BGP Assistance / Verification

Unanswered Question
Oct 9th, 2008

I am looking for some assistance and or verification with eBGP. Any assistance is greatly appreciated.

Example BGP topology on a private network:

Router A is the primary site. AS 1.

Router B is the hot stand by site. AS 2.

Router C is at a remote site. AS 3.

Router A & B advertises the same networks to router C via BGP.

Router C advertises various networks back to routers A & B.

Router A & B redistributes BGP into OSPF.

I need to control path selection.

Circuit between router A & B is the primary path and must be used. Circuit between B & C is the secondary circuit for failover.

Since this is purely eBGP, I believe local pref is not an option. I am also not able to use weights as router C at the remote end is not Cisco.

Any Suggestions? Your assistance is most appreciated.

Cheers,

Chris

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Richard Burts Thu, 10/09/2008 - 16:36

Chris

I am not clear when you say that A & B redistirbute BGP into OSPF whether this is into the same OSPF or into different OSPFs. Describing A as the primary and B as the hot standby sounds like it probably is the same OSPF. In that case I do not understand the logic of A being in a different AS than B. If they share a common Interior Routing Protocol how can they be different External Systems?

And I find this description quite confusing:

Circuit between router A & B is the primary path and must be used. Circuit between B & C is the secondary circuit for failover.

It really makes it sound like A and B and C are all part of the same AS.

But beyond those issues I am not sure why you think that local pref would not work. If C receives the same routes from A and B and if C sets local pref to favor A, then C would send its traffic through A and use B as a backup. Is that not what you said you wanted?

HTH

Rick

christopher.clayden Thu, 10/09/2008 - 18:00

Hi Rick,

Thank you for your response, I sincerely appreciate it.

Yes you are correct; BGP is being redistributed into the same OSPF area.

Site "C", which is the remote site; there are hundreds of these sites. Router A at the prime site and Router B at the Hot Standby Site each have a point to point frame PVC mapped to each remote site.

For each remote site, site "C" in this example, each have a different AS.

As this is a Legacy config, I also do not understand why they used different AS's for the prime site (Site A) and for the hot stand by site (Site B). They should both be in the local AS, but I am not permitted to change this.

I need the capability to control path selection at Router A - the primary site.

Perhaps I could set local pref in the out bound direction at Router A?

Cheers,

Chris

PS: What is HTH?

Richard Burts Fri, 10/10/2008 - 13:45

Chris

As long as someone insists that you must keep A and B in separate ASes and each remote in its own separate AS then I am not sure that there is a good way to have A do route selection. Setting local pref on A is only effective if it is advertised to an IBGP peer. Local pref is not propagated to external peers.

HTH (Hope This Helps)

Rick

Giuseppe Larosa Fri, 10/10/2008 - 13:46

Hello Chris,

HTH = Hope to help (I think)

if RA, RB and RCi are all in different BGP ASes you cannot use local-preference that is used within a single AS domain.

To influence choices you can:

use AS path prepending on RB to influence Rci to prefer the shorter AS path from RB.

When redistributing BGP (all routes learned from RCi routers) into OSPF:

use a route-map that will

set metric-type to O E1

set metric 50 (RA) / 1000 (RB)

this will allow all internal routers to choice RA if it provides a remote branch site route.

on RA makes RC preferred using weight over RB.

if you want to be sure that also RB will route via RA :

have RB prefer RA over each RCi using weight

(not recommended for faster convergence see later)

Rci will prefer RA

RA is preferred by all internal routers and even by RB

if RA to RCi BGP session goes down:

OSPF route generated by RA will disappear

RB will install RCi routes and will redistribute them into OSPF.

For convergence reasons I would keep RB to prefer RCi over RA: doing this with an higher seed metric in OSPF as explained above

note:

if you also redistribute OSPF into BGP you will need to use route filters because you have multiple redistribution points on RA and RB.

Edit:

you will not need to set weight AS path via RCi is shorter then going via RB and then RCi (all eBGP sessions)

Hope to help

Giuseppe

Actions

This Discussion