FIB compression makes it go over the ASIC’s official limit - as far as I know the 1M+ is from the ASIC’s spec
Pedro Martins Prado [email protected] / +353 83 036 1875 > On 4 Dec 2025, at 13:58, Mike Hammett via NANOG <[email protected]> wrote: > > Is FIB compression accounted for in their "over 1M" routes claim, or is that > on top of the spec? > > > > ----- > Mike Hammett > Intelligent Computing Solutions > http://www.ics-il.com <http://www.ics-il.com/> > > Midwest-IX > http://www.midwest-ix.com <http://www.midwest-ix.com/> > > ----- Original Message ----- > From: "Jon Lewis" <[email protected] <mailto:[email protected]>> > To: [email protected] <mailto:[email protected]> > Cc: [email protected] <mailto:[email protected]>, "Mike Hammett" > <[email protected] <mailto:[email protected]>> > Sent: Wednesday, December 3, 2025 7:40:34 PM > Subject: Re: Arista 7280QR-C36 Viability > > >> On Dec 1, 2025, at 8:57 AM, Mike Hammett via NANOG <[email protected]> >> wrote: >> >> At the risk of the Streisand effect, what am I missing about the Arista >> 7280QR-C36? It looks like a great router for small ISPs (great price, large >> packet buffers, good port selection, meaningful hardware routes). That said, >> it looks to be right on the border of DFZ viability. It supports "over 1M" >> routes, but I currently have about 1,036,824 in my route table. How much >> over is "over"? What happens in EOS when it goes over? > > If you let it run out of FIB space, those routes that don’t make it into the > FIB blackhole. FIB compression works quite well though, especially if you > enable it. :) > > > > > > _______________________________________________ > NANOG mailing list > https://lists.nanog.org/archives/list/[email protected]/message/YCMPBILFA7RASKAU5RFND62YBUHWBLGO/ _______________________________________________ NANOG mailing list https://lists.nanog.org/archives/list/[email protected]/message/3XBZCXCTG7OZKOMC6FRTQQ6OAT6RJPJH/
