When discussing routing protocols, a lot of literature refers to routers as
"nodes." I think you were right the first time, Chuck. A RIP network with a
few thousand router nodes would work, depending on the topology, as you
mention below. The 15-hop count limits the distance between networks, not
the max number of routers.
In addition to the other scalability problems Chuck mentions, if there were
any slow links you might run into a bandwidth problem. RIP sends the whole
table (after applying split horizon) every 30 seconds. Each packet fits
just 25 routes, so multiple packets are required to send 1000s of routes.
Each packet is 532 bytes long.
Don't know how we got here from a question on IS-IS versus OSPF. One design
challenge with OSPF that Chuck alluded to is that sometimes a network
topology doesn't have a good candidate for the area 0 network. If the
network grew without much concern for hierarchy, there may not be any
obvious high-speed core that could work as Area O.
Priscilla
At 12:33 AM 5/22/01, Chuck Larrieu wrote:
>I rechecked the NANOG archive, and I believe you are correct. It was several
>thousand nodes.
>
>As to the number of routers theoretically possible in a RIP domain, you
>might be surprised if you were to think through the math.
>
>Take a router. Connect ten routers. Connect ten routers to each of those
>ten. You can do this seven times, and the max distance from any router on
>the periphery to any other router on the periphery is 15 hops - seven in
>and seven out again.
>
>10^7 = 10,000,000
>
>if that center router began with 100 directly connected routers, the number
>grows astronomically, and yet the max diameter would remain 15 hops.
>
>the real restraint would be the ability of the router to hold a routing
>table that big. Along with the problem of convergence. Even if there were no
>network problems ever, I bet that sucker would take forever to converge!
>Literally!
>
>Chuck
>
>-----Original Message-----
>From: Curtis Call [mailto:[EMAIL PROTECTED]]
>Sent: Monday, May 21, 2001 6:38 PM
>To: Chuck Larrieu
>Cc: [EMAIL PROTECTED]
>Subject: RE: How is IS-IS more scalable than OSPF? [7:5207]
>
>
> >BTW, I have been told by folks who work in really big networks that none
of
> >the routing protocols scale beyond 4-5K routers. As an interesting aside,
a
> >few weeks ago on NANOG there was a discussion about the largest RIPv1
> >network in existence. It was revealed that until a year or two ago, Xerox
> >used RIPv1 and had a few thousand routers running RIPv1 on the network.
>
>
>I believe that the RIP network you are referring to had a few thousand
>nodes, not a few thousand routers. I doubt a few thousand routers could
>handle RIPs max 15 hop limitation.
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=5463&t=5207
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
