[nsp] BGP question

[Jeff Chan]

Yes, that's right.  MED is checked after path length so it affects
equal path length routes.

Jeff C.
__

On Friday, September 12, 2003, 9:37:47 AM, Stephen Wilcox wrote:
> Ok so yuo need to change something thats checked after the as path length. Your 
> best bet is the metric (MED):

> route-map prov1-feed-in permit 10
>  set metric 0
> route-map prov2-feed-in permit 10
>  set metric 50


> lowest med wins so prov1 will get the traffic all other things (inc as-path 
> length being equal)

> Steve

> On Fri, 12 Sep 2003, Krzysztof Adamski wrote:

>> That is not what Kevin asked (I have the same question).
>> He wants to only influance the routes with the same AS path length to
>> always go to provider1, but in all other cases the AS path length wins
>> with default behavior.
>> 
>> K
>> 
>>  On Fri, 12 Sep 2003, Mark Tinka wrote:
>> 
>> > Sorin CONSTANTINESCU wrote:
>> > > Hi,
>> > >
>> > > Kevin wanted his egress traffic (upload) to go through preferred
>> > > upstream, not his ingress (download) traffic.
>> >
>> > Well then, in that case, even easier; simply modify the LOCAL_PREF attribute
>> > of the routes learned from ISP 2, and set that value lower.
>> >
>> > Cisco defaults to 100, so you can set local preference for routes learned
>> > from ISP 2 [either default information, or a full BGP feed] to something
>> > like 80. The routes with a higher local preference, win, and get installed
>> > from the BGP Loc-RIB into the router's active routing table.
>> >
>> > You can do that with:
>> >
>> > route-map bgp-feed-in permit 10
>> >  set local-preference 80
>> >
>> > Then apply this to the inbound annoucements coming from ISP 2, the less
>> > prefered circuit. I'd recommend using the Route Refresh capability to effect
>> > the new policies :-).
>> >
>> > >
>> > > Mark Tinka said:
>> > >> cisco-nsp-bounces at puck.nether.net wrote:
>> > >>> I'm still a newbie at bgp, so please bear with me.
>> > >>> I have two links to the internet.
>> > >>> A lot of routes from both providers have equal AS paths.
>> > >>> If one bgp session gets reset all that traffic (the equal AS path
>> > >>> routes) goes out to provider2 and does'nt reset back to provider1,
>> > >>> providing that the bgp session to provider1 was the first one to
>> > >>> come up. I don't have any metrics, weights or localpref on any
>> > >>> routes. Is there any way to have those equal path routes default to
>> > >>> provider1 rather than provider2 if provider1's link and bgp session
>> > >>> are up? Right now in order to reset those routes back to provider1
>> > >>> I have to do a hard "clear ip bgp provider2" reset.
>> > >>>
>> > >>> Kevin,
>> > >>> Honeycomb Internet services
>> > >>
>> > >> Hi Kevin.
>> > >>
>> > >> You can use communities, with your ISP, to influence the path
>> > >> selection process. But even easier, you can make the AS path over the
>> > >> less preferred ISP, longer, by prepending it along with your BGP
>> > >> announcements to their BGP speaker.
>> > >>
>> > >> Basically, the path with the shortest AS path, wins. So, by making
>> > >> the AS path over ISP 2 longer, you automatically ensure it's only
>> > >> used for return traffic when ISP 1 is totally unavailable. You can
>> > >> do this by:
>> > >>
>> > >> route-map PREPEND-ISP-2 permit 10
>> > >>  set as-path prepend xxx xxx xxx xxx
>> > >>
>> > >> Where xxx = your own ASN. The more times you prepend your ASN, the
>> > >> longer the path over ISP 2 appears. IIRC, Cisco route maps will
>> > >> support a maximum of 10 prepends + 1 which is sent along by default.
>> > >>
>> > >> You will then need to apply this route map to your outbound policies
>> > >> to ISP 2, and you're in business.
>> > >>
>> > >> Regards,
>> > >>
>> > >> Mark Tinka - CCNP
>> > >> Network Engineer, Africa Online Uganda
>> > >>
>> > >>
>> > >> _______________________________________________
>> > >> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> > >> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> > >> archive at http://puck.nether.net/pipermail/cisco-nsp/
>> >
>> > Regards,
>> >
>> > Mark Tinka - CCNP
>> > Network Engineer, Africa Online Uganda
>> >
>> >
>> > _______________________________________________
>> > cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> > https://puck.nether.net/mailman/listinfo/cisco-nsp
>> > archive at http://puck.nether.net/pipermail/cisco-nsp/
>> >
>> 
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>> 

> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/

-- 
Jeff Chan
mailto:cisco-nsp at jeffchan.com
http://www.supranet.net/