Since this old thread recently became alive (momentarily), and I read through the posts, (perhaps, again!) ...
Patrick, I would like to understand why you said that routers handling 10G traffic in DFZ are not bothered (much) by a few extra prefixes? Isn't this counter-intuitive? For example, for the worst case packet size of 40 bytes, a router has only 32ns to completely process a packet (including lookup!) in order to support 10Gb/s line rates. The higher rates leave with even smaller time, which makes me think that it's the "slow running" routers that should not be bothered *much* by a small increase in the number of prefixes. Or, were you referring to 10G routers "slow running" by comparing them with 100G routers? I do not except anyone to have such a long memory, so you may want to skim through the following :) Zartash On Sat, Dec 20, 2008 at 9:37 AM, Patrick W. Gilmore <patr...@ianai.net>wrote: > On Dec 19, 2008, at 10:48 AM, Joe Greco wrote: > > As for routing table size, no router which can handle 10s of Gbps is >>> at all bothered by the size of the global table, >>> >> >> ... as long as it isn't something like a Cisco Catalyst 6509 with SUP720 >> and doesn't have a PFC3BXL helping out ... >> >> ... or if we conveniently don't classify a Catalyst 65xx as a router >> because it was primarily intended as a switch, despite how ISP's commonly >> use them ... >> >> so only edge devices >>> or stub networks are in danger of needing to filter /24s. And both of >>> those can (should?) have something called a "default route", making it >>> completely irrelevant whether they hear the /24s anyway. >>> >> >> A more accurate statement is probably that "any router that can handle >> 10s of Gbps is likely to be available in a configuration that is not at >> all bothered by the current size of the global table, most likely at some >> substantial additional cost." >> > > Good point! I should have said "10s of Gbps and tables associated with > default-free networks". > > Or are there lots of people using 6500s without 3BXLs in the DFZ? I admit > I have not audited every router in the DFZ, so perhaps someone with factual > info can help out here. > > If not, then we're back to where we started. The DFZ isn't worried about > table size this cycle, and the edges can (should?) have default. I'm sure > that will change in a couple years, but everything always does. > > Oh, and before anyone jumps all over me, I am NOT implying you should > deaggregate and blow up the table. Just that 300K prefixes is the DFZ is > not a reason to start filtering /24s. Today. :) > > -- > TTFN, > patrick > > >
