[c-nsp] OSPF redist customer routes

Jay Hennigan jay at west.net
Tue Nov 13 03:21:18 EST 2012


On 11/12/12 9:55 PM, CiscoNSP_list CiscoNSP_list wrote:

> Thanks Jay - We already run iBGP(Full mesh under VPNv4) across our POPs
> for vrf solutions....how best to migrate our customer routes from
> ospf->iBGP? (And how to separate our infrastructure IPs(Keep in OSPF)) 

Without knowing the details of your network it's going to be tough to go
step-by-step.  Assuming that you already have loopbacks on your routers
in OSPF, BGP points to the loopbacks, and that you have full mesh iBGP
or route reflectors in the global table, start with one router and
redistribute static and connected into BGP.  Use a route map limiting
redistribution to customer prefixes or a single customer prefix for
testing.  The same route map can inject communities as needed (no-export
would likely be nice).  These would be in the global table unless in a
VRF but you're already doing that.

Take that prefix out of OSPF and verify that it propagates to your POPs,
is reachable throughout your network and doesn't leak outside your AS.
Repeat until you have all OSPF customer routes removed from a single
router, then on to the next.   iBGP is distance 200 and OSPF is 110 so
you won't see the BGP route in the forwarding table until you remove the
OSPF one.

>> Customers with redundant connections can use a private AS into iBGP or
>> tracked floating statics redistributed.
> 
> A lot of our customers CE's dont support BGP (Or require a license
> upgrade)...so we are stuck(to a degree) with having to support OSPF?

For non-redundant customers a static default at the customer edge is all
that you need.  For redundant customers either upgrade to BGP at the CE
or use a floating static for the backup with the inverse at the PE.  For
backup routes we use a tagged floating static distance >200 on the PE
and a route map to match the tag, set weight to 0 and de-pref local pref
so that the backup doesn't propagate until the primary goes down.

And as Andrew pointed out, if you use a private AS for BGP to the
customer prem, then it is actually eBGP.

I seem to recall a fairly good presentation writeup on OSPF-BGP
migration in the NANOG archives but a quick search comes up empty.

--
Jay Hennigan - CCIE #7880 - Network Engineering - jay at impulse.net
Impulse Internet Service  -  http://www.impulse.net/
Your local telephone and internet company - 805 884-6323 - WB6RDV


More information about the cisco-nsp mailing list