[Irtf-rr] IP Table Size
Poh Tze Ven
tvpoh at essex.ac.uk
Tue Feb 4 17:48:05 EST 2003
Hello,
> It would help if your categorization made more sense. With a few
> hundred thousand routes, why exclude the few hundred (at most) ASBR?
The size of IP table from ASBR can be very different from an internal router. I
only need data from internal routers, and do not wish the information from ASBR
to bias it.
> Unless you mean exclude the ASE routes and count the intra-area and
> inter-area routes.
No, I try to study the ASE routes visible to internal routers.
> The routing information base (RIB) is what really
> matters, not the FIB, unless you are trying to build hardware and
> looking for what the ASIC FIB sizes should be.
I think I have made myself clear, I need FIB not RIB. Your last point is close
to what I am interested in (but not designing ASIC of that sort). Anyway if you
study ASBR carefully (which certainly you do), most of them use more than 95% of
RIB for FIB.
> If you are interested
> in the IGP size you need to know number of nodes (routers in an OSPF
> area, IS in an ISIS level) and adjacencies, which is relevant if your
> interest is computational scaling. It might help to first learn the
> terminology and get a clue as to what questions make sense to ask
> before asking.
I know what I am asking for. Thanks. The IGP size is not all that important, I
only want to do some rough correlation with the table size.
> You might also have more success looking at the various sites that
> provide full routing information and counting the routes yourself.
> You won't find the number of unaggregated routes inside any given ISP,
> just the global routing information.
I do not need the details and I am aware of the availability of such ASBR
routing information but that is not I am after. I need internal router routing
information and that cannot be found in the Net.
> You also won't find the number
> of nodes and adjacencies if that is what you are after.
No
> Last I heard it was around 250,000 routes for the largest ISPs with
> some routers having 1,000,000 route instances (more than one way to
> get there from here). Its a moving target. You can find number of
> routers and a rough idea of out degree (leading to number of links)
> from some past presentations at NANOG or elsewhere.
Again I am not seeking for such info. The reason for (iii) is that if given (ii)
I would like to correlate it with (iii)
My requirements are nothing as complicated as what you have thought.
cheers!
More information about the irtf-rr
mailing list