EIGRP(CCNP)

Página 1

Basic:

- Multicast IP:224.0.0.10

- Use own IP protocol =88

-

NBRSHIP:

- we can configure nbr's statically. multicast ability of the interface will go away

Nbr'ship requirement:

- interface primary IP must be in same subnet

-AS number should be same

-connected interface must not be passive

-authentication must pass

- K-values must match:

R1(config)#router eigrp 123

R1(config-router)#metric weights 0 1 0 1 0 1

HELLO packets:

- on LAN, send every 5 seconds

- on slow links like T1, 60 seconds by default

- hold/dead interval = 3xhello interval

- Change hello interval:

R1(config-if)#ip hello-interval eigrp

-Change hold interval:

R1(config-if)#ip hold-time eigrp

-the dead interval tells neighbor device when to consider this device dead

- Only hold interval is send inside hello packet to neighbors

-

NON STOP FORWARDING

-

EIGRP Authentication

-key chain funda:

Step 1) create key chain:

!

key chain MK

key 0

key-string KL

key 1

key-string NB

!

Step 2) Enable MD5 authentication on interface

R1(config-if)#ip authentication mode eigrp 123 md5

Step 3) apply key chain to interface

R1(config-if)#ip authentication key-chain eigrp 123 MANOJ

PASSIVE INTERFACE

- Hello won't be send on passive interface. received hello will be dropped.

- Network command enables EIGRP hellos on interfaces which are included in network statement.

- Redistribute connected won't enable hello on interfaces

two ways to stop hello packets while still advertising the networks to eigrp:

1) passive interface

2) redistribute connected

- we can use passive interface default command too

R3#sh run | sec router eigrp

router eigrp 123

passive-interface default

no passive-interface FastEthernet0/0

network 0.0.0.0

no auto-summary

-

STATIC NBr and router id:

- configuring static nbr on an interface will remove EIGRP multicast abilities on that interface.

- even if we configure static route, we need network command to send/receive hellos. nbrship won't come up without network command.

router eigrp 123

network 10.0.123.0 0.0.0.255

no auto-summary

neighbor 10.0.123.1 FastEthernet0/0

-check:

show ip eigrp neighbor detail

-

Router-id:

- even if router-id is same,nbrship will come up between directly connected routers.

- in case of internal eigrp routes, routes will be exchanged even if router-id is same.

- in case of external eigrp routes (via redistrbution command), routes wont be exchanged btw directly connected routers and indirectly connected routers.

EIGRP over WAN

EIGRP topology Table

-Once EIGRP ngrship is up:

a) router will put its subnets in its topology table

using:

i) network command

ii) redistribute command

b) EIGRP ngrs will exchange topology tables.

-

EIGRP Packets

-EIGRP uses RTP to send updates and ack

1) Hello

2) Update: send multicast

3) ack: send unicast

5) query:

6) reply":

- EIGRP doesn't re-send topology table at all once topology table has been exchanged. only in case any change in topology, then send update packet.

- Summarization kills ngrship and bring back the ngrship

- update are send to 224.0.0.10 multicast IP. receiving router will send ack using unicast. if no ack, router will keep sending update for 16 times. if still no reply,kill that relationship even if hello are being received.

-

EIGRP and Split horizon

- NBMA networks will have issue with forming EIGRP nbrship

-we can disable split horizon on an interface:

#no ip split-horizon eigrp

- by default, EIGRP msgs will use only 50% of the bandwidth of the interface

- change bandwidth using ip bandwidth-percent eigrp

-

Feasible distance and advertised distance

- Advertised distance= distance advertised by ngr for a given prefix. keep in mind that ngr won't give distance but parameters like bandwidth,delay,load,reliability ,mt. local router will calculate distance using parameters given by the nbr.

- Feasible distance: take lowest bandwidth and sum of delay. calculate total distance

Metric Manipulations

1) change bandwidth and delay:

-keep in mind that bandwidth is in kilobits/second and delay is in 10 raise to power -5. if needs to change, change delay rather than bandwidth. bandwidth is used by QoS,OSPF.. so dont change bandwidth.

-changing bandwidth/delay will only change FD and not RD/AD

2) change K-values: change K-value to 2,3,4...

3) offset list:

-New distance = distance calculated by the router + offset value in offset list given by the operator

-offset list works only with standard access list

- nbr relationship will go down and come back

-offset list basically changes the delay value

configuration:

step 1) create standard access list:

R4(config)#access-list 99 permit 1.1.1.0 0.0.0.255

step 2) apply access list in offset list in eigrp process:

R4(config-router)#offset-list 99 in

- this will affect all downstream nbrs

step 3) option: can apply offset list per interface

R4(config-router)#offset-list 99 in 2000 fa0/1

- if offset list is applied inbound to an interface connected to upstream provider, only feasible distance will change

-if offset list is applied inbound, it will change FD for current router and downstream routers

- if offset list is applied outbound, it will change AD of current router to downstream routers

Convergence

- successor will be added to topology table, if no feasible successor.

- in case of feasible successor, it will be added to routing table after checking

a) the Administrative distance : if EIGRP AD 170, route will be added in topology table but not routing table

b) feasibility condition : RD < FD

- if feasibility condition is not matched, we can still see the route using below command:

R4#show ip eigrp topology all-links

Convergence by going ACTIVE

- if no feasible successor, then router will send query msg to its neigbor for missing route. router change to active.

- neighbor send query to all neighbors and wait for reply from all ngrbs.

-

Stuck in Active

- Active timer: when I send a query to my nbr, I'll wait for 3 minutes to wait for reply.

- If i don't get reply in 3 minutes, I'll stuck in active

- at 90 sec ( half of active timer), I'll send SIA-query to my neighbor. if issue with the nbr,i'll wait for 180 sec and bring down the nbrship. if issue with the ngbr's ngr - my ngr will send SIA-reply - my nbr will timeout his ngb

change active timer:

#router eigrp 123

# timers active time <>

Limiting Query Scope

Option 1) stub router

- stub router: stub router will tell nbr that its a stub router using hello packet. Nbgrs will not send queries to stub router.

- A stub router doesn't advertise EIGRP learned routes from one nbr to another nbr.

#router eigrp <>

# eigrp stub

- can be used to advertise static, redistrbuted routes using

#eigrp stub

Option 2) Summarization

-

EIGRP VARIANCE

- Cisco CEF does per flow load sharing as default

- EIGRP: we can do unequal cost load balancing

- Variance adds Feasible successor routes to routing table

- be default set to 1 - equal cost load balancing

- for unequal cost load balancing - variance > 1

Steps:

1) Take FD.

2) multiply FD with variance

3) for all feasible successor routes, if FD*variance > feasible successor - add the Feasible successor to routing table

4) any route which don't comply with Feasibility condition will not be added to routing table

EIGRP Route filtering: to filter inbound/outbound EIGRP updates

Use Distribute list:

Format:

distribute-list {ACL | prefix-list | route-map} {in|out} {interface}

1) Distribute list with access list:

-if distribute list is applied as "in", all incoming EIGRP updates from

ngbrs will be filtered before putting into topology table

-if distribute list is applied as "out", all outgoing EIGRP updates will be filtered before sending to ngrsa) standard access list:

standard access list:

step 1) create access:list

access-list 99 deny 5.5.5.0 0.0.0.255

step 2) apply distribute list:

distribute-list 99 out

verify:

R1#show ip protocols

Routing Protocol is "eigrp 123"

Outgoing update filter list for all interfaces is 99

Incoming update filter list for all interfaces is not set

extended access-list:

- in case of extended access-list, source will act as EIGRP neighbor while destination will act as the prefix which needs to be denied/permitted:

step 1) create extended access:list

access-list 100 deny ip host 10.0.15.5 5.5.5.0 0.0.0.255 ( 10.0.15.5 is nbr IP)

step 2) apply

distribute-list 100 out

2) Distribute list with prefix list:

- prefix list are usually used with routing protocols

-Format:

ip prefix-list list-name {seq seq-number}{deny|permit}{prefix/prefix-length} {ge ge-value}{le le-value}

- if ge/le is not used, subnet mast is prefix-length

- if ge/le is used,prefix-length is used to match the prefix and subnet mask will be between ge and le

-

3) Distribute list with route map:

EIGRP ROUTE SUMMARIZATION:

- EIGRP summarisation can be done at any router

- with summarisation query scope reduces

- router adds summary route with null0

-Manual summarisation

(config-if) # ip summary-address eigrp asn

auto-summary:

- only network orginating from the router will be summarized

- only prefix with network command will be summarized

- with eigrp auto-summary, eigrp works as classful routing protocol

EIGRP Default-route

method 1: use default route and advertise to EIGRP using network 0.0.0.0 command or redistribute static command

method 2: use default network command. A classful network needs to be advertised to EIGRP.

step 1) create a classful network

step 2)

use command:

ip default-network

step 3) advertise classful network to eigrp process

method 3: using summary address

(config-if)# ip summary-address eigrp 0.0.0.0 0.0.0.0

- only default route will be advertised

NAMED mode EIGRP

- no need to go to interface for eigrp config. but IPv6 - still needs to go interface

- version supports : 15.0.1(M), 12.2.32SRE, 12.3.32XNE

#router eigrp CISCO

# address-family ipv4 unicast automous-system

-By default, EIGRP stub routers advertise information about two types of

routes back to the hub - directly connected networks and summary routes.