RE: BGP MED alternative (with sample output)

From: Khalid Siddiq (khalid@xxxxxxxxxx)
Date: Wed Jul 31 2002 - 06:17:52 GMT-3


   
But the restriction is that we have not to change the configuration of router R
1.
I think we can solve this by using a route-map at R2 with R1, in which we set t
he origion attribute of all the route send to R1 as IGP. also i am supposing th
at the routes learn via R4 has origion attribute set to incomplate.
this make R1 to prefer the lowest origion attribute path.
regards,
khalid

-----Original Message-----
From: ying c [mailto:bf5tgh1@yahoo.com]
Sent: Wednesday, July 31, 2002 5:43 AM
To: ccielab@groupstudy.com
Subject: RE: BGP MED alternative (with sample output)

Hi,
Below is some output for those of you might be interested. In order for Sandra'
s solution to work, we have to modify R1's bgp to include "bgp bestpath compare
d-routerid" (have to bend the rule a little bit), otherwise there's no guarante
e the lowest RID will win. Brian's solution works under the assumption we are n
ot using network command to advertise r3-r5:
Before change ORIGIN:
* 172.16.10.0/24 172.16.35.3 0 0 200 ?
*> 140.10.1.1 0 0 200 ?
-------------------------------------------------------------------
After change ORIGIN:
*> 172.16.10.0/24 172.16.35.3 0 0 200 i <<< IGP wins
* 140.10.1.1 0 0 200 ?
===================================================================
Before change RID:
*> 172.16.10.0/24 140.10.1.1 0 0 200 ?
* 172.16.35.3 0 0 200 ?
r1#sib 172.16.10.0
BGP routing table entry for 172.16.10.0/24, version 4
Paths: (2 available, best #1, table Default-IP-Routing-Table)
  Advertised to non peer-group peers:
  172.16.35.3
  200
    140.10.1.1 from 140.10.1.1 (172.16.102.2) <<<< Lower RID wins
      Origin incomplete, metric 0, localpref 100, valid, external, best
  200
    172.16.35.3 from 172.16.35.3 (172.16.103.3)
      Origin incomplete, metric 0, localpref 100, valid, external
-------------------------------------------------------------------
After change RID on both OSPF & BGP from 172.16.103.3 to 172.16.100.3:
* 172.16.10.0/24 172.16.35.3 0 0 200 ?
*> 140.10.1.1 0 0 200 ?
r1#sib 172.16.10.0
BGP routing table entry for 172.16.10.0/24, version 4
Paths: (2 available, best #2, table Default-IP-Routing-Table)
  Advertised to non peer-group peers:
  172.16.35.3
  200
    172.16.35.3 from 172.16.35.3 (172.16.100.3) <<<< Older wins
      Origin incomplete, metric 0, localpref 100, valid, external
  200
    140.10.1.1 from 140.10.1.1 (172.16.102.2) <<<< Lower RID did not win
      Origin incomplete, metric 0, localpref 100, valid, external, best
======================================================================
Add bgp bestpath compare-routerid on r1:
*> 172.16.10.0/24 172.16.35.3 0 0 200 ?
* 140.10.1.1 0 0 200 ?
r1#sib 172.16.10.0
BGP routing table entry for 172.16.10.0/24, version 4
Paths: (2 available, best #1, table Default-IP-Routing-Table)
  Advertised to non peer-group peers:
  140.10.1.1
  200
    172.16.35.3 from 172.16.35.3 (172.16.100.3) <<<< Lower RID wins
      Origin incomplete, metric 0, localpref 100, valid, external, best
  200
    140.10.1.1 from 140.10.1.1 (172.16.102.2)
      Origin incomplete, metric 0, localpref 100, valid, external

  ying c <bf5tgh1@yahoo.com> wrote: Sandra and Brian have the right answers. lo
cal_pref is
not the answer because that's used to influence path
selection going into AS100 from AS200.

Thanks,
Chang
--- Daniel Garrity wrote:
> Couldn't you just set the local_pref higher for that
> route on R2?
>
> Something like setup a Route-map on R2 matching the
> network in question
> and setting the local_pref to 300 or something?
>
> Dan
>
>
>
> -----Original Message-----
> From: ying c [mailto:bf5tgh1@yahoo.com]
> Sent: Tuesday, July 30, 2002 11:00 AM
> To: ccielab@groupstudy.com
> Subject: BGP MED alternative
>
>
> Hi,
>
> Here's one of the question that I wonder one of you
> smart guys may have an answer:
>
> R1------R2-------R3
> \ | |
> \-----R4-------R5
>
> R1 is in AS100, R2 and R4 are in AS200, R1, R2, R4
> are
> running BGP. R2, R4, R3 and R4 are running OSPF. R2
> and R4 both peer with R1 and send full ospf specific
> routes to R1.
>
> The requirement is to have R1 always use R1-R2 link
> as
> the primary route to reach the network between R3
> and
> R5. Normally, we would use MED, AS-PATH or aggregate
> routes to achieve this. However, here's the tricky
> part: you are not allowed to use MEDs and not to
> change the configuration on R1 or R4 to accomplish
> this.
>
> I did it by using a lower metric number when I
> redistribute ospf into bgp in R3, but I think this
> pretty much violated "not to use MEDs" rule. Is
> there
> any other way to solve it?
>
> Thanks,
> Chang
>



This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:36:49 GMT-3