RE: how does as-set change the bgp best path selection?

From: Omer Ansari (omer@xxxxxxxxxx)
Date: Thu Aug 29 2002 - 18:36:29 GMT-3


   
Brian,

two things:

1.

> Weight
> Local-Pref/Locally Originated <---
> AS-Path
> Origin
> MED
> EBGP over iBGP
> Shortest IGP path
> Router-ID (lowest)

I thought there was a distinction b/w the local-pref and the locally
originated criteria (steps2 and 3) as in:

http://www.cisco.com/warp/public/459/25.shtml

2. this is the reasoning behind my question:

without as-set the aggregate address bgp route shows up as a locally
originated via igp (just like one generated using the network command) and
all is well in the best path selection.

but, with the as-set, show ip bgp would show this aggregate address with
all the ASes appended [am sure this is how the neighbors would see it, not
sure how the local router would see it]

if indeed the local router would see the route with all the ASes
appended, wouldn't that disqualify this aggregate address as being a
locally generated route then?

> > -----Original Message-----
> > From: Omer Ansari [mailto:omer@ansari.com]
> > Sent: Thursday, August 29, 2002 3:31 PM
> > To: Brian McGahan
> > Cc: ccielab@groupstudy.com
> > Subject: RE: how does as-set change the bgp best path selection?
> >
> > Brian,
> > thanks that makes sense.
> >
> > however, my question was targetting more the bgp best path selection
> than
> > the eventual routing of the packets.
> >
> > say we have a bgp route 12.1.2.0/24, and we are also learning a
> 12.0.0.0/8
> > network from a neighbor.
> >
> > we also have "aggregate-address 12.0.0.0 255.0.0.0" command on the
> > router.
> >
> > if local_pref of both the local and learnt aggregate are the same and
> no
> > weights have been applied on routes, the bgp best path algo should put
> > the [local] aggregate address going to Null0 in the route table
> instead
> > of the "learnt 12.0.0.0/8" route.
> >
> > this entry ofcourse was never competing with the 12.1.2/24 entry in
> the
> > best path algo.
> >
> > ..
> >
> > if the above is true, my question was if the aggregate address was
> used
> > with as-set, does this drop the aggregate address further down in the
> best
> > path algo selection?
> >
> > e.g if the "learnt 12.0.0.0/8" route has a low bgp metric, and that
> turns
> > out to be the next criteria for path selection would that route be put
> in
> > the table or the aggregate-address for 12.0.0.0/8 with as-set..
> >
> > regards,
> > Omer
> >
> >
> > On Thu, 29 Aug 2002, Brian McGahan wrote:
> >
> > > Omer,
> > >
> > > You will never route traffic to a locally generated aggregate,
> > > since the outgoing interface is null. Besides, if the local router
> is
> > > doing the aggregation, it should always have a longer match to the
> > > destination network in the first place. This is the reason the null
> > > route is used, to in fact prevent the local router from routing
> traffic
> > > for which the longest match is your locally generated aggregate.
> > >
> > > HTH
> > >
> > > Brian McGahan, CCIE #8593
> > > Director of Design and Implementation
> > > brian@cyscoexpert.com
> > >
> > > CyscoExpert Corporation
> > > Internetwork Consulting & Training
> > > http://www.cyscoexpert.com
> > > Voice: 847.674.3392
> > > Fax: 847.674.2625
> > >
> > >
> > > > -----Original Message-----
> > > > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On
> Behalf
> > > Of
> > > > Omer Ansari
> > > > Sent: Thursday, August 29, 2002 3:39 AM
> > > > To: ccielab@groupstudy.com
> > > > Subject: how does as-set change the bgp best path selection?
> > > >
> > > > All,
> > > >
> > > > we know that local routes from the bgp router are 3rd in the
> selection
> > > > list:
> > > > http://www.cisco.com/warp/public/459/25.shtml
> > > >
> > > > this includes bgp routes generated using the aggregrate-address
> > > command.
> > > >
> > > > however if the as-set option is used, does that route still
> qualify as
> > > a
> > > > locally generated route and if not, does that drop this route
> further
> > > down
> > > > in the bgp path selection list?
> > > >
> > > > thanks
> > > > Omer



This archive was generated by hypermail 2.1.4 : Sat Sep 07 2002 - 19:48:42 GMT-3