RE: [nsp] how to fool the SPF in OSPF?

From: George Robbins (grr@shandakor.tharsis.com)
Date: Tue Jun 12 2001 - 23:49:57 EDT


I'm not sure why you think that would vary between protocols, the routes
need to exist one way or the other, the are then made known to the iGP
either implicitly or explicitly. If your network is well organized in
an cidr-block vs. geography sense, you can aggregate routes and have
things work correctly, you can also "redistribute" into BGP if you
take some care with a route-map to select what you want to suck in
(such as /30's) and set appropriate clever communities such as no-export,
to routes that you don't want to escape.

If your real question is "do I have to put in a network statement for
every customer route", the answer depends on how you want things to work,
the bricks are all there for whatever you want to build.

Of course, you can make arguments for the adequacy of OSPF, ISIS or
EIGRP for most ISP level iGP requirements, but only OSPF and BGP have
any real universiality.

                                                        George

PS: BGP confederations are the key when you get beyond the point where
        iBGP mesh rules start to become deadly.

> From cisco-nsp-request@puck.nether.net Tue Jun 12 23:00:54 2001
> Resent-Date: Tue, 12 Jun 2001 23:00:49 -0400
> Received-Date: Tue, 12 Jun 2001 22:58:38 -0400
> Date: Tue, 12 Jun 2001 22:58:28 -0400 (EDT)
> From: <jlewis@lewis.org>
> X-Sender: <jlewis@redhat1.mmaero.com>
> To: "Martin, Christian" <cmartin@gnilink.net>
> cc: "'Chris Davis'" <chris.davis@computerjobs.com>,
> <cisco-nsp@puck.nether.net>
> Subject: RE: [nsp] how to fool the SPF in OSPF?
> In-Reply-To: <94B9091E1149D411A45C00508BACEB359CDBC7@entmail.gnilink.com>
> Resent-From: cisco-nsp@puck.nether.net
> X-Mailing-List: <cisco-nsp@puck.nether.net> archive/latest/6792
> X-Loop: cisco-nsp@puck.nether.net
> Precedence: list
> Resent-Sender: cisco-nsp-request@puck.nether.net
>
> On Tue, 12 Jun 2001, Martin, Christian wrote:
>
> > This bit me a while back. I was trying to build a nice, "scalable",
> > multi-area network. The reasoning was that it was world-wide, with some
> > slot routers and WAN links. Enterprise networks are always painful in the
> > end when trying to engineer isp-like ideas into them...
>
> Our network's not _that_ big, though it does span several states. It's
> basically currently a star-like layout with one central point connecting
> to the internet and lots of spokes going out to POPs...some of which are
> actually chains or trees of POPs. i.e. grossly simplified with lots of
> spokes omited:
>
> POPa POPf
> | /
> POPb---center---POPc---POPd---POPe
> / | \ \
> internet POPg
>
> Things are interconnected with a mix of T1's and T3's. Having all our
> transit in one place has kept things relatively simple, but we will
> eventually be adding transit at multiple points, at which point it's kind
> of hard to define a center. We already have private peering at POPs other
> than the center. I like OSPF since it's an open standard and supported by
> lots of vendors. From what I've heard, it sounds like most of the larger
> networks run ISIS rather than OSPF. Is it time to start looking into
> switching?
>
> --
> ----------------------------------------------------------------------
> Jon Lewis *jlewis@lewis.org*| I route
> System Administrator | therefore you are
> Atlantic Net |
> _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________
>
>



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:41 EDT