advances in eigrp

8/14/2019 Advances in EIGRP

1/46

Copyright 2001, Cisco Systems, Inc. All rights reserved. Printed in USA.

9859_05_2004_X.scr

1 2004 Cisco Systems, Inc. All rights reserved.

RST-4300

9859_05_2004_X

ADVANCES IN EIGRP

RST-4300


RST-4300

9859_05_2004_X

Advances in EIGRP

Scaling Enhancements

Network StabilityEnhancements

Other Enhancements


2/46


9859_05_2004_X.scr

SCALING ENHANCEMENTS


RST-2310

9636_05_2004_c1


RST-4300

9859_05_2004_X

Scaling Enhancements

EIGRP Stubs

Single Peering Over ParallelLinks


3/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

When a router runningEIGRP loses its connectionto a network, it firstsearches for alternate loopfree paths

If it finds none, it then sendsqueries to each of itsneighbors, looking for analternate path

BA

10.1.1.0/24

router-a#sho ip eigrp topo

IP-EIGRP Topology Table

....

P 10.1.1.0/24, 1 successors, FD is 281600

via Connected, Ethernet1/2

router-a#show ip eigrp events

Event information for AS 100:

....

12 Active net/peers: 10.1.1.0/24 1

14 FC not sat Dmin/met: 4294967295 128256

15 Find FS: 10.1.1.0/24 128256

....

18 Conn rt down: 10.1.1.0/24 Ethernet 3/1


RST-4300

9859_05_2004_X

EIGRP Stubs

If the neighbor has no path tothis destination, it replies

The router then removes allreferences to this route from itslocal tables

In large hub and spokenetworks, the hub routers haveto build queries and processreplies from each of the spokes

This limits scaling!

router-a#show ip eigrp events

Event information for AS 100:

1 NDB delete: 10.1.1.0/24 1

....

12 Active net/peers: 10.1.1.0/24 1

14 FC not sat Dmin/met: 4294967295 128256

15 Find FS: 10.1.1.0/24 128256....

18 Conn rt down: 10.1.1.0/24 Ethernet 3/1

BA

10.1.1.0/24


4/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

If these spokes are remotessites, they have twoconnections for resiliency, notso they can transit trafficbetween A and B

A should never use the spokesas a path to anything, sotheres no reason to learnabout, or query for, routesthrough these spokes

BA

10.1.1.0/24

dont use these paths


RST-4300

9859_05_2004_X

EIGRP Stubs

To signal A and B that the pathsthrough the spokes should notbe used, the spoke routers canbe configured as stubs

router#config t#

router(config)#router eigrp 100

router(config-router)#EIGRP stub connected

router(config-router)#

BA

10.1.1.0/24


5/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

Marking the spokes as stubsallows them to signal A and Bthat they are not valid transitpaths

A will not query stubs,reducing the total number ofqueries in this example to 1

Marking the remotes as stubsalso reduces the complexity ofthis topology; B now believes itonly has 1 path to 10.1.1.0/24,rather than 5

marke

das

stubs

BA

10.1.1.0/24


RST-4300

9859_05_2004_X

EIGRP Stubs

If stub connectedisconfigured

B will advertise 10.1.2.0/24to A

B will not advertise10.1.2.0/23, 10.1.3.0/23, or10.1.4.0/24

If stub summaryisconfigured



ip route 10.1.4.0 255.255.255.0 10.1.1.10

!

interface serial 0

ip summary-address eigrp 10.1.1.0 255.255.254.0

!router eigrp 100

redistribute static 1000 1 255 1 1500

network 10.2.2.2 0.0.0.1

network 10.1.2.0 0.0.0.255

10.1.2.0/24

A

B

10.2.2.2/31

10.1.3.0/24

eigrp stub connected

eigrp stub summary


6/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

If stub staticis configured



If stub receive-onlyisconfigured

B wont advertise anythingto A, so A needs to have astatic route to the networksbehind B to reach them

ip route 10.1.4.0 255.255.255.0 10.1.1.10

!

interface serial 0

ip summary-address eigrp 10.1.1.0 255.255.254.0

!

router eigrp 100

redistribute static 1000 1 255 1 1500

network 10.2.2.2 0.0.0.1network 10.1.2.0 0.0.0.255

eigrp stub receive-only

eigrp stub static

A

B

10.2.2.2/31

10.1.3.0/24

10.1.2.0/24


RST-4300

9859_05_2004_X

EIGRP Stubs

Any combination of the route types can be specified on theeigrp stub statement, except receive-only, which cannot beused with any other option

For example:

eigrp stub connected summary redistributed

If eigrp stubis specified without any options, it will sendconnected and summary routes


7/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

At A, you can tell B is a stubusing show ip eigrpneighbor detail.

10.1.2.0/24

A

B

10.2.2.2/31

10.1.3.0/24

router-a#show ip eigrp neighbor detail

IP-EIGRP neighbors for process 100

H Address Interface Hold Uptime SRTT RTO Q Seq Type

(sec) (ms) Cnt Num

0 10.2.2.3 Et1/2 10 00:00:50 320 1920 0 7

Version 12.2/1.2, Retrans: 0, Retries: 0

Stub Peer Advertising ( CONNECTED ) Routes


RST-4300

9859_05_2004_X

EIGRP Stubs

EIGRP stubs are available in 12.0(6.3)T 12.0(6.1)PI12.0(6.3)XE01(0.35)

http://www.cisco.com/en/US/partner/products/sw/iosswrel/ps1830/products_feature_guide09186a00800ab721.html


8/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

Future Enhancement

Dual router remotes are notsupported with EIGRP stubstoday

C and D are a single remote site

If we mark C and D as stubrouters

C wont advertise a defaultlearned from A to D

D wont advertise a defaultlearned from B to C

If the B to D link fails, routing

from 10.1.1.0/24 failsD doesnt have any routetowards the core

BA

10.1.1.0/24

C

D

remote site

stub

0.0.0.0/0 0.0.0.0/0

nodefaultadvertised


RST-4300

9859_05_2004_X

EIGRP Stubs

Future Enhancement

What if we have a mixture ofstub capabilities on a singleinterface?

Some routers can bedeclared as stubs

Other routers cannot bedeclared stubs

EIGRP stubs currently dont

support mixed modes on thesame interface

BA

can be marked as stubs

cannot bemarked asstubs


9/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Stubs

Future Enhancement

CSCdx74716 would allowmixing stub and non-stubneighbors on shared media

Useful for Ethernet in thedata center

Multipoint frame relay usedto connect to remotes andlateral connections withinthe same network layer

BA

can be marked as stubs

cannot bemarked asstubs


RST-4300

9859_05_2004_X

EIGRP Stubs

Future Enhancement

EIGRP Stub Site

Route-maps used on C and D toidentify which routes are leakedthrough which interfaces

C would allow advertising10.1.1.0/24 on interface towardA, and 0.0.0.0/0 on interfacetoward D

C would not advertise 0.0.0.0/0toward A, since its not on the

allowed list This allows both routers to be

marked as stubs

The site wont ever be used totransit traffic between A and B

BA

10.1.1.0/24

C

D

remote site

stub

0.0.0.0/0 0.0.0.0/0

leak 0.0.0.0/0

leak 10.1.1.0/24 to A


10/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Single Peering

Future Enhancement

EIGRP currently peers overevery link between a pair ofrouters

Each route learned at A willbe advertised to B overevery neighbor relationship

Each route placed in theactive state will cause aquery through everyneighbor relationship

A

B

new route


RST-4300

9859_05_2004_X

paths installedin Bs routing

table

paths notinstalled in Bsrouting table

Single Peering

Future Enhancement

Slows down networkconvergence

EIGRP only split horizons onroutes actually installed in therouting table

If there are more paths than themaximum paths EIGRP caninstall in the routing table....

The routers end up queryingeach other, which can cause

major problems in networkconvergence times

A

B

queries due tolost route

queriesbecause splithorizon isdisabled onthese links


11/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Single Peering

Future Enhancement

EIGRP will eventually peerbased on router IDs rather thaninterface addresses

Only one neighbor relationshipno matter how many linksbetween the routers

Reduces routing traffic

Increases convergence speed

Decreases active timesthroughout the network

A

B

singlerelationship

NETWORK STABILITYENHANCEMENTS


RST-2310

9636_05_2004_c1


12/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Network Stability Enhancements

EIGRP SIA Rewrite

Multiple AutonomousSystem Support

Neighbor ContinuityEnhancements

EIGRP 3-way Handshake

EIGRP Graceful Shutdown

EIGRP Graceful Restart


RST-4300

9859_05_2004_X

EIGRP SIA Rewrite

A loses its route to 10.1.1.0/24

No FS, mark route active

Set a 3 minute active timer

Query all neighbors (B)

B receives As query

No FS, mark route active

Set 3 minute active timer

Query all neighbors (C)

C receives Bs query

Examine local topology table

No feasible successors

No neighbors to query!

A

B

C

10.1.1.0/24

10.1.10/24 gone; no FS

query

query

10.1.10/24 gone; no FS

10.1.10/24 gone


13/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP SIA Rewrite

C has no alternate path to10.1.1.0/24

Remove from local tables

Reply to querying neighbors

B receives Cs reply

No outstanding queries


Reply to querying neighbors

A receives Bs reply

No outstanding queries


A

B

C

10.1.1.0/24

10.1.10/24 gone; no FS

query

query

10.1.10/24 gone; no FS

10.1.10/24 gone

reply

reply

remove 10.1.1.0/24

remove 10.1.1.0/24

remove 10.1.1.0/24


RST-4300

9859_05_2004_X

EIGRP SIA Rewrite

If C sends the reply, and Bnever receives it, whathappens?

As active timer (3 minutes) isstill counting down while B andC are trying to get the replyback

When this timer expires, Adeclares an SIA

The A/B neighbor relationship isreset

Why reset A/B when B/C is theproblem??

A

B

C

10.1.1.0/24

10.1.10/24 gone; no FS

query

query

10.1.10/24 gone; no FS

10.1.10/24 gone

reply

remove 10.1.1.0/24

bad link, replynever makes it

reset relationship!


14/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP SIA Rewrite

After the SIA rewrite

12.1(4.0.3)T and 12.1(4.1)

A sets its active timer to halfthe configured active time (1.5minutes, normally)

After this time has passed, Asends an SIA Query

If B acknowledges this query, Aresets its timer, and the A/Bneighbor relationship stays up

Bs relationship with C will failat some point

This clears the query from Bspoint of view

B replies to A

A

B

C

10.1.1.0/24

10.1.10/24 gone; no FS

query

query

10.1.10/24 gone; no FS

10.1.10/24 gone

reply

remove 10.1.1.0/24

bad link, replynever makes it

SIA query


RST-4300

9859_05_2004_X

Multiple AS Support

Do You Really Want to Do This?

Okay, maybe its not that bad

But we still wouldnt recommend it


15/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Multiple AS Support

A route is redistributed fromRIP into AS 200

At A, it is redistributed into AS100

B receives this route as well;which of the two externals will itprefer?

There are two routes learnedthrough separate routingprocesses with the sameadministrative distance, so theroute installed first wins

AS 100

AS 200

RIP

A B

C


RST-4300

9859_05_2004_X

Multiple AS Support

If router B prefersthe route through AS 100, itwill redistribute the routeback into AS200

If the redistribution metric atB is lower than theredistribution metric at C, Awill prefer the path throughB

We have a permanent loop!

AS 100

AS 200

RIP

A B

metr

ic1000

C

met

ric500


16/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

AS 100

AS 200

RIP

A B

Multiple AS Support

External routescan also carryadministrative tags; as theexternal route isredistributed into AS 100 atA, it can be tagged

This tag can then be used toblock the redistribution ofthe route back into AS 200 atB

tag 100

route-map settag deny 10match tag 200

route-map settag permit 20

!

router eigrp 200

redistribute eigrp 200 route-map settag

route-map settag permit 10set tag 200

!

router eigrp 100


....


RST-4300

9859_05_2004_X

AS 100

AS 200

RIP

A B

Multiple AS Support

This blocks the formation ofthe loop, since A will nolonger receive theredistributed from B throughAS 200

B still receives both routes,however, and could stillchoose the path through AS100, resulting in suboptimal

routing

tag 100


set tag 200

!

router eigrp 100


....

route-map filtertag deny 10

match tag 200

route-map filtertag permit 20

!

router eigrp 200

redistribute eigrp 100 route-map filtertag


17/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Multiple AS Support

CSCdm47037 resolves the routing loop and thesuboptimal routing

If two routes with the same administrativedistances are compared, and the process type isthe same (both EIGRP), then compare the metricsof the routes as well

http://www.cisco.com/cgi-bin/Support/Bugtool/onebug.pl?bugid=CSCdm47037


RST-4300

9859_05_2004_X

Multiple AS Support

If the redistribution metric isnot manually set at A, it willbe carried from AS 200 into100

The cost of the pathbetween A and B is thenadded at B

At B, the route through AS200 wins; it has the lower

metric

AS 100

AS 200

RIP

A B

metric 1000

metric 1500

IP-EIGRP Topology Table forAS(100)/ID(10.0.17.10)

....


via 10.0.6.4 (1500/1000), FastEthernet0/0

....

IP-EIGRP Topology Table forAS(200)/ID(10.2.17.10)

....


via 10.2.8.20 (1000/256256), FastEthernet0/1


18/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Multiple AS Support

CSCdt43016, Support for Incoming Route FilteringBased on Route Maps, makes it possible to filterroutes based on any route map condition before itis accepted into the local routing protocol database

This is listed as an OSPF feature, but it works for allrouting protocols

http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122relnt/xprn122t/122tnewf.htm#33626


RST-4300

9859_05_2004_X

AS 100

AS 200

RIP

A B

Multiple AS Support

This blocks the formation ofthe loop, since B will nolonger have the pathredistributed from A into AS100 in its topology table

This also prevents thesuboptimal routing

tag 100


set tag 200

!

router eigrp 100


....

route-map settag deny 10

match tag 200


!

router eigrp 100

distribute-list filtertag filter in


19/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Neighbor Continuity Enhancements

Future Enhancement

EIGRP currently resets itsneighbor relationships for

Filtering configurationchange

Summarizationconfiguration

Others....

B

A

router#config trouter(config)#router eigrp 100router(config-rtr)#distribute-list 100 in serial 0/0

*Aug 27 13:06:26.758: %DUAL-5-NBRCHANGE:IP-EIGRP(0) 100: Neighbor 10.1.1.1(Serial0/0) is down: peer restarted*Aug 27 13:06:27.976: %DUAL-5-NBRCHANGE:

IP-EIGRP(0) 100: Neighbor 10.1.1.1(Serial0/0) is up: new adjacency


RST-4300

9859_05_2004_X

Neighbor Continuity Enhancements

Future Enhancement

EIGRP will use gracefulrestart techniques in thefuture to reduce neighborresets to the minimum

B

A

router#config trouter(config)#router eigrp 100router(config-rtr)#distribute-list 100 in serial 0/0

Graceful Restart


20/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP 3-Way Handshake

During normal adjacencyformation, multicast helloscause the EIGRP process toplace new neighbors in theneighbor table

Unicast packets are then usedto exchange known routinginformation, and complete theneighbor relationship

But what happens if the link isunidirectional, or one peerrestarts while building therelationship?

A

B

multicasthello

unicastupdate+init

unicastacknowledgement

unicasttopologytableinfo


RST-4300

9859_05_2004_X


If the link is unidirectional, Awill receive Bs hello, andtransmit an update with the initbit set

A never receives anacknowledgement for thispacket, so it retransmits untilthe retransmission timeout isexceeded

Once this timeout is exceeded,the neighbor is torn down; the

process repeats when Areceives another hello from B

A

B

multicasthello

unicastupdate+init

unicastupdate+init

unicastupdate+init

....


21/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


A receives a hello, sends theupdate with the initializationbit set, and B acknowledgesit. A begins sendingtopology table information

B restarts; once it comesback up, it will send anupdate with the initializationbit set

A acknowledges this packet,and continues sendingtopology information to B

A

B

multicasthello

unicastupdate+init



Brestarts

unicastupdate+init




RST-4300

9859_05_2004_X


A and B are out ofsynchronization, but Adoesnt know this

B ignores the routinginformation A is sending,since it has lost its neighborstate with A

Eventually, after A hasretransmitted the samepacket several times, it will

tear down the neighborrelationship

A

B

multicasthello

unicastupdate+init

un

icastacknowledgement


Brestarts

unicastupdate+init

un

icastacknowledgement



22/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


To prevent EIGRP from formingneighbor relationships undereither of these conditions, anew neighbor state is created:the pending state

When A receives the firstmulticast hello from B, it placesB in the pending state, andtransmits a unicast update withthe initialization bit set

While B is in this state, A willnot send it any queries orrouting information

A

B

multicasthello

unicastupdate+init

Binpending


RST-4300

9859_05_2004_X


When B receives this updatewith the init bit set, it sends anupdate with the init bit set aswell

The acknowledgement for Asinitial update is piggybackedonto this packetit is nevertransmitted by itself

There is no way for A to receivethe acknowledgement for its

initial update without alsoreceiving Bs initial update

A

B

multicasthello

unicastupdate+init

Binpending

un

icastupdate+init+ack


23/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


Once the acknowledgement forits initial update is received, Atakes B out of the pendingstate, and begins sending ittopology information

If this acknowledgement isntever received, hellos from Bare ignored while A attempts toretransmit the initial update

Eventually, A will time B out,and the process will start over

A

B

multicasthello

unicastupdate+init

Binpending

unicastupdate+init+ack

Boutofpending



RST-4300

9859_05_2004_X

EIGRP 3-way Handshake

Committed in 12.2(13.7)T2, 12.2(15.1)S, 12.2(16.1)Bas CSCdy45118

http://www.cisco.com/cgi-bin/Support/Bugtool/onebug.pl?bugid=CSCdy45118


24/46


25/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

GR/NSF Fundamentals

NonStop Forwarding (NSF) is a way to continueforwarding packets while the control plane isrecovering from a failure.

Graceful Restart (GR) is a way to rebuild forwardinginformation in routing protocols when the controlplane has recovered from a failure.

The fundamental premise of NSF/GR is to routethrough temporary failures, rather than aroundthem!


RST-4300

9859_05_2004_X

EIGRP Graceful Restart/NSF

Router A loses its controlplane for some period oftime.

It will take some time forRouter B to recognize thisfailure, and react to it.

Control Data A

Control Data B


26/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


During the time that A hasfailed, and B has notdetected the failure, B willcontinue forwarding trafficthrough A.

Once the control planeresets, the data plane willreset as well, and this trafficwill be dropped.

NSF reduces or eliminatesthe traffic dropped while Ascontrol plane is down.

Control Data A

reset

Control Data B


RST-4300

9859_05_2004_X


If A is NSF capable, thecontrol plane will not resetthe data plane when itrestarts

Instead, the forwardinginformation in the data planeis marked as stale.

Any traffic B sends to A willstill be switched based on

the last known forwardinginformation.

Control Data A

no reset

Control Data B

mark forwardinginformation as stale


27/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


While As control plane isdown, the routing protocol holdtimer on B counts down....

A has to come back up andsignal B before Bs hold timerexpires, or B will route around it

When A comes back up, itsignals B that it is stillforwarding traffic, and wouldlike to resync

This is the first step in GracefulRestart (GR)

Hold Timer: 1514131211109876

Control Data A

Control Data B


RST-4300

9859_05_2004_X


The signal in EIGRP is anupdate with the initializationand restart (RS) bits set

A sends its hellos with therestart bit set until GR iscomplete

B transmits the routinginformation it knows to A

When B is finished sendinginformation, it sends a specialend of table signal so A knows

the table is complete

Control Data

Control Data

A

B

topologyinformation

hello+Restart

init+Restart

endoftable


28/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


When A receives this end oftable marker, it recalculatesits topology table, andupdates the local routingtable

When the local routing tableis completely updated,EIGRP notifies CEF

CEF then updates theforwarding tables, andremoves all informationmarked as stale

Control Data A

Control Data B


RST-4300

9859_05_2004_X


eigrp nsfenables gracefulrestart

show ip protocolsverifiesgraceful restart isoperational

http://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a0080160010.html

A

B

router eigrp 100

eigrp nsf

....

router eigrp 100

eigrp nsf

....

router#show ip protocols

Routing Protocol is "eigrp 100

....

Redistributing: eigrp 100

EIGRP NSF-aware route hold timer is 240s

Automatic network summarization is in effect

Maximum path: 4

....


29/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


Routing protocol graceful restart is supported in IOS 12.2(15)T

NonStop Forwarding is Supported on the:

Cisco 10000 and Cisco 12000 12.0(22)S

Cisco 7500 in 12.0(22)S, with the caveat that inserting a newstandby RSP will cause some traffic loss, and switching from theprimary to standby RSP will cause a microcode reload on the linecards

http://www.cisco.com/en/US/partner/tech/tk826/tk364/technologies_white_paper09186a008016317c.shtml

2 January 2003


RST-4300

9859_05_2004_X


Be careful with partialdeployments of GR/NSFcapability

If B restarts, A will reset itssession, removing all therouting information it learnedfrom B. However, D willcontinue to forward trafficthrough B

This will, at best, causeasymmetric routing. At worst, it

could cause a routing loop A must be GR capable

network core

GR/NSF capable

A

B C

D

session reset

D continues

forwarding

asymmetricreturn path


30/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

service provider

A

B C

D

eBGP

eBGP

iBGP

iBGP

EIGRP


EIGRP is configured forGR/NSF, while BGP is not

Ds next hop for all routes isA; the path to A is learnedvia EIGRP

If the control plane on Brestarts, D will continuelearning BGP routes with anext hop of A through C; itwill also maintain the bestpath to that next hopthrough B

best pathto A BGP learnedroutes

OTHER CURRENTENHANCEMENTS


RST-2310

9636_05_2004_c1


31/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Other Current Enhancements

EIGRP PE/CE Deployment

EIGRP PE/CE BackdoorLinks

EIGRP Third Party Next Hop


RST-4300

9859_05_2004_X


In this network, we have twocorporate sites, connectedby a leased line and VPNthrough a service provider

EIGRP routes redistributedinto BGP at B, and back intoEIGRP at C, appear asexternal routes at Site 2

We want them to appear as

internal routes

Service Provider

Site 1

Site 2

A

B

C

D

VPN

external


32/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


As routes are redistributed intoBGP as B, extendedcommunities containing theEIGRP metrics are attached tothem

As routes are redistributedback into EIGRP at C, theseextended communities are usedto reconstruct the routes asinternals

The VPN is considered a 0 costlink in this configuration

Service Provider

Site 1

Site 2

A

B

C

D

VPN

internal


RST-4300

9859_05_2004_X


Service Provider

Site 1

Site 2

A

B

C

D

VPN

internal

ip vrf VRF-RED

rd 172.16.0.1:20

exit

....

router eigrp 1

address-family ipv4 vrf VRF-RED

autonomous-system 101

network 172.16.0.0 255.255.0.0

redistribute BGP 101 metric 10000 100 255 1 1500

exit-address-family

router-c#show ip eigrp vrf VRF-RED topology

IP-EIGRP Topology Table for AS(1)/ID(192.168.10.1)

Routing Table:VRF-PINK


via 50.10.10.2 (409600/128256), Ethernet3/0



33/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

no backdoor link

Service Provider

Site 1

Site 2

A

B

C

D

VPN


12.0(27)SV 12.0(21.1)SY212.0(21.1)S2

Backdoor links are notsupported

http://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a0080154db3.html


RST-4300

9859_05_2004_X

Service Provider

Site 1

Site 2

VPN

EIGRP PE/CE Backdoor Links

The biggest danger withbackdoor links is possiblerouting loops

Site1 advertises a networkthrough the back door to site 2

C prefers this route, andredistributes it into BGP

B prefers the BGP route, andredistributes it into EIGRP,forming a loop

The solution is to automatically

tag all the routes originating insite 1 so they will be rejected byC

This tag is called the Site ofOrigin (SoO)

A

B

C

D


34/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Service Provider

Site 1

Site 2

A

B

C

D

VPN


The SoO is set on all PErouters on the interfaceconnecting to the PE, and onbackdoor link routers

The CE will always reject themarked EIGRP learned routes,and prefer the BGP learnedroutes.

You can then set the backdoorlink so the path through theVPN is always preferred overthe backdoor link.

route-map SoOrigin permit 10

set extcommunity soo 100:1

....

interface FastEthernet 0/0

ip vrf sitemap SoOrigin

....


RST-4300

9859_05_2004_X


12.0(27)SV 12.0(26)SZ 12.0(26.1)S

http://www.cisco.com/en/US/products/sw/iosswrel/ps1829/products_feature_guide09186a00801eff60.html


35/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


C advertises 10.1.1.0/24 to A,which it readvertises it to B,and sets the next hop to0.0.0.0

When B receives this route,it sets the next hop to10.1.3.1

If B is receiving packets for10.1.1.1, it will need toforward them through A,even though it has a directconnection to C

EIGRPneighbors

EIGRPneighbors

no EIGRPrunning

A

B

C

10.1.1.0/24

10.1.3.2

10.1.3.1

IP-EIGRP Topology Table for AS(100)

....

P 10.1.1.0/24, 1 successors ....

via 10.1.3.1 ....


RST-4300

9859_05_2004_X


EIGRP third party next hopallows A to leave the nexthop at 10.1.3.2

B can then use the directlink between B and C toforward traffic to 10.1.1.0/24,even though EIGRP isntrunning between the tworouters

EIGRP


....

P 10.1.1.0/24, 1 successors ....

via 10.1.3.2 ....

interface Serial 0

no ip next-hop-self eigrp 100

....

EIGRP

no EIGRPrunning

A

B

C

10.1.1.0/24

10.1.3.2

10.1.3.1


36/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


Here A, B, and C are allconnected to the samebroadcast segment, 10.1.1.0/24

A is redistributing Rip intoEIGRP

B isnt running RIP

C isnt running EIGRP

B would normally show A asthe next hop, rather than C,although it can reach C directly

With eigrp no next hop self onthe Ethernet, A will send itsupdates to B with C as the next

hop

A

B C

10.1.1.0/24


....

P 10.1.1.0/24, 1 successors

via 10.1.2.1 ....

router eigrp 100

redistribute rip metric ...

....

interface Ethernet 0

no ip next-hop-self eigrp

EIGRP

.1.2

.3

RIP


RST-4300

9859_05_2004_X


Applications for third party next hop:

Dynamic Multipoint Virtual Private Networkshttp://www.cisco.com/en/US/products/sw/iosswrel/ps1839/products_feature_guide09186a0080110ba1.html#1039490

Preserving the next hop in redistribution from broadcastnetworks

http://www.cisco.com/cgi-bin/Support/Bugtool/onebug.pl?bugid=CSCdk23784


37/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

EIGRP Enhanced Route Map Support

CSCdw22585 providesenhanced support of routemaps for EIGRP

This allows setting themetric using a route map toprefer one path overanother, for instance

10.1.1.0/24

route-map setmetric permit 10

match interface serial 0/0

set metric 1000 1 255 1 1500


match interface serial 0/1set metric 2000 1 255 1 1500


....

router eigrp 100

distribute-list route-map setmetric in

s0/0 s0/1

A

B C

D


RST-4300

9859_05_2004_X


matches against tags on internalroutes

match tag 100

matches against tags on externalroutes

match tag external 100

matches against the external metric ofan external route

match metric external 1000

matches routes with metrics from 900to 1100

match metric 1000 deviation 100

matches routes sourced from BGPautonomous system 65500

match route-type external route-typebgp 65000

matches routes sourced from BGPautonomous system 65500

match route-type external route-typebgp 65000


38/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X


matches against the next hop listed inthe route

match ip next-hop 10.1.1.1

matches against the interface theroute was learned through

match interface serial 0/0

sets the component metrics for aroute

set metric 1000 1 255 1 1500

sets the next hop listed in the routeset ip next-hop 10.1.1.1

sets the tag on internal routesset tag 100

sets the tag on external routesset tag external 100

OTHER FUTUREENHANCEMENTS


RST-2310

9636_05_2004_c1


39/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Other Future Enhancements

EIGRP IPv6

MIB Support

Loadable EIGRP

Dynamic Metrics

Bundled Metrics

Summary Only

Leak Through a Summary

Default InformationOriginate


RST-4300

9859_05_2004_X

EIGRP IPv6

Future Enhancement

Initially, EIGRP IPv6 will

Run EIGRP over an IPv6multicast address

Add additional TLVs to theEIGRP packets to carry IPv6addresses

Interface based configuration

In the future, all configurationwill be moved under the EIGRP

processInterfaces grouped by range,rather than network statements

IPv4 configuration will followthe IPv6 configuration style

router#conf t

Enter configuration commands, one perline. End with CNTL/Z.

router(config)#ipv6 unicast

router(config)#int fastEthernet 0/0

router(config-if)#ipv6 eigrp 1

router(config-if)#exit

Per-interface configuration

router(config)#!

router(config)#ipv6 router eigrp 1

router(config-rtr)#router-id 1.1.1.1router(config-rtr)#no shutdown

router(config-rtr)#end

router#

Assign router-id and no shutdown


40/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

MIB Support

Future Enhancement

EIGRP will support 68 MIB objects in 4 major tables

EIGRP traffic statistics

EIGRP topology data

EIGRP neighbor data

EIGRP interface data

A fifth table, the EIGRP VPN Table, is included for indexing

eigrpRouteSIA and eigrpAuthFailure can trigger SNMP traps


RST-4300

9859_05_2004_X

MIB Support

EIGRP Traffic Statistics

AS Number

Hellos Sent/Received

Updates Sent/Received

Queries Sent/Received

Replies Sent/Received

EIGRP Topology Data

Destination Net/Mask

Active State

Feasible Successors

Origin Type

Distance

Reported Distance

EIGRP Interface Data

Peer Count

Reliable/Unreliable Queues

Pacing

Pending Routes

Hello Interval

EIGRP Neighbor Data

Peer Address

Peer Interface

Hold Time

Up Time

SRTT/RTO

Version

And many more....


41/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Loadable EIGRP

Future Enhancement

Cisco IOS Software is currently monotlithic

To get a new version of any specific module, you need a newversion of IOS

When you get a new version of IOS, you get a new version ofeverything, not just the module you wanted

Development is planned to break EIGRP off as a loadable unit

This means you could load a new version of EIGRP into anexisting running instance of IOS

New features and bug fixes could be targeted

You load only EIGRP


RST-4300

9859_05_2004_X

Dynamic Metrics

Future Enhancement

EIGRP uses manually configured bandwidth anddelay metrics

EIGRP only reads the load and reliability metricsoff an interface

When the neighbor relationship comes up

When the bandwidth or delay are manually changed

Why cant we make these dynamic?


42/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Dynamic Metrics

Future Enhancement

The problem is EIGRP caneasily get into positivefeedback loops if metrics areread dynamically from theinterface

As traffic is shifted to10.1.1.0/24, the load increases

This drives traffic back to10.1.2.0/24, increasing its load,decreasing 10.1.2.0/24s load

This, in turn, drives traffic backto 10.1.1.0/24, increasing itsload, and decreasing

10.1.2.0/24s load

Constant Churn

10.1.1.0/24

10.1.2.0/24

load 10 load 20

shift traffic here

load 20 load 10

shift traffic here

load 10 load 20

shift traffic here


RST-4300

9859_05_2004_X

Dynamic Metrics

Future Enhancement

What we need is a way tobalance the traffic properly

We dont want any churn

Cisco has a patentedmechanism to allow the shiftingof load without the churn

This capability will be includedin a future feature allowing

dynamic metrics to be used inEIGRP

10.1.1.0/24

10.1.2.0/24

load 10 load 20

shift traffic here

load 20 load 10

shift traffic here

load 10 load 20

shift traffic here


43/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Bundled Metrics

Future Enhancement

If there are multiple links usedfor load sharing, only a singlelinks bandwidth and delay willbe advertised to neighbors

In this network, B and Cadvertise the same metrics to10.1.1.0/24

C actually has more bandwidthavailable to reach 10.1.1.0/24

EIGRP will be able to bundle themetrics of the equal cost linksbetween C and D

A will get a more true picture ofthe paths available to10.1.1.0/24

BW:1000D:500

BW:2000D:???

BW:1000D:500

BW:1000D:500

BW:1000D:500

BW:1000D:500

10.1.1.0/24

A

B C

D


RST-4300

9859_05_2004_X

Summary Only

Future Enhancement

A would like to advertise justthe 10.1.0.0/16 to B and C

B and C dont need reachabilityto 10.2.1.0/24

To do this, build

A summary for 10.1.0.0/16

A distribute list blockingeverything except the summarytowards B and C

Create a new keyword on thesummary statement

Just the summary would beadvertised, and no other routes

Eliminates the distribute list

A

B C

10.1.0.0/2410.1.1.0/2410.1.2.0/2410.2.1.0/24

interface serial 0/0ip summary-address eigrp 100 10.1.0.0 255.255.0.0!access-list 10 permit 10.1.0.0 0.0.255.255!router eigrp 100distribute-list 10 out

interface serial 0/0ip summary-address .... summary only


44/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Leak Through a Summary

Future Enhancement

A wants to advertise thesummary 10.1.0.0/16 and10.1.1.0/24

Use a pair of overlappingsummaries, and play with theadministrative distance

Use a static to null 0 for10.1.0.0/16, rather than asummary, and then build thecorrect distribution list

Theres no real easy way to dothis today

Create a new command to allowa set of routes matching a routemap to leak through a summary(or stub, etc.)

A

B C

10.1.0.0/2410.1.1.0/2410.1.2.0/24

10.2.1.0/24

overlapping summarieswith administrativedistances

redistributed static routewith a distribute list

new command to allow acomponent to leak past asummary


RST-4300

9859_05_2004_X

Default Information Originate

Future Enhancement

The only options for originating a default route into EIGRPtoday are:

A redistributed static route, which produces an external

A summary, which produces an internal, but isnt conditional

A new feature will be added to create the command default-information originateunder router eigrp

A route map will be able to control when the default route isgenerated or not

The default route generated will be an EIGRP internal


45/46


9859_05_2004_X.scr


RST-4300

9859_05_2004_X

Recommended Reading

Continue yourNetworkers learningexperience with furtherreading for this sessionfrom Cisco Press.

Check theRecommendedReading flyer forsuggested books.

Available on-site at the Cisco Company Store


RST-4300

9859_05_2004_X

Complete Your Online Session Evaluation!

WHAT: Complete an online session evaluationand your name will be entered into adaily drawing

WHY: Win fabulous prizes! Give us your feedback!

WHERE: Go to the Internet stations locatedthroughout the Convention Center

HOW: Winners will be posted on the onsiteNetworkers Website; four winners per day


46/46


RST-4300

9859_05_2004_X

advances in eigrp

Documents