Newsgroups: comp.dcom.sys.cisco
Path: utzoo!utgpu!cunews!bnrgate!bwdls61!bwdls56!fortinp
From: fortinp@bwdls56.bnr.ca (Pierre Fortin)
Subject: Re: RIP vs IGRP
Message-ID: <1991May7.144345.21683@bwdls61.bnr.ca>
Sender: usenet@bwdls61.bnr.ca (Use Net)
Reply-To: fortinp@bwdls56.bnr.ca (Pierre Fortin)
Organization: Bell-Northern Research, Ottawa, Canada
References:  <34800@boulder.Colorado.EDU>
Date: Tue, 7 May 1991 14:43:45 GMT

In article <34800@boulder.Colorado.EDU>, scsabir@tvgurus.hdtv.zenithe.com (Andrew Birner) writes:
|> 
|>  I was all set to implement my wide-area network using IGRP; now that Novell has
|> released NetWare 386 v3.11, which uses RIP in its TCP/IP implementation, I fear
|> I have to revisit the routing issue.  The problem, for me, is that Novell just
|> dumped about a dozen RIP-based routers onto my network; I need a way to deal
|> with them.  My choices seem to be: (1) run IGRP on the WAN, and force the Net-
|> Ware routers to use their local cisco box for a default route; or (2) run RIP
|> everywhere.  Option (2) appears easier to deal with, since there are no static
|> routes to deal with; it is also more consistent.  However, IGRP (and OSPF) is
|> a better routing protocol than RIP.  My network is simple enough now that I
|> would probably have no trouble running as a RIP-based net; that may change,
|> though, so I don't want to shoot myself in the foot by choosing RIP.  (Does any-
|> one know of an IGRP or OSPF implementation for NetWare v3.11....)
|>  Anyway, I'd appreciate any comments on or solutions to this dilemma; thanks in
|> advance for any and all helpful input.
|> 

Check the cisco manual for "redistribute" and "default-metric" (router) 
subcommands.  We have set up our network (over 260 subnets) with IGRP.  Most
devices use Proxy ARP to gain access to remote devices.  However, for those 
devices which can't use Proxy ARP, we redistribute the IGRP routes into RIP
with a default-metric of 8 (hey, you can pick anything you want as long as it 
doesn't result in RIP-unreachables...:)   Note that cisco also allows 
redistributing RIP into IGRP, but *don't* do it unless you want to chase flapping
route problems when your network becomes larger and more complex.  The last 
thing you'll think of checking for is that new router/link...  :(

The point is:  use IGRP within the network, then send RIP routing packets out
on your Ether interfaces, but *don't* listen to those same RIP updates on a 
neighbouring router.  You'll get IGRP metrics squished into RIP's 16 and expanded
back out to one of 16 IGRP values.  This action is responsible for some route
flapping is a complex network topology.  
 
Another (router) command you should be aware of is "passive-interface"; it can 
be used to eliminate redundant (non-IGRP) updates from wasting bandwidth on 
your serial links (and possibly some ether ports).

|> - Andrew E. Birner, Zenith Electronics Corporation -

-- 
Cheers,                      
Pierre Fortin       fortinp@bnr.ca         (613)763-2598