On 11/1/07, Nicolaj Kamensek <[EMAIL PROTECTED]> wrote: > Chris Kawchuk schrieb: > > Chris, > > > You are correct. cFEB can be either 128 or 256. Again, since the cFEB > > has all the actual forwarding routes for the router's ASIC's, you mya be > > able to get away with only 128M for now, but again, as the global > > routing table gets bigger and bigger, you may run into a limit. > > but not because of the DRAM on the FEB. The forwarding-tables itself are > stored in the SRAM from the FEB and on M7i CFEB it's 8MB in size(maybe > available with 16mb, I don't know) which can hold up to approximately > 550.000 routes. The dram only has a copy of the SRAM content. > But the dram is important for other things like arp entries.
550k IPv4 routes, I guess... so if all IPv4 prefixes get an IPv6 counterpart which has a prefix size twice as IPv4's, that would map to a 183k (IPv4 + IPv6) routes. Hummmmmmm... > > Likewise, if you start adding L3VPNs, and add more and more MPLS/VPN > > routes, you will run into the 128 Mb limit quickly. > > > > Hence, 256M is strongly recommended. > > I agree. Even on a Internet Exchange where there few ARP entries ? A cable scenario would use thousands of ARP entries... or there are any other reasons for having 256M on the FEB ? Rubens _______________________________________________ juniper-nsp mailing list juniper-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/juniper-nsp