Hi Arnaud, You nearly got it right. Only small thing however is such packets will be rate limited to the CPU (software), so we will drop all packets not conforming to the rate limiting.
The reason it will be used is only for management purposes only (hence the header should never be used for data traffic - which seems like the only reason why people want to use the routing header). Thanks, Vishwas On 6/4/07, Ebalard, Arnaud <[EMAIL PROTECTED]> wrote:
Hi Vishwas, Le 4 juin 07 à 04:20, Vishwas Manral a écrit : > The idea is that for every router the packet goes through, we need to > check the IP address of all the interface addresses, and make sure > that the none of the interface address either before or after in the > source routing header match any of the IP address of the packet. Can you tell me if I understand correctly by commenting that example: for every packet that includes a Routing Header with something like 88 addresses in it (feasible with a 1500 PMTU), a router with 5 addresses will check every address it is configured with against those 88 ones (440 comparisons). With 100 Mbit/s of traffic, the router can see almost 9000 such packets per second. Where I come is that it will have to perform something like 4 millions IPv6 address comparison per sec, and in slow path. Is that correct ? Regards, a+ -- Arnaud Ebalard EADS Innovation Works - IT Sec Research Engineer PGP KeyID:047A5026 FingerPrint:47EB85FEB99AAB85FD0946F30255957C047A5026
-------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
