Hi Ladislav, Yes I think this is a crux of this WG. I saw your mail in the past but no precise answer from anyone ;)
Moreover even if one provides a RIB definition on one box today how do we make it extensible so new network applications can be deployed on it tomorrow. Worse how do we accomplish consistent RIBs being supported across multiple vendors ? Otherwise I can't see how read/write of any data can be abstracted for network wide services provisioning using I2RS. Regards, R. On Thu, Mar 14, 2013 at 3:22 PM, Ladislav Lhotka <[email protected]> wrote: > > On Mar 14, 2013, at 10:13 AM, Robert Raszuk <[email protected]> wrote: > >> Hi, >> >> As we have seen today I2RS is scoped to interface with RIB. Great. But >> what is the precise definition of RIB here ? Are we talking about > > I already raised pretty much the same question before: > > http://www.ietf.org/mail-archive/web/i2rs/current/msg00554.html > > I think one of the initial documents (architecture?) should include a > detailed terminology section. > > Lada > >> currently known RIBs or is the RIB definition yet to be defined. I >> think it should be something defined now as pretty much all other work >> will depend on it. I bet everyone in the room has different opinion on >> it ;) >> >> Example: >> >> Assume I have invented a routing protocol which routes based on tuple >> of dst IP address + port number. (Latency sensitive applications go on >> different links that non time sensitive ones). >> >> Is I2RS effort going to help me to push such routing data to the box's >> RIB or not ? Note that I am not talking about pushing bunch of ACLs to >> 100s of my boxes doing PBR. >> >> Thx, >> R. >> _______________________________________________ >> i2rs mailing list >> [email protected] >> https://www.ietf.org/mailman/listinfo/i2rs > > -- > Ladislav Lhotka, CZ.NIC Labs > PGP Key ID: E74E8C0C > > > > _______________________________________________ i2rs mailing list [email protected] https://www.ietf.org/mailman/listinfo/i2rs
