BGP Questions

[carl_townshend]

Hi All

Couple of BGP questions,

Firstly, If I have 2 links via IBGP to another router, what do most people do to make it prefer one path over the other, weight is only local, so do people use local pref in and out the router so the traffic is preferred both ways over the link? can you do the same with MED?

What if this was EBGP, would you use local pref for outbound and MED or AS path prepend for inbound?

Also I have been playing with GNS3 to try to load share traffic between 2 links on 2 ibgp neighbours.

There are 2 interfaces on each router so 2 neighbours from the routers perspective.

For some reason it doesn't seem to load share, do you need to use a loopback so it looks like one neighbour to do this?

Cheers

[Giuseppe Larosa]

Hello Carl,

in iBGP local preference or weight can be used.

as you noted for two eBGP sessions you could use MED and local preference in opposite directions

>>

Also I have been playing with GNS3 to try to load share traffic between 2 links on 2 ibgp neighbours.

There are 2 interfaces on each router so 2 neighbours from the routers perspective.

For some reason it doesn't seem to load share, do you need to use a loopback so it looks like one neighbour to do this?

BGP defaults to select one single path so either you

configure

maximum-paths ibgp  2

under router bgp

or you use  a single iBGP session terminated on loopbacks + appropriate static routes or IGP  +  and update-source neighbor commands

use of loopbacks for iBGP sessions is very common and handy for mulltiple reasons.

Hope to help

Giuseppe

[carl_townshend]

Hi

I tried the maximum-paths 2 when peering using 2 interfaces and it didn't appear to work on gns3, it just kept picking 1 interface as the best path.

Also, whilst you are there , I have been looking at vrf's in BGP also, I have noticed they are not working, it only lets me use the address-family ipv4 vrf xxxx command, it doesnt appear to advertise any routes etc, does it need to be the address-family vpnv4 vrf xxx command? if so why does the former one let you put in vrf's etc?

cheers

[MHM Cisco World]

Hi friend 
R1 with AS 100 want to iBGP with R4 and R5 (same AS100), both R4 and R5 have connect to same subnet for example 10.10.10.0/24
1- we will config R1 with loopback 
2- config the R1-R2,R2-R4,R1-R3,R3-R5 & loopback in R1 with EIGRP or OSPF (I prefer OSPF).`

NOTES:- 10.10.10.0/24 not include in OSPF config
3- config BGP neighbor from loopback in R1 to both R4 and R5.
4-config BGP neighbor from R4 to R1 loopback, and config network 10.10.10.0 mask 255.255.255.0
5-config BGP neighbor from R5 to R1 loopback, and config network 10.10.10.0 mask 255.255.255.0
6-NOW R1 have two path for 10.10.10.0/24, we need to override select one by config max-path 2

here the trick 
when the R1 send packet to 10.10.10.0/24 and select path through the R4, it will forward to R2!! R2 will receive packet without know how to forward this destination IP.
in order to Solve this we will config  10.10.10.0/24 with OSPF 
BUT
here R1 will have two route ONE from BGP other from OSPF ?
OSPF is better than BGP because it have AD=200 
so R1 will use OSPF not BGP.
Hmmm
how about load sharing ??
load sharing is happened in OSPF not BGP
R1 now have two OSPF path for same route 10.10.10.0/24 
I hope you get idea about load sharing in the IBGP.

CASE R2 and R3 is not present
same as above but here the load sharing is BGP not OSPF.