>This is actually for a practical issue, I have a customer that wants to >implement +-400 remote sites connected with redundancy to two core routers. >Each router will have three T1's and the 400 sites will be split between the >three T1's. This still brings the EIGRP to +-133 EIGRP neighbors per >interface and 400 neighbors per router. The customer wants to run EIGRP. I >am asking this question to determine if this will be an issue and to find >documentation to back this up. The alternative would be to run OSPF or BGP >but I need backup info to get the customer to change. > >Thanks Doug
I'm afraid this won't be a specific answer, but some things worth looking at. My experience both as a network designer and a router architect makes me very nervous here. Fairly recently, I was working on a router design, and sales came up with the desire to have more neighbors per interface than the other guy. It happened to be OSPF. If you assume a 1500 byte MTU, which is a fairly safe real-world limiting assumption, you can only fit 47 (IIRC) neighbors into a single OSPF hello packet. From a code implementation standpoint, yes, we could have designed for "fragmented" hello packets, but that greatly complicated things. I don't know offhand if EIGRP has some of the same architectural limits. It's less that I can flatly say the router won't work, but that you are getting into areas where the code probably hasn't been fully tested. Sometimes it's actually cheaper to use several small routers. I had one installation which again was a hub-and-spoke, admittedly with a far fewer number of neighbors, but essentially 3 lines coming into the hub (well, actually 3 lines with 3 PRI ISDN as backup). I set up 3 routers, each with one dedicated line and one PRI interface. In your case, it might be interesting to cost out the configuration with 3 routers linked by an Ethernet. My intuitive sense is that it MIGHT work, but any major network disruption with flapping could drive the routers crazy. It will also depend if the routers use the same processor for forwarding and routing. Incidentally, I can say with confidence this would be extremely unlikely to work for BGP. It's not even the routing complexity -- the neighbor AS sending few or many routes -- as the processor overhead of maintaining that many TCP sessions. > >-----Original Message----- >From: MADMAN [mailto:[EMAIL PROTECTED]] >Sent: Tuesday, January 15, 2002 4:49 PM >To: [EMAIL PROTECTED] >Subject: Re: EIGRP neighbor limitations [7:32058] > > >I don't know about a hard limit but me thinks you'll hit the practical >limit first anyway:) Is this an acedemic question??? > > Dave > >"Robertson, Douglas" wrote: >> >> Does anyone know of limitation in the amount of EIGRP neighbors on a >router. >> If there is, is this a limitation per physical interface or a limitation >> per router. I found a document on CCO a couple of months ago that >mentioned >> these limits but I have now searched and searched but cannot find that >> document again. >> >> Appreciate any input >> >> D. Robertson >-- >David Madland >Sr. Network Engineer >CCIE# 2016 >Qwest Communications Int. Inc. >[EMAIL PROTECTED] >612-664-3367 > >"Emotion should reflect reason not guide it" Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=32167&t=32058 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
