What are the *current* limits?
curtis at ans.net
Fri Feb 9 16:31:04 UTC 1996
In message <m0tiafL-000198C at zachs.place.org>, Paul Jimenez writes:
> Excuse my newbiness, but with all the current talk of reducing route table si
> and such, I have a few questions:
> What is a current router table size? How is it measured? (in routes? or
> kilobytes? or megabytes?) How big a router is needed to handle it? By
> 'big' I mean how much RAM and/or cpu is needed to deal with just routing
> and route table updates and flap control?
> Sepearte, but related, is the issue of bit shoveling. Given the minimum
> router that will handle a full routing table, how much bandwidth can
> it shovel through its various ports? I'm assuming this is related more to
> CPU speed than to memory constraints. Given a full routing table, how
> much bandwidth can routers of various sizes handle?
> I guess what I'm looking for are the real 'just-this-side-of-breakage'
> limits. So we need to reduce route table size - okay, fine, but _how much_
> do we need to reduce it? at what size does it become handlable?
> I hope these questions make sense, I'm fairly new at this.
> See you at NANOG!
> --Paul Jimenez
> Freeside Communications
I don't think anyone answered your question and with good reason.
This topic is the source of endless debate and those in the know don't
want to go through it again. Check the archives. There is no
agreement as to what the limits are in terms of X number of forwarding
entries or BGP paths, or routes changes per unit of time, but those
are the factors. The limits themselves differ among router vendors
and individual products and even software builds.
The traffic scaling issues are related to routing scaling. The
performance of most routers (all that I know of) is affected by the
routing load, if nothing else because forwarding entries must be
changed somehow. I don't want to comment on this further.
I can't comment on how routers will change. It safe to say the limits
of available technology (routers you can pick from the glossy and
actually buy) will be raised in 1996, though deployment of the latest
technology will proceed at different rates in parts of the Internet
and growth will continue. Administrative efforts such as those to
better aggregate must also continue.
More information about the NANOG