Selon Basavaraj Patil <[EMAIL PROTECTED]>: > > Inline: > > > On 8/8/06 10:54 AM, "ext Pars Mutaf" <[EMAIL PROTECTED]> wrote: > > > Hello, > > > >> From your draft: > > > > |Routers that implement the current recommendations would send the > > |periodic multicast router advertisements every 30 minutes, which can > > |be a significant problem in mobile/cellular network environments. > > > > Why do you think 1 multicast RA per 30 minutes is a significant problem > > for a host? Personally, I would think that 1 RA per 30 mins has a > > negligible impact on energy cost. 30 minutes is really longtime IMHO. > > So if you consider for example a GGSN which may have several hundred > thousand hosts attached to it (GGSN the AR for all these hosts), having to > send a periodic RA to all these hosts is an expense with no real benefit. > In order to deliver the RA, the host has to be paged, radio resources > allocated and a traffic channel established to deliver the RA. > > So what is the benefit of doing this operation? I agree that waking up a > host every 30 minutes to deliver an RA is not a significant drain on the > battery, but it is a factor. Additionally the fact that you have to deliver > the RA as a unicast message to a very large number of hosts is a waste of > the radio resources. Having a host establish a traffic channel only for the > purpose of receiving the periodic RA is sub-optimal. Why have a specific > interval (MAX) of 30 minutes only? Why not let it be 180 mins or 240 mins? > 30 Mins is a value IMO which is as random as choosing 180, 240 or something > else.
OK. Flooding the network with RA messages unicasted to several hundred thousands hosts (wow!) "at the same time" is a significant problem. But, I think you will agree, the problem is not the period of RAs. 30 or 180 minutes later you will have the same serious problem. Your subnet is too large IMHO. Regards, pars > > > > In addition this may be useful: > > Every 30 minutes (or less), my host will wake up, look around i.e. > > receive the RA, thus make sure that the access router is *still* > > with me, and sleep again. > > If the host wants to make sure whenever it wakes up (not periodically just > for receiving an RA), it can just as well send a RTR solicitation to confirm > that it still is connected to the AR and be happy. > > -Raj > > > > > What's wrong with that? > > > > Thanks > > pars > > > > > > > > > > > > On Mon, 2006-08-07 at 22:33 -0500, Basavaraj Patil wrote: > >> Hello, > >> > >> The I-D: > >> > http://www.ietf.org/internet-drafts/draft-madanapalli-ipv6-periodic-rtr-advt > >> s-00.txt > >> > >> proposes several changes to ND procedures and parameters. > >> > >> Pls review and comment. > >> > >> -Raj > >> > >> > >> -------------------------------------------------------------------- > >> IETF IPv6 working group mailing list > >> ipv6@ietf.org > >> Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 > >> -------------------------------------------------------------------- > > > > ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
