P43 IGRP EIRGP - v3 5 PDF

Institute of Computer Technology - Vienna University of Technology
L43 - IRGP and EIGRP
IGRP Background
IGRP stands for Interior Gateway Routing Protocol

designed
and deployed successfully in 1986 by

Cisco Systems
proprietary protocol
IGRP and EIGRP
Reason for IGRP?

At
Cisco Routing Protocols
this time there was no alternative for RIP available

disadvantages:
RIP
Metric limitation
{
{
{
{
max. 15 hops (16 hops = network unreachable)

Hop count doesn't reflect the capacity of the transmission medias
takes the least hop path instead of the fastest or best path
=> didn't allow flexible routing in complex environments
much routing overhead (full routing table every 30seconds)

2005, D.I. Manfred Lindner
IGRP-EIGRP, v3.5
IGRP-EIGRP, v3.5
Agenda
IGRP at a glance
IGRP
EIGRP
Distance vector protocol (can only be used within

an Autonomous System)
Composite metric
Bandwidth
Delay
Reliability
Loading
Implementation of loop avoidance mechanisms

Support of multiple unequal-metric paths
Faster convergence than RIP
IGRP-EIGRP, v3.5
IGRP-EIGRP, v3.5
Page 43 - 1
Page 43 - 2
IGRP / EIGRP Metric calculation
Bandwidth
unit:
10 microseconds
is the sum of the transmission delays along the path
and is stored in a 32-bit field, in increments of 39.1
nanoseconds
unit:
bits/sec
default values for LANs
it
corresponding to real bandwidth
default
values for serial lines
all 1s indicates an unreachable destination
corresponding to bandwidth of 1.544Mbps (T1 line)
default
configuration of the real bandwidth on serial lines is a must!!!

cisco interface command: bandwidth <number in kbit/s>
default
bandwidth along a path is taken

BWIGRP is expressed by (1/bandwidth)*1017
BWEIGRP is expressed by (1/bandwidth)*1017*256
configuration of the real delay on serial lines is an option

cisco interface command: delay <number in tens of usec>
DelayIGRP
range is from a 1200bps line to 10 terabits per

second
DelayEIGRP
5
is expressed by (delay/10)
is expressed by (delay/10)*256
IGRP-EIGRP, v3.5
Bandwidth
metric
values for serial lines
corresponding to delay of 1.544Mbps (T1 line)
The
IGRP-EIGRP, v3.5
values for LANs
corresponding to real delay
minimum
Delay
values for bandwidth as part of composite metric:
Delay
metric
assumption: bandwidth for serial links is configured properly
values for delay as part of composite metric:
assumption: delay for serial links is default value for T1
otherwise all serial links will have metric of T1

Delay
Bandwidth
Satellite (500 Mbit/s)
Ethernet (100 Mbit/s)
Ethernet (10 Mbit/s)
Token Ring (4 Mbit/s)
Token Ring (16 Mbit/s)
FDDI (100 Mbit/s)
1.544 Mbps
128 kbps
64 kbps
56 kbps
10 kbps
1 kbps
BWEIGRP
BWIGRP
5.120
256.000
256.000
640.000
160.000
256.000
1.657.856
20.000.000
40.000.000
45.714.176
256.000.000
2.560.000.000
20
100
1.000
2.500
625
100
6.476
78.125
156.250
178.571
1.000.000
10.000.000
IGRP-EIGRP, v3.5
DelayEIGRP
51.200.000
100 Mbit Ethernet (0,1 ms) 25.60
10 Mbit Ethernet (1 ms)
25.600
Token Ring 4 ( 2,5ms)
64.000
Token Ring 16 ( 0,6ms)
16.000
FDDI 100 ( 0,1ms)
2.560
serial links:
1.544 Mbps (20 ms)
512.000
128 kbps (20 ms)
512.000
64 kbps (20 ms)
512.000
56 kbps (20 ms)
512.000
10 kbps (20 ms)
512.000
1 kbps (20 ms)
512.000
Satellite (2 sec)
IGRP-EIGRP, v3.5
Page 43 - 3
Page 43 - 4
DelayIGRP
200.000
10
100
250
62,5
10
2.000
2.000
2.000
2.000
2.000
2.000
8
IGRP Metric calculation
Reliability
arbitrary
number
means 0%
worst
reliability between source and destination

dynamically measured
k1=1, k2=0, k3=1, k4=0, k5=0

K1 til K5 are arbitrary numbers which can be configured
metric1 = k1 min BWIGRP +
keepalives are sent off the interface every 10 seconds, frame has
k 2 min BWIGRP
+ k 3 sumDelay IGRP
256 - load
If k5 doesn
doesnt equal 0, an additional operation is done
CRC
samples are calculated over 5 minutes
time
defaults:
255 means 100%

1
Formula for metric calculation
compositem etric = metric1
interval needs reconfiguration, when Reliability is
k5
reliabilit y + k 4
used
For default values of k parameters:
compositem etric = k1 min BWIGRP + k 3 sumDelay IGRP
IGRP-EIGRP, v3.5
IGRP-EIGRP, v3.5
11
IGRP Metric example 1
Loading
arbitrary
Prefered Path!!!
number
Load is given as a fraction of 255. A load of 255 indicates a
Metric=BW+delay=78125+2000=
80125
Metric=BW+delay=78125+2000=80125
completely saturated link

dynamically
measured
128kbps
calculated over 5 minutes

128kbps
time interval needs reconfiguration, when Loading is used
64kbps
Metric=min.BW
Metric=min.BW along the path+sum of delays =156250+2000+2000=160250
=156250+2000+2000=160250
IGRP-EIGRP, v3.5
10
IGRP-EIGRP, v3.5
Page 43 - 5
Page 43 - 6
12
IGRP Metric example 2
IGRP multiple paths

Configured variance=2
80125
80125
IP
IP
IP
IP
128kbps
64kbps
!!!
th
Pa
d
e
er
ef
Pr
Metric=min.BW
=156250+2000+2000=160250
Metric=min.BW
=6476+2000+2000=10476
IGRP-EIGRP, v3.5
13
IGRP multiple paths
IGRP-EIGRP, v3.5
15
IGRP Metric
Support of up to 6 parallel paths (default:4) to the

destination for load balancing
default:
IP
IP
1,544Mbps
Pr
efe
re
dP
ath
!!!
128kbps
IP
1,544Mbps
IP
IP
128kbps
same metric for parallel paths is necessary
Routing updates also include a count of hops and a

computation of the path MTU
=>
max. network diameter 255 hops (IP TTL-field)

100 hops
default:
Support of unequal-metric load balancing

Prerequisite:
configuration of a variance factor

alternative path metric must be within the specified
variance of the best local metric
The
IGRP-EIGRP, v3.5
14
IGRP-EIGRP, v3.5
Page 43 - 7
Page 43 - 8
16
IGRP & default routes
IGRP Loop avoidance & timers
RIP and OSPF are using 0.0.0.0 as their default

route (=>metric is not related to a distance)
IGRP allows to mark real networks as candidates
for being default
Routing update timer: 90 seconds
Hold-down timer: (3*90sec)+10sec=280sec
Invalid timer: 3*90sec=270sec
e.g.
if multiple exit points to the Internet exist, then this

implementation allows to choose the optimal border
router
(starts
in the absence of routing information about a

specific route)
candidate for being default
IGRP
194.96.4.4/30
Internet
Flush timer: 7*90sec=630sec

How
much time should pass before a route should be

flushed from the routing table
194.96.4.8/30
candidate for being default

IGRP-EIGRP, v3.5
17
IGRP Loop avoidance & timers
19
IGRP Protocol stack

OSI stack
Periodical Routing updates

destination
IGRP-EIGRP, v3.5
address: 255.255.255.255
Topology changes are reported with triggered

updates
Hold-downs
Split horizon
Route Poisoning Updates
IGRP
9
are
intended to defeat larger routing loops

sent if a route metric has increased by a factor of 1.1
or greater
Data link layer
are
IGRP-EIGRP, v3.5
IP protocol number 9 (decimal)
IP
Physical layer
18
IGRP-EIGRP, v3.5
Page 43 - 9
Page 43 - 10
20
Agenda
EIGRP at a glance
IGRP
EIGRP
Advanced distance vector protocol (can only be

used within an AS)
uses exactly the same metric as IGRP
Loop avoidance by DUAL
DUAL:
Diffusing Update Algorithm
evolved by Mr. J.J. Garcia-Luna-Aceves
IGRP-EIGRP, v3.5
21
EIGRP Background
advantages:
disadvantages:
IGRP-EIGRP, v3.5
23
Support of multiple unequal-metric paths

Classless routing protocol (supports route
summarization)
proprietary protocol
tries to combine the advantages of the distance vector and the link
state protocol world without their specific problems
Distance vector
EIGRP at a glance
EIGRP stands for Enhanced Interior Gateway Routing

Protocol
Event triggered updates (Multicasts)

Fast convergence
Update
contains network plus prefix
It supports IP, IPX and Appletalk
less CPU power and memory usage, simple configuration

slow convergence, lot of routing overhead, possibility of loops
Link state
advantages:
disadvantages:
fast convergence, no loops, better metric (cost factor)

high CPU power and memory usage (SPF algorithm, LS database), not so
easy to configure (area concept)
IGRP-EIGRP, v3.5
22
IGRP-EIGRP, v3.5
Page 43 - 11
Page 43 - 12
24
EIGRP Concepts
EIGRP Concepts
R2
R8
Hello
1st : Every EIGRP router has to discover its

neighbors
Hello
Hello
R3
its
done with a Hello protocol

a Router doesnt expect an acknowledge for its Hello
network type dependent
R1
R5
Hello
Hello
Hello
Hello
R4
Hello
FDDI
NBMAs nonbroadcast multiaccess network
2nd: based on this Hellos it builds a neighbor table

When
a newly discovered neighbor is learned, the

address and interface of the neighbor is recorded.
The HoldTime is the amount of time a router treats a
neighbor as reachable and operational: 3* Hello interval
IGRP-EIGRP, v3.5
25
EIGRP Concepts
N
e
t
w
o
r
k
X
R7
Hello
Multiaccess with broadcast/multicast support (BMA)
Hello
Hello
Point-to-point network
Hello
R6
Hello
Hello
Hello
Token
Ring
Hello
Neighbor table R1
Neighbor table R2
Neighbor table R3
Neighbor table R4
R2
R1
R1
R1
R3
R3
R2
R2
R4
R4
R8
R4
R5
R3
R7
IGRP-EIGRP, v3.5
27
EIGRP Concepts
PointPoint-toto-Point:
Point: Neighbor relationship is formed
with the router on the other end
Hello time interval: 5 sec
3rd:After the neighbor discovery a Topology table is

built
Neighbor
routers exchanging their complete routing

tables and store these informations in a Topology table
information exchange through Update packets
Update packets
Broadcast multiaccess:
multiaccess: Neighbor relationships
are formed dynamically using multicast hellos
Hello time interval: 5 sec
Multicast address: 224.0.0.10
contain a sequence number field in the header and must be
acknowledged by the receiver (reliable transmission)

are sent in the following instances:
{
Nonbroadcast multiaccess:
multiaccess: Neighbor relationships
are formed by manual configuration
Hello time interval: 60sec
Dest. Address: Unicast address of the
neighbor
IGRP-EIGRP, v3.5
F/R
X.25
ATM
26
when a neighbor first comes up (packets dest. addr is an unicast)

when a network has failed (packets dest. addr. is 224.0.0.10)
when there is a metric change for a certain destination (packets
dest. addr. is 224.0.0.10)
IGRP-EIGRP, v3.5
Page 43 - 13
Page 43 - 14
28
EIGRP Concepts
EIGRP Concepts
Update packets
Topology Table
In
contrast to OSPF every EIGRP router is modifying any

received update packet. It adds its own local distance to
the information and sends the packet with an own
sequence number to its neighbors.
Stores
routing information that neighbors exchange after

the first Hello exchange (=> smaller compared to an
OSPF topology table). DUAL acts on the Topology table
to determine Successors and FSs.
Successor:
{
A neighbor that has been selected as the next hop for a destination,
it ends up in the Routing Table
Feasible Successor (FS):

{
A neighbor that has satisfied the Feasibility Condition and has a path
to the destination (is an alternate route to the current successor)
Feasibility Condition (FC):

{
IGRP-EIGRP, v3.5
29
EIGRP Concepts
A condition that is met when the lowest of all the neighbors' costs
plus the link cost to that neighbor is found, and the neighbor's
advertised cost is less than the current successor's cost.
IGRP-EIGRP, v3.5
31
EIGRP Topology table and feasible successor

R2
R2
Ack.
NetY ad:20
Update
Ack.
10
R3
R1
NetY ad:10
Update
Ack.
NetY ad:20
Update
R5
NetY ad:120
Update
100
10
R4 NetY ad:20
Ack.
NetY
ad:40
Update
R6 NetY ad:220
Update
100
Ack.
FDDI
R3
R1
10 Net X ad:216
R7
Net X ad:17
Token 6
Ring
FDDI
Update
Part of R1
R1s Topology Table
Network
Update propagation
Successor
ad = advertised distance
Feasible Successor
IGRP-EIGRP, v3.5
10
R6
100
N
e
t
w
o
r
k
X
NetY ad:21
R5
100
R4
Token 6
Ring
R7
Ack.
16
Update
New network Y
N
e
t
w
o
r
k
10
Ack.
R8
Net X ad:26
R8
16
30
Advertised Distance
Feasible Distance
Neighbor
17
27
R4
216
226
R3
26
36
R2
ad = advertised
distance
IGRP-EIGRP,
v3.5
Page 43 - 15
Page 43 - 16
32
EIGRP Topology table without

feasible successor
EIGRP Active state
R2
R8
Net X ad:36
26
R3
10 Net X ad:216
R1
R5
100
10
R6
100
Active
N
e
t
w
o
r
k
that destination and has no other feasible successor (FS)

available. The router is forced to compute a route to the
destination.
Its sending a query packet to all its neighbors.
Query packet (will be acknowledged from the receiver)
R4
R7
Net X ad:17
Token 6
Ring
FDDI
Part of R1
R1s Topology Table
Successor
no Feasible
Successor!!!
Advertised Distance
Feasible Distance
Neighbor
17
27
R4
216
226
R3
36
46
R2
ad = advertised
distance
IGRP-EIGRP,
v3.5
Sent to all neighbors when a router goes into Active for a destination
and is asking for information on that destination. Unless it receives
replies back from all its neighbors, the router will remain in Active
state and not start the computation for a new successor.
Reply packet (will be acknowledged from the receiver)

{
Network
state
A router's state for a destination when it has lost its successor to
33
Sent by every EIGRP neighbor which receives a query. If the

neighbor doesn't have the information, it queries its neighbors
indicating that it is also performing route recomputation
IGRP-EIGRP, v3.5
35
EIGRP Topology table

without feasible successor
EIGRP Active state
R2
Passive
Active State
R1
state
R3
Query 1
Ack 9
destination available in its Topology table
R5
100
10
R6
100
Ack 1
N
e
t
w
o
r
k
X
R4
R7
Token 6
Ring
FDDI
10
Part of R1
R1s Topology Table
Network
Successor
no Feasible
Successor!!!
IGRP-EIGRP, v3.5
Reply 17
Ack 17
If the successor disappears from the topology table

because of a network change and there is no
feasible successor, DUAL puts the route into the
active state.
26
Ack 1
A router's state after losing its successor when it has an FS to the
R8
Reply 9
If the successor disappears from the topology table

because of a network change and there is a
feasible successor, DUAL keeps the route in a
passive state.
34
Advertised Distance
Feasible Distance
Neighbor
17
27
R4
216
226
R3
36
46
R2
IGRP-EIGRP, v3.5
Page 43 - 17
Page 43 - 18
36
EIGRP Topology table

without feasible successor
EIGRP Compatibility
R2
R8
Ack
26
R3
R1
Update
Ack
R5
100
R6
10
100
automatic redistribution
N
e
t
w
o
r
k
manual redistribution
X
Passive State
R4
10
R7
IGRP
AS100
EIGRP
AS100
Token 6
Ring
IGRP
AS 397
EIGRP
AS100
FDDI
Part of R1
R1s Topology Table
manual redistribution
Network
Advertised Distance
Successor
Feasible Distance
Neighbor
216
226
R3
36
46
R2
IGRP-EIGRP, v3.5
37
EIGRP Compatibility
IGRP-EIGRP, v3.5
39
EIGRP Protocol stack

OSI stack
Route tagging
EIGRP
all other IP
routing
protocols
EIGRP
AS100
has the notion of internal and external routes.
Internal routes
are ones that have been originated within an EIGRP autonomous

system (AS).
External routes
are ones that have been learned by another routing protocol or
reside in the routing table as static routes.
EIGRP
88
Route redistribution
in
the case of IGRP is done automatically, when EIGRP

and IGRP are belonging to the same Autonomous
System (compatible metric!!!). IGRP derived routes are
treated as external routes in EIGRP (also OSPF, RIP,
EGP, BGP...)
IP protocol number 88 (decimal)
IP
IGRP-EIGRP, v3.5
Data link layer

Physical layer
38
IGRP-EIGRP, v3.5
Page 43 - 19
Page 43 - 20
40

P43 IGRP EIRGP - v3 5 PDF

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

P43 IGRP EIRGP - v3 5 PDF

Diunggah oleh

Hak Cipta:

Format Tersedia

Institute of Computer Technology - Vienna University of Technology

Institute of Computer Technology - Vienna University of Technology

L43 - IRGP and EIGRP

L43 - IRGP and EIGRP

IGRP stands for Interior Gateway Routing Protocol

and deployed successfully in 1986 by

IGRP and EIGRP

Reason for IGRP?

Cisco Routing Protocols

this time there was no alternative for RIP available

max. 15 hops (16 hops = network unreachable)

much routing overhead (full routing table every 30seconds)

2005, D.I. Manfred Lindner

Distance vector protocol (can only be used within

Implementation of loop avoidance mechanisms

2005, D.I. Manfred Lindner

2005, D.I. Manfred Lindner

2005, D.I. Manfred Lindner

2005, D.I. Manfred Lindner

Institute of Computer Technology - Vienna University of Technology

Institute of Computer Technology - Vienna University of Technology

L43 - IRGP and EIGRP

L43 - IRGP and EIGRP

IGRP / EIGRP Metric calculation

IGRP / EIGRP Metric calculation

corresponding to real bandwidth

values for serial lines

all 1s indicates an unreachable destination

corresponding to bandwidth of 1.544Mbps (T1 line)

configuration of the real bandwidth on serial lines is a must!!!

bandwidth along a path is taken

configuration of the real delay on serial lines is an option

range is from a 1200bps line to 10 terabits per

IGRP / EIGRP Metric calculation

2005, D.I. Manfred Lindner

IGRP / EIGRP Metric calculation

values for serial lines

corresponding to delay of 1.544Mbps (T1 line)

values for LANs

corresponding to real delay

2005, D.I. Manfred Lindner

values for bandwidth as part of composite metric:

assumption: bandwidth for serial links is configured properly

values for delay as part of composite metric:

assumption: delay for serial links is default value for T1

otherwise all serial links will have metric of T1

2005, D.I. Manfred Lindner

2005, D.I. Manfred Lindner

2005, D.I. Manfred Lindner

Institute of Computer Technology - Vienna University of Technology

Institute of Computer Technology - Vienna University of Technology

L43 - IRGP and EIGRP

L43 - IRGP and EIGRP

IGRP Metric calculation

IGRP Metric calculation

reliability between source and destination

k1=1, k2=0, k3=1, k4=0, k5=0

metric1 = k1 min BWIGRP +

255 means 100%

Formula for metric calculation

compositem etric = metric1

interval needs reconfiguration, when Reliability is