From: Petr Lapukhov (petr@internetworkexpert.com)
Date: Thu Jul 19 2007 - 03:34:21 ART
Correct point - this is just a scalability issue.
Just like with "classic" OSPF area 0, OSPF superbackone is used to "scale"
OSPF over MPLS core.Surely enough, if you were to run a single area over
MPLS core (e.g. over a mesh of GRE tunnel), that may lead to unnecessary
LSA flooding (LSA1/LSA2) and link state database synchronization, full SPF
recalculations on all routers, lack of information hiding and, therefore,
inability
to summarize intra-area routes.
Then again, BGP is distance-vector by it's nature, so it's pretty well
motivated
to simulate OSPF inter-area route redistribution with it :) Just imagine
that
you have a core network running RIP with OSPF areas connected on the edges -
pretty similar situation, with the except of the fact that BGP is capable of
carrying
much more information than RIP does (thus allowing to carry full set of OSPF
attributes)
Of course, sham-links are used to simulate the very behavior of *single*
area, but
this is for traffic-engineering purposes, and for use in a well-understood
and
controlled environment.
HTH
-- Petr Lapukhov, CCIE #16379 (R&S/Security/SP) petr@internetworkexpert.comInternetwork Expert, Inc. http://www.InternetworkExpert.com
2007/7/18, Alex <alex.arseniev@gmail.com>: > > Hi there, > RFC4577 might help http://www.ietf.org/rfc/rfc4577.txt > (O) routes might appear as (IA) or (E) on remote CE depending on OSPF > domain-id value. The design is probably taking into consideration the fact > that OSPF uses SPF algorithm for intra-area paths and distance-vector for > inter-area paths so injecting Type 3/Type 5 LSA does not lead to full SPF > recalc. > My 0.02$ > Rgds > Alex > > > ----- Original Message ----- > From: "Ian Blaney" <ian.blaney@gmail.com> > To: "Mohamed, Liban [NTK]" <Liban.Mohamed@sprint.com> > Cc: "Cisco certification" <ccielab@groupstudy.com> > Sent: Wednesday, July 18, 2007 4:45 PM > Subject: Re: MPLS VPN - OSPF PE-CE > > > > Mohamed > > > > Sorry you have misunderstood my question. > > > > This is not a real life scenario but just a question in my mind about > how > > OSPF works in a MPLS VPN setup. Why does an internal OSPF type 1 (o) on > > one > > site always appear as a type 3 (IA) on another within a VPN. It is > > designed > > this way. My question is really why is it designed this way? > > > > Cheers > > Ian > > _______________________________________________________________________ > Subscription information may be found at: > http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Aug 18 2007 - 08:17:41 ART