[c-nsp] MPLS VRF EIGRP scale

Bruce Pinsky bep at whack.org
Mon Aug 15 15:37:36 EDT 2005


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Jessup, Toby wrote:
> Does anyone have a feel for the maximum CE-PE EIGRP adjacencies on PE
> platforms recommended by Cisco?
> 
> To what degree, if any, is the recommended limit of CE-PE EIGRP
> adjacencies per-box less than the maximum recommeded CE-PE BGP
> adjacencies?
> 
> Many customers want CE/PE EIGRP over SP "MPLS VPN" product offerings. Is
> there any compelling technical reason why an SP would not embrace CE/PE
> EIGRP as a standard service for customers? I understand the provisioning
> objections ....
> 

A few that come to mind...

OSPF and IS-IS offer fast convergence options (potential for subsecond).
BGP eliminates the redistribution requirement (and hence the CPU load) from
EIGRP to MP-BGP on the PE.
EIGRP (or any IGP derived route) routes from a dialup backup connection
will be preferred over the VPN-derived route since that route will be an
external (via the redistribution from MP-BGP to EIGRP).  Can be worked
around in OSPF via sham-links. EIGRP requires Site-of-origin with EIGRP
backdoor link.
Multi-homed sites may require the use of EIGRP Site-of-origin to prevent loops.

- --
=========
bep

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (MingW32)

iD8DBQFDAO8AE1XcgMgrtyYRAic2AKDOpLqijHTO4Vt0/hCoaxRyAGH38gCfQt/w
nZT1aK2peYqZbVN+ZLv/GGg=
=jjc8
-----END PGP SIGNATURE-----


More information about the cisco-nsp mailing list