants to help address.
gap between business and IT, Nocker It’s difficult to go to a business leader
created a team of internal consultants who’s under pressure and say, ‘You
IQ|1OQ2|P14

really need to think about this new small number of new employees. (Nocker in tandem with their business partners.
capability.’ So, the consultants have to is a strong believer in skills transference; “This helps ensure we all have a
reassure their clients that they’re tak- her undergraduate degree is in nutrition. common understanding of what we
ing care of current priorities, but also The connection to her current career? “I mean by analytics – and what ‘analytics
coach them on how they can move the am fundamentally driven to understand nirvana’ would look like to all of us.”
needle. It becomes my responsibility how things work and how behaviors
to ensure that my team can articulate a drive results,” she says.) So just what is analytics nirvana to
compelling story about the benefits of Nocker? “For me, part of that vision
business analytics.” Exactly what are the skills that Nocker includes an end-to-end analytical busi-
sees shaping this new breed of busi- ness process,” Nocker says. “So from
Like any good change leader, Nocker ness analytics experts? “A passion for the manufacturing floor to receiving the
recognizes the extreme value to be understanding cause and effect in the order to shipping the product and get-
found in small wins. “We wanted to business and a mindset of continuous ting it into the customer’s hands, even
spark the interest by generating value improvement,” Nocker answers. “Also registering the revenue for the product,
immediately, so we created a Rapid critical is the ability to function within what are the analytics throughout that
Value Realization team to pilot projects both the business and IT sides of the business process that we could pro-
in selected business areas and help business. They need to understand vide at the right time to the right person
people understand the capabilities in that it’s not just about the technology to help them make the right decisions?
their own context. They ran some sup- – technology is just an enabler. They Having those analytics embedded in
ply chain analysis for one factory on need to be able to hear a business a business process for the decision
data around one product and they requirement like ‘I need to manage makers is key to truly making us an
performed workforce utilization analysis eroding margins’ or ‘Help me im- analytic company.”
for our human resources department. prove our customer satisfaction’ and
understand the data and technology So when her CFO asks, “How are we
“This is where we are now,” Nocker that can make that happen.” going to change the way people think?,”
adds. “So while we continue to Nocker is ready, albeit with an ambi-
perform additional pilots to get and In retraining her staff toward developing tious answer: “We need to get people
share those quick successes, our next this new role of internal consultant, Nock- focused on seeking the most optimal
step is to consider how we can scale er had staff members shadow her as outcome for every decision – and that
the pilot into a global process. We she modeled her consultative approach can only be done with the right software
are continuing on this path of Rapid with internal business clients, seeking to capabilities and the right skills. Educate
Value Realization while we build out understand their objectives and educat- them on what is possible, demonstrate
the long-term enterprise platforms ing them about how her team’s analy- results and celebrate successes. That’s
and competencies to support the sis could help. Then, she says, she just how you make it contagious.” q
analytics framework across all busi- sat back and watched. “Over time, you
ness processes and functions. Each see the people who naturally gravitate
Rapid Value Realization takes us toward the client-facing consulting role
one step closer by solidifying the and the ones who prefer to focus on the
stakeholder buy-in and validating back-end development.”
the course we are on to establish a
global, analytics-driven environment.” For all of her staff, Nocker is an avid
believer in continuous training and
New skill sets required often has her team attend conferences
As she reshaped the definition of her
department’s role from IT order-taker to
business consultant, Nocker became Kelly LeVoyer is Senior Manager of the SAS Marketing Edi-
one of a growing contingent of IT ex- torial department and editorial director of sascom maga-
ecutives faced with the need for a new zine at SAS’ corporate headquarters. She and her team of
writers cover current business and technology trends in a
skill set – one that combines the quan-
variety of print, online and social media channels.
titative talent of a statistical analyst with
the business mindset of an MBA. She
focused the majority of her effort on
retraining current staff and hired only a
P15|IQ|1OQ2

Toward an integrated approach


to economy statistics
Modernized government statistics office delivers
macroeconomic data to businesses and citizens

Since 1962, the National Institute of accelerates data collection significantly,


Statistics (NIS) of the Federal Public enables in-depth analysis of data and
Service (FPS) Economy has provided produces a much wider array of statistics
statistics regarding the Belgian econ- with far less manual effort.
omy. NIS collects, processes and
delivers comprehensive information Comprehensive data collection has
on a wide range of topics such as always been the core business of the
population sizes, business demograph- NIS. But it has often been rather labor
ics, profitability indicators, employment intensive, since much of the data had
rates, energy consumption patterns to be gathered using single-purpose
and monetary exchange figures. questionnaires. This time-consuming
method has now been almost entirely
A business intelligence platform to abandoned in favor of automated data
“SAS clearly is a prepare for the future collection from secondary data sourc-
Today, demand for this information is es such as income tax and VAT admin-
motivator for our internal growing rapidly and expectations run istration records, the Crossroads Bank
collaborators, which is high. Information and Communication for enterprises, energy consumption
important to keep them Technology (ICT) Director Frank De Saer databases, and various other relational
engaged for our business.” of the FPS Economy observes that “to- databases in use by federal and re-
day’s businesses need accurate informa- gional authorities. “These databases
Frank De Saer tion to define and optimize their strategy, are available to us; we only have to en-
Information and Communication and they want it without delay. The same sure that we use them efficiently,” says
Technology Director, Belgium’s goes for public policy makers, research De Saer. “We chose SAS because of
Federal Public Service Economy centers, students and even families. We the powerful integration features. It of-
realized several years ago that it would fers the perfect platform to bring all of
be difficult to maintain our service quality these disparate sources together in a
if we didn’t radically modernize our plat- single data warehouse for further anal-
form and processes of collecting, pro- ysis and processing.”
cessing and making data available.”
Better time-to-market by building on
Early in 2007, NIS made the decision a data warehouse
to implement an integrated SAS Busi- Thanks to this data warehouse, new
ness Intelligence platform. The system statistics can be created and published to
IQ|1OQ2|P16

the Web much faster. De Saer observes The integrated SAS reporting and query
that SAS has enabled the NIS to com- tools are already empowering the inter-
pletely revise and improve its production nal NIS staff. The NIS statisticians have
of analytical information. “We advanced easy access to the data, which can be
from an era where every statistic was easily understood without the need for
built up from scratch, going from ICT specialists to build ad hoc queries,
the preparation for data collection, as it is documented in business metada-
to processing, analysis, reporting and ta. Power users also have access to sta-
publishing. We often needed a time tistical analysis toolsets to prepare new
frame of approximately a full year to statistics with available or enriched data.
produce a new indicator.”
De Saer also points out that the
Now with SAS, the time-to-market for FPS Economy attached great impor-
new statistics is much quicker – and tance to controlling and minimizing
more accurate and comprehensive. They operational and technical risks when
often can be delivered within as little as choosing and deploying the platform.
a month, or in some cases even a week, “SAS was the only supplier to offer a
by taking advantage of the integrated completely end-to-end integrated sys- “SAS was the only supplier
tem,” he emphasizes, “including OLAP
SAS architecture. The SAS solution has
functionality, comprehensive reporting
to offer a completely end-
a complete set of inflow tools to collect
data and store it in the data warehouse. capabilities, ETL, data integration and to-end integrated system.”
Furthermore, integrated analysis and re- Web solutions. Other suppliers offered
Frank De Saer
porting tools enable the NIS to effectively a collection of tools from different ven-
Information and Communication
query this data warehouse. “We no longer dors, requiring additional interfacing
Technology Director,
need to calculate every statistic manually. efforts. The integrated SAS solution
Belgium’s Federal Public
Thanks to SAS, this is now accomplished was technically superior and prom-
Service Economy
automatically. In addition, all information ised to be more stable, guaranteeing
integrated security and metadata in
is stored in our data warehouse so that
both the short and the long run.”
it is instantly available anytime it is need-
ed,” states De Saer. The transition started with a com-
prehensive proof of concept (POC),
Easy to adapt and expand
enabling SAS to demonstrate full
The integrated platform is a win-win for
functionality with real NIS data very
customers and NIS collaborators alike.
early into the project. “It was a head
According to De Saer, “SAS is easy to
start for the project, which is now
adapt and expand. This allows us to
already on full cruising speed, prepar-
rapidly adapt to new requirements that
ing the NIS service for the future.” q
always seem to pop up.” For instance,
new statistics can easily be added. And
regarding accessibility, “SAS enables us
to build a secure Web portal to make our ONLINE
statistical information readily available to White paper: Business Analytics for the CIO
customers and NIS collaborators.” www.sas.com/resources/whitepaper/wp_3521.pdf
P17|IQ|1OQ2

Four methods for


high-performance computing
How to choose the right high-performance computing
method for your business analytics scenario
By Keith Collins, Senior VP and Chief Technology Officer, SAS

Anyone who advanced past basic math- This configuration is used primarily as a
ematics in school has learned this simple way to solve batch window time prob-
concept: Large problems can often be lems. If your current process doesn’t
divided into groups of smaller problems. run as quickly as you need it to, you
split it up and run each piece in parallel.
In computing, this concept is gaining
traction on a very large scale. Parallel When you use it: Typically, the cal-
processing, grid computing and virtual- culations in these scenarios can be
ization are all different methods used to partitioned pretty naturally. You can
spread large computational problems partition the problem by something
among multiple resources. But let’s move inherent in the data, such as time,
past the hype and talk about how you geographic region or products. There
can best use some of these architectures is an obvious way to break the prob-
in your IT environment. lem into smaller pieces or a natural
way of splitting up the algorithm.
At SAS, we’ve identified and developed A second type of partitioning scheme
four unique ways to distribute high- takes advantage of discrete com-
performance computing resources to putational steps or subparts in the
give you the right amount of computing algorithm (sometimes called threads).
power, right where you need it for your Subparts can be calculated in parallel
unique analytical problems. In this col- and then brought back together.
umn, I’ll explain each of those methods
and describe specific scenarios where The SAS difference: We do this today
you might use each one. with SAS Grid Manager, and we can
automate analytic processes to help you
Method 1: Shared storage do this in SAS Data Integration Studio
What it is: The first configuration is and SAS Enterprise Miner .
® ™

basic grid computing. In this scenario,


multiple machines are pulling data from Business benefit: As opposed to
a single data source, and each machine running all of your calculations as a
is running different pieces of the bigger sequential process, you’re computing
algorithm or mathematical equation. things in parallel. Whether you use data
Essentially, you’re breaking one big partitioning or algorithm partitioning, the
problem into multiple pieces and main point is that independent com-
running each of those pieces against putational processes can take place at Figure 1:
the same data source at the same time. the same time from a single data source. Traditional storage vs. shared storage
IQ|1OQ2|P18

Now that you’re processing data faster than the window


of time needed, you can start asking what-if questions of
real-time streaming data.

Method 2: Moving compute


to partitioned data
What it is: Partitioned data architectures type of architecture. For example, one
use a single process for distributing the SAS retail customer is using real-time
data. Each processor or “worker node” OLAP for merchandise planning, which
then accesses its portioned data and takes all that data, puts it in memory
performs its computing. In many cases, and then does what-if computations in
the data is already partitioned and exists a large grid architecture.
in separate data stores before you apply
analytics to it. Business benefit: When there is a busi-
ness need for data to be partitioned in
When you use it: These architectures a certain way, allowing analytics to be
are useful for situations where the data used within that existing partitioning
is fairly static or can be easily bulk par- scheme has many natural advantages,
titioned. We currently have a lot of retail including speed and accuracy. In other
clients using this type of architecture for cases, analytics need to be moved to
markdown optimization and merchan- where the data is because governance
dise planning. They have a lot of data around data and the cost to move data
that needs to be stored and associ- is too high.
ated with appropriate computations,
and there’s a high correlation between
the analytics needed and the data that
persists on that node. Using this setup,
retailers have been able to reduce batch
processing times for promotion optimi-
zation and price markdown problems. In
many cases, they have already split the
data by division, which is the right gran-
ularity for the optimization problem, so
they can optimize prices by division.

The SAS difference: Advancements


with SAS In-Database Analytics make it
possible to move the relevant analytic, Figure 2:
data integration and business intel- Traditional storage vs. partitioned,
ligence tasks closer to the data in this distributed data architecture
P19|IQ|1OQ2

We’ve identified and developed four unique ways to


distribute high-performance computing resources to give
you the right amount of computing power, right where
you need it for your unique analytical problems.

Method 3: Moving data to compute The SAS difference: This architecture


What it is: This type of distributed distributes work across many ma-
processing spreads a combination of chines and uses threads to leverage
threads and processes across multiple the hardware on each machine. SAS
machines. Essentially, you break the has developed patent-pending algo-
problem up into a bunch of small pieces rithms that maximize the use of this
along with the data that is needed for architecture. We also use a variant of
each subproblem, and send both the the classic message passing interface
data and the algorithm to different nodes (MPI) by breaking problems down
to be processed. In the most complex into smaller subproblems and using
cases, this involves a “tree” of workers threads to further break down and
(or nodes), with some workers automati- solve the subproblems.
cally delegating tasks to subworkers.
Business benefits: Now that you’re
When you use it: In general, this con- processing data faster than the win-
figuration is good for large computations dow of time needed, you can start
of smaller amounts of data where you asking what-if questions of real-time
distribute all or most of the data to all the streaming data.
different worker nodes. In other words,
the amount of data being analyzed is
relatively small but the computational
tasks involve many parts and subparts.
Customers are using this today for com-
plex risk calculations. For example, large
international banks can recalculate their
entire risk portfolios at very high speeds
with this grid configuration handling hun-
dreds of predictive computations for a
pricing portfolio in a very short amount of
time. One key factor in determining the
value of this setup is to consider whether
the amount of data being moved around Figure 3:
is a manageable size for the available traditional storage vs. message
network bandwidth. passing architecture
IQ|1OQ2|P20

Method 4: Adding the present to past


and future calculations
What it is: Methods 1 through 3 look at credit in different way, for example, you
historical data and traditional architec- may want to respond. If their phone has Systems Devices
tures with information stored in the ware- been working fine, but you can see that
house. In this environment, it often takes its performance is starting to deteriorate,
months of data cleansing and prepara- what can you do to improve performance
tion to get the data ready to analyze. on the fly? You can only take immediate
Now, what if you want to make a decision action if you have a system for analyzing
or determine the effect of an action in real data in real time.
time – as a sale is made, for instance, or
at a specific step in the manufacturing The SAS difference: There are a lot of
process. With streaming data architec- classic statistical process control ideas
tures, you can look at data in the pres- that originated in manufacturing in the
ent and make immediate decisions. The 1980s that can be applied to the way
larger flood of data coming from smart information enters systems and the way
phones, online transactions and smart- we communicate with people who re-
grid houses will continue to increase the ceive the information. We have a lot of
Systems
amounts of data that you might want to experience with CEP and rules engines
analyze but not keep. Real-time stream- that influences our work in this area.
ing, complex event processing (CEP) and
Business benefits: The attention toward
analytics will all come together here to let
in-database analytics fits within this area
you decide on the fly which data is worth
of high-performance computing as well.
keeping and which data to analyze in real
Our strategy for the future will be to put
time and then discard. Systems Devices
the computing power as close to the data
When you use it: Radio-frequency iden- as possible, recognizing that as volumes
tification (RFID) offers a good user case of data increase, you need to move data
for this type of architecture. RFID tags management processes and analytic
processes to the right place. Sometimes

Data
provide a lot of information, but unless
the state of the item changes, you don’t those processes are directly connected
need to keep warehousing the data about to the devices coming in, where analyt-
that object every day. You only keep data ics needs to be applied before you ever
when it moves through the door and out store the data. q
of the warehouse. The same concept Systems
applies to a customer who does the ONLINE
same thing over and over. You don’t IDC White Paper: Raising the Bar on Business Analytics:
need to keep storing data for analysis on Innovation Powered by Grid
a regular pattern, but if they change that www.sas.com/reg/wp/corp/10210
pattern, you might want to pay atten- Data Warehouse/Database
tion. If you can detect that they’re using

Figure 4:
Keith Collins, Senior Vice President and Chief Technology Traditional architecture
Officer (CTO) at SAS, is responsible for driving corporate vs. streaming architecture
technology strategy through a focus on customer- and
partner-facing activities. Collins fosters close working
relationships with marketing and sales to ensure that
SAS technologies are aligned with customer needs and
market demand. He has been instrumental in leading SAS’
evolution as a provider of industry-specific solutions.
P21|IQ|1OQ2

Why business analysts are


so important for IT and CIOs
Molding today’s business analysts into tomorrow’s business technology analysts
by Thomas Wailgum, Senior Editor, CIO Magazine

For two decades, the CIO has been on the importance of the business current and former business analysts
viewed as the ultimate broker between analyst role,” the analysts write, “but few and reviewed more than 29,000 business
the business and technology functions. know exactly what it is that business analyst job postings. What Schwaber and
But while that may be an accurate per- analysts do.” Karel found out is that, at present, there
ception in the executive boardroom, are “many different breeds of business
down in the trenches, business analysts The 21st century business analyst is a
analysts, each native to a particular silo
have been the ones tasked with devel- liaison, bridge and diplomat who
within the enterprise, and each focused
oping business cases for IT application balances the oftentimes incongruous
on addressing the most critical concerns
development, in the process smoothing supply of IT resources and demands
within that silo,” they write in the report,
relations among competing parties and of the business. Forrester’s research
The New Business Analyst.
moving projects along. found that those business analysts
who were most successful were The Forrester analysts also discovered
According to a new Forrester report, the ones who could “communicate, that like many technology-intensive roles
however, the reality is less precise facilitate and analyze.” inside companies today, the line between
than this description. The business a pure business analyst and a pure
analyst position varies depending Some business analyst positions tilt
IT business analyst has blurred. The
on the organization, and the line be- more toward business functions such as
waters are muddied even more because
tween pure business functions and IT operations, marketing, finance or engi-
business’ IT needs (such as ERP sys-
functions has eroded. neering; other analysts seem to fit better
tems consolidation or enterprisewide
in more IT-oriented positions such as in
What is clear: The most successful data warehouse rollouts) span not
applications and architecture groups, or
business analysts are the ones who only different departments but across
in project management offices.
blend the temperament and commu- entire companies. In addition, newer
nications savvy of a diplomat with According to the Forrester analysts, technology methodologies, such as ser-
the analytical skills of an intelligence however, not many people, including vice-oriented architecture (SOA), require
officer. And business analysts are a business analysts themselves, are a deep understanding of both business
hot commodity. able to figure out a standard definition and IT as well as close attention to chang-
(complete with typical skill sets, proper ing business conditions, write Schwaber
The in-depth April 2008 Forrester training methods and set career paths) and Karel.
Research report by analysts Carey for the business analyst position.
Schwaber and Rob Karel provides a So just where should a business ana-
better understanding of this crucial yet To better understand the business lyst reside on an org chart – business
largely undefined role. “Everyone agrees analyst function, Forrester surveyed 338 or IT? “Although distinguishing among
IQ|1OQ2|P22

breeds of business analyst makes sense analysts into tomorrow’s highly evolved role might require some education and
in theory,” note Schwaber and Karel, “in BT analysts. Historically, some CIOs even evangelism.”
practice, trends in both business and IT have struggled with how to best use
are forcing business analysts to assume business analysts. Establish specialized centers
responsibilities outside of their siloed of excellence for business
But CIOs have to do something right technology analysts
comfort areas.”
now to influence the crop of future Given the nature and demands of the
The ultimate blurring of the business- business analysts because the stakes role, business technology analysts will
oriented business analyst and the are too high. “Your future business have to collaborate with a wide range
IT-oriented business analyst, contend technology analysts will be your most of critical business and IT stakeholders.
the Forrester analysts, is what they term valuable business analysts because “As a result, they will likely spend their
the business technology analyst. And the they can single-handedly turn business- days scattered across the enterprise,”
person in this role can be a CIO’s and requested, IT-delivered applications into note Schwaber and Karel. “To ensure
the IT department’s ace in the hole, as well tomorrow’s dynamic business applica- that the business technology analyst role
as a better-equipped business liaison. tions,” write Schwaber and Karel. is coherent, supported and ultimately
attractive, CIOs should establish a fo-
These new and converged business To do this, CIOs and IT managers can rum in which these folks can share best
technology (BT) analysts, write Schwab- do several things right now, the analysts practices, such as a business technology
er and Karel, are the “key to making say. Here are a few of the Forrester analysis center of excellence.”
dynamic business applications a reality analysts’ ideas:
by both accelerating the speed at which In the end, the more business tech-
Look in their own backyards nology analysts that are working in the
business applications can be changed Those employees already working as
and assuring the engagement of the business, the better off the CIO and
business analysts are well-suited to IT function will be– no matter if the BT
business customer in these changes.” the role of business technology analyst analysts are reporting into IT or the
BT analysts possess a blend of business “because they’re already familiar with business side. That’s because those
and operational know-how and a high both the business functions in question IT-savvy analysts, who will have a
degree of tech know-how. In addition, and with business analysis disciplines more in-depth understanding of and
BT analysts have more “cross-functional like process modeling,” write Schwaber more expertise in technologies, will
and cross-domain” business experience, and Karel. “The best candidates are “ultimately help the business make
rather than just focusing on one area or busi-ness-oriented business analysts better decisions when it comes to
function within the business. Schwaber who want more direct control over how its interactions with IT,” contend the
and Karel predict that the different breeds business processes are automated, and Forrester analysts. And, “CIOs have
of business analysts (such as solely busi- IT-oriented business analysts who want new allies in the business.” q
ness-focused or solely IT-focused) “will to move from IT into the business.”
slowly dissolve as projects increasingly This article originally appeared on CIO.com.

demand knowledge that spans business Look for potential analysts in typically
functions like marketing and sales and IT untapped areas of the business
domains like process, information and IT executives also should try fishing in
the pools of business subject-matter ONLINE
experience,” they write. “As time passes, Special 11-page report on staffing for intelligence:
fewer and fewer business analysts will experts for business technology analyst www.sas.com/resources/asset/104328_1209.pdf
have the luxury of working only in a single talent. “Of course, business-oriented
business function or IT domain.” business analysts and business subject
matter experts don’t report to CIOs,” the
The challenge for CIOs, the analysts analysts write. “Convincing them of the
point out, is molding today’s business value of the business technology analyst
P23|IQ|1OQ2

Ready for takeoff


Fraport on the up with SAS Business Analytics
®

As owner and operator of Frankfurt Air- These – and other questions – could only
port, Fraport AG is one of the world’s be partially answered with Fraport’s pre-
leading airport companies. Frankfurt vious heterogeneous IT infrastructure,
Airport had 53.5 million passengers in which had evolved over many years.
2008, making it the third largest airport With a variety of applications developed
in Europe. It has become Germany’s for specific purposes, it was difficult
largest employment complex at a single for the company to see the big picture
location, with more than 500 compa- of flight operations in Frankfurt. And
nies and organizations providing jobs for the uncontrolled growth of applications
approximately 70,000 people. With its resulted in high operating costs.
associated companies and subsidiaries,
Fraport AG is active at more than 60 Roland Krieg, PhD, Senior Vice President
locations around the world. of Information and Telecommunication
and CIO at Fraport AG, states the chal-
Framework supports decision making lenge facing the company IT department
Frankfurt Airport is definitely on the up: very clearly: “For me, it was important to
• Fraport consistently achieves record obtain an answer to the question, ‘What
numbers for passengers and freight. can IT do to provide the information
• A fourth runway is planned, as is a third that will enable management to make
terminal. decisions faster?’” The answer was
• The airport’s operational managers equally unambiguous: Fraport needed
depend on correct, consistent and reli- a business intelligence architecture – a
able information to guide this dynamic framework for applications that would
growth to further success. support decision making by senior man-
agers and specialized departmental
Questions managers ask often include: employees alike – now and in the future.
How fast will passenger numbers grow?
Where is the hidden potential for optimiz- One big picture from many sources
ing current business? What interactions Fraport’s Business Intelligence
are there between individual flight Architecture Framework (BIAF) is the
management processes? solution that now serves as Fraport’s
IQ|1OQ2|P24

Terminal 2 at Frankfurt Airport. Image provided by Fraport


press center photo service.

information backbone. Based on SAS, a Web-based reporting environment, For instance, Fraport relies on SAS
BIAF integrates information on past individually tailored to the employees’ Data Integration Server to deliver fully
and future flight movements, passen- tasks. Currently BIAF has 660 total us- consistent and accurate data. This data
ger and employee data, flight delays, ers, a 95 percent increase since 2007. integration platform supports a central,
weather reports and more – all infor- continuous metadata structure, display-
mation extracted from operational These information consumers can ing not only technical but also business
applications – giving a common overview download predefined reports, contain- management and administrative infor-
of flight operations. ing summaries of the relevant data. mation – such as archive data or access
Key users – employees in specialist rights. Likewise, the SAS Enterprise BI
For example, BIAF tells air traffic con- departments who wish to produce Server creates business intelligence for
trollers whether storms are expected, their own, more in-depth reports, everybody. Multiple buttons meet the
and gives managers an overview or ask ad hoc questions – can make various user requirements and provide
of all key performance indicators in multidimensional analyses at the click reporting and analysis functions in every
near-real time, including average times of a button. degree of detail.
for luggage delivery, delay causes for
all flights and airport security levels. How did the Fraport IT experts meet Dieter Steinmann, Senior Manager of
The business data is refreshed every these varied user requirements and Business Systems at Fraport, sum-
five minutes. Both managers and ensure the flexibility and scalability of marizes the decision to use SAS: “We
operational experts monitor reports via the architecture while guaranteeing the decided to extend BIAF with SAS for
their PCs and mobile devices. consistency of the overall picture? two reasons – from a strategic stand-
point, because the SAS solution
The airline’s overall IT architecture is The answer is to be found in the SAS
offered the necessary guarantee for the
designed for the challenges of the future, Business Analytics Framework, which
future, and from a functional standpoint
so Fraport experts can quickly display provides business intelligence, data
because it is sufficiently powerful to pro-
new processes and integrate further integration and analytic technologies.
duce one big picture out of the myriad
reporting solutions with much less effort. SAS integrates all the necessary
items of information.”
steps for knowledge production –
Carefully defined interfaces and com- from data management, data quality, Why didn’t Fraport develop BIAF with
panywide standardized metadata warehousing, analysis and forecasting, the SAP Business Warehouse, which
guarantee the flexibility and scalability through to reporting and distribution was already installed in the company for
of BIAF. Information is accessed via of the information. operational business tasks? Steinmann
P25|IQ|1OQ2

explains: “We saw advantages of SAS – a reporting application for flight inci-
“We can answer and SAP in different domains, and there- dents, where current and historical data
fore used the systems where they have are summarized, including the destina-
new questions from their respective strengths.” tion, customer structure and delays.
management better
Since bringing the power of SAS to BIAF With BIAF, Fraport’s IT experts have
and quicker, produce more than two years ago, Fraport has created a strong basis for directing
reports and reduce IT noticed the following benefits: flight operations. Fraport has also linked
infrastructure costs, all data from SAP BW with SAS. Now the
• Consistent, reliable overview of flight questions about costs, revenues and
without any major effort.” operations. process profitability can be answered –
Roland Krieg, • Display of complete processes over the information that cannot be derived from
PhD, Senior Vice President of whole company. SAP BW alone.
Information and Telecommunication • Simple and fast integration of new tasks
and requirements. “We can answer new questions from
and CIO, Fraport AG
• Faster, better and simpler reporting on management better and quicker, produce
a common database. reports and reduce IT infrastructure costs,
• Individual items of information in real all without any major effort,” says Krieg.
time available to management and
Krieg advises his peers in other organiza-
specialist departments.
tions to support management by setting
• Business intelligence for everybody;
up similar BI architectures.“I recommend
in-depth analysis for specialist users.
that business intelligence be seen as
• Lower operating costs for the business
a strategic task for the CIO, thereby
intelligence infrastructure.
avoiding the situation where you have a
• Financial control enabled through
mass of individual solutions delivering in-
integration of data from SAP BW.
consistent data, which will cause people
Strategic intelligence at Fraport to make bad decisions.” q
Fraport and SAS specialists deliberately
invested quite a lot of time in the design
of the architecture during the concep- ONLINE
tual phase. Carefully defined standards SAS Business Analytics Framework:
for data control, security questions, www.sas.com/businessanalytics
authorizations and definitions of who
does what ensure that, later on, IT
experts can react very quickly and
flexibly to new requests.

The payback on this investment in-


cludes very short development and
implementation times for subsequent
individual reporting applications. The
first BIAF solution went live in July 2005
IQ|1OQ2|P26

Postmodern storage
Use business analytics across your entire storage landscape
by Allan Russell, SAS Fellow

For the past fifteen years, data ware- To address these issues, organizations
houses have been built and optimized are looking at a more diverse range of
to store and handle data from transac- storage approaches. Instead of forcing
tional systems. For certain kinds of data all data to trickle into data marts from an
and certain kinds of questions, the data enterprise data warehouse (EDW), IT
warehouse makes sense, especially leaders are learning how to design,
when business users are asking: build and manage a range of analytical
databases built specifically to house
• How many dryers were sold last year? data for analytical applications. Some
of the data may indeed come from the
• When did Suzanna Smith become
EDW. However, it’s likely that the struc-
a customer?
tures will be different, incorporating
• How much business did we do with our wider tables with many variables, and
top supplier in 2009? also including data which has not come
through the classic EDW.
As the business becomes more complex,
With more and more data flooding the
however, so do the questions asked by
organization every day, it’s simply not
and about the business, including:
necessary to store it all in the EDW. Rath-
• Why is this happening? er, some of it will be stored in temporary
databases or data marts, some of it will
• What if these trends continue? be stored in larger tables that pull in only
the most recent attributes for decision
• What actions are best to take based
making (see p 28), and some of it will be
on these trends?
stored off site in “the cloud.”
In addition to asking more complex Ideally, your storage decisions will be
questions, businesses are deal- made with analytics as a priority. After
ing with dramatic increases in data all, many IT analyst firms, from Gartner
sources, including data that may to IDC, are saying that analytics are
come from outside the organization or the future. If you intend to be part of
may not originate from classic finan- that future, and you’re planning to run
cial-based sources. your business on analytics, you need
P27|IQ|1OQ2

a business analytics framework that be great for managing data that is used One important question we hear a
can work with every type of storage in for basic reporting, but the future will lot is, “How do you ensure that com-
your IT environment. see more and more storage grids, cloud mon entities are in fact described in a
computing and “shared nothing” con- common way when you may have
Attributes of new cepts for storage. them represented in multiple stores?”
analytic infrastructures Here, we recommend Master Data
When applying analytics, you employ With the growing number of different Management to force a common set of
a series of different analytical methods physical storage architectures, it’s un- definitions and to ensure a level of data
to answer different types of questions: likely that any one option will be used quality (see page 9) across systems.
predictive modeling to identify your consistently throughout an organiza-
most profitable customers, statistical tion. You need to be able to flexibly opt It is also important to set your data qual-
forecasting to estimate the right level of between the different storage architec- ity standards based on how the data will
inventory and optimization to reduce tures depending on the type of analytical be used. Certainly, some data, such as
shipping costs while satisfying demand. method that you’re using – and be flex- financial data and customer data, should
ible enough to choose the right analytics be held to high data quality standards.
Likewise, in the high-performance for any type of storage architecture. However, as the range of data we need
computing article on page 17 of this to manage continues to expand, it may
publication, Keith Collins talks about The good news is that SAS business be that certain data is useful even though
four different ways to break analytic analytics can work with all of these its quality is not (and possibly never can
problems down into smaller parts architectures. If you use one platform for be) anywhere near 100 percent. This
to achieve faster results. As these dif- your enterprise data warehouse, another is particularly true of much of the data
ferent grid-enabled methods for analyt- for the data you’re storing temporarily used in analytical processes. Analyzing
ics increase in popularity, your storage and another for real-time analytics, you massive amounts of data to discover
approaches need to become flexible can use SAS across the full spectrum of broad patterns will not be affected by
enough to handle the many types of storage environments. the occasional error in the detail in a few
analytical problems you’re solving. records out of the, say, 3 billion that you
Business implications
From the origins of database systems, With the emergence of diverse storage are analyzing. This again is an issue of
the idea has been to store data on discs architectures, businesses do need to governance: deciding what levels of data
where it could be used by more than one guard against some of the governance quality are appropriate for the different
application. Then we evolved to the EDW issues that the single EDW approach types of data you will be using, and en-
approach, which allowed us to combine was designed to combat. Modern data suring that quality levels are managed
and standardize data in situations where quality and data governance approaches consistently across the infrastructure.
different definitions and formats had help make the most of diversity while still Flexibility and adaptability are important
emerged. These systems will continue to retaining control. attributes in your storage environment,
especially if you want to reduce
storage costs and deliver only the
data you need for decision making. As
Allan Russell, Senior Vice President of Strategy at SAS, was your storage environment continues to
selected in 2007 by CEO Jim Goodnight to be a SAS Fellow,
evolve, remember that SAS Business
researching and reporting on emerging trends in the soft-
ware industry. A company veteran for more than a quarter Analytics offer the flexibility and the
century, Russell works with customers and technology data management capabilities to turn
innovators in many industries. data into insight, regardless of whether
your data comes from the data ware-
house, a data mart or the cloud. q
IQ|1OQ2|P28

Why you need a customer state vector


Improve modeling with a cross-departmental view of customer data
by Jim Goodnight, CEO, SAS

In most organizations, customer data is is now and projects where it is going. response rates and what types of
stored in multiple disparate systems, re- Vectors are an excellent prediction tool bank services are used together with
sulting in separate product-related data for launch, orbit and landing positions. different banking products.
stores about each individual customer. In
a bank, for example, data about a cus- Likewise, a customer state vector acts as In most organizations today, those
tomer’s credit card history is often stored a central repository for time-based and variables for credit card fraud, banking
separately from checking account data, “customer state” data. This approach to fraud and marketing would all be stored
home loan data and car loan data. customer-centric data storage gathers separately. The customer state vec-
data from various systems and stores tor, on the other hand, works across all
On a practical level, this limits your un- the entirety of a person’s current inter- operational systems and combines the
derstanding of each customer to a set actions with the organization based on data needed for all the models into one
of discrete touch points. On a technical integrated, real-time information across table. Every time a change is made in a
level, it limits the number of variables all departments and divisions. customer’s state, a message is sent to
that analysts can use to build models for the state vector manager detailing what
fraud detection, campaign management In business analytics, state vectors just occurred. The manager uses that
and customer support programs. are used for modeling. In the process information to update the state vector
of building predictive models for fraud for the enterprise.
Often, organizations believe they are detection or marketing, you discover
building an efficient, customer-centric the underlying set of variables that are The idea of the state vector is to collect
environment when they are, in fact, in- important for use in those predictive all the data for all the decision making
creasing the costs and complexities of models. Once you’ve built the models, for an entire enterprise in one place. The
managing and maintaining widespread you get a good picture of the data and CSV is not necessarily storing raw data
data. To reduce those complexities, we variables you need to collect. That set on every check that’s written. Instead, it
propose a real-time storage option called of data is labeled a state vector for the might include a summarization of some
the customer state vector (CSV). models you’re building. statistics that are built up from that ac-
count. Rather than saving every last bit of
The customer state vector is based on With credit card fraud, for instance, you’re
customer data from operational systems,
an engineering concept that is popular going to need quite a few variables from it only saves data used in the models the
in the science community. For example, the credit card side, including purchase organization is running.
NASA uses a state vector to control the history and frequency of use. For fraud
space shuttle during operations. Vari- in the banking area, you look at size Once the vector is updated, a number of
ables in the shuttle state vector show the of deposits, average check size, the different models can be executed and – if
present position, velocity components average number of checks written per necessary – send information back to,
and other factors of the orbital trajectory week and similar variables. In market- say, the charge department on whether
at snapshots in time. It analyzes where ing, you try to build models with all or not to accept that charge. The purpose
an orbiting vehicle has been, where it the relative data available on campaign of the CSV is to supply all the variables
P29|IQ|1OQ2

The idea of the state vector is to collect all the data


for all the decision making for an entire enterprise
in one place.

needed for all the models throughout the of using just card data, for example,
bank, so you can be ready to score any you can use card data and banking
event that occurs using modeling. deposit data.

One of the important things about this ap- When you pull in variables from multiple
proach is that it leaves all the operational areas across the organization, you’ll start
systems untouched. Traditionally, there to see an uplift in results as models are
are so many reports that are dependent updated. I’d predict a 15 percent to 20
upon those silos of data that organiza- percent lift in model performance.
tions never merge everything together in
one giant database because of the huge As expectations for CIOs continue
amounts of surrounding reporting and to become more strategic and more
verification that goes on within the silos. business-focused, a state vector proj-
ect is one way for IT to bring value to
The CSV helps the organization break the business as a whole. IT leaders
out of its silo approach and bring fraud aren’t likely to remove the independent
data together with customer support silos in their organization, but they can
data across the organization. easily gather the data needed from
each of those silos and start maintaining
By contrast, most data warehousing is one CSV to see what every customer
not done in real time. Updates are done looks like as a whole.
on a nightly or weekly extract of opera-
tional data bases. The CSV becomes The results in model uplift will be seen
more of an operational data store to keep across divisions, and operational sys-
track of the state of everything that’s tems will not be affected. If, like most
going on for a different customer. CIOs, you’re being asked to focus
on business gain and long-term
Once you’ve accumulated all the data strategic plans, consider the benefits
used in all the different models, you of a customer state vector. q
could expand the models to make use
of more data from that state vector and
improve your modeling results. Instead

Jim Goodnight has served as SAS CEO since he


co-founded SAS in 1976. A worldwide leader in
enterprise intelligence solutions and services, SAS has
more than 11,000 employees in 53 countries. In 2009,
the company’s 34th consecutive year of growth and
profitability, its revenue was $2.31 billion. Goodnight
has a PhD in statistics and speaks internationally
on leadership, education and innovation.
SAS® for Cyber Security
Real-Time Monitoring of Networks | Historical Trending and Analysis | Aggregation of Massive Data Volumes | Detection of Future Threats

What if you could more accurately detect the 10% of


security threats that require immediate attention?

You can. SAS gives you The Power to Know.®

SAS for cyber security combats threats using cyber analytics for complete and real-time situational
awareness of your entire network through monitoring, integrating, uncovering and predicting.

www.sas.com/cybersecurity
for a free white paper

SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies. © 2010 SAS Institute Inc. All rights reserved. 53502US.0510
SAS Institute Inc. World Headquarters   +1 919 677 8000
To contact your local SAS office, please visit: www.sas.com/offices
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA
and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies.
Copyright © 2010, SAS Institute Inc. All rights reserved. S53502.0510

Anda mungkin juga menyukai