The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] [mpls] LDP Multicast to WG Document
Hi Thomas, > I believe liushuying was perhaps more worried about control plane load, > at reconvergence, for a high number of P2MP tunnels. > Though the concern is valid, I would say that things should roughly be > comparable to what happens with PIM. But I think it is fair to say that > this aspect requires consideration in the solution design. LDP uses TCP to send the label mappings to its peers. So in case of a large number of P2MP LSP we will take maximum benefit of the flow control en reliability in TCP, I don't think we can do better then that. I think the real problem behind the need to send many updates is that during that time, which increases with the number of updates, you have traffic loss. Thx, Ice. > > Cheers, > > -Thomas > > IJsbrand Wijnands: >> >> We have some ideas on how to minimize the traffic loss during >> switchover, but I'm not sure if we can prevent it without some kind of >> FRR mechanism. >> >> Thx, >> >> Ice. >> >> BTW, I support make this a WG document. >> >> >> On Feb 9, 2006, at 7:56 AM, liushuying wrote: >> >>> Hi folks, >>> I support this draft becoming a WG document. >>> >>> However,i am about to ask a question. >>> >>> 1) 2.3.2.4. Upstream LSR change >>> >>> If, for a given node Z participating in a P2MP LSP <X, Y>, the >>> upstream LSR changes, say from U to U', then Z MUST update its >>> forwarding state by deleting the state for label L, allocating a >>> new >>> label, L', for <X,Y>, and installing the forwarding state for L'. >>> In >>> addition Z MUST send a Label Map <X, Y, L'> to U' and send a Label >>> Withdraw <X, Y, L> to U. >>> >>> The question: >>> When a large quantity of P2MP LSPs exists in MPLS networks, >>> if unicast routing information changes, there may be many LSRs >>> whose upstream LSRs have changed in every P2MP LSP, each of them >>> sends a Label Withdraw message >>> to its old upstream LSR and a Label Map message to its new upstream >>> LSR at the same time, >>> so the load of every LSR will enhance sharply and last for a long >>> time. >>> As a result, the time of traffic disruption in every P2MP LSP will >>> last for a long time, >>> and so many packets will be lost. >>> >>> Thanks >>> Shuying Liu >>> >>> >>> ----- Original Message ----- >>>> Date: Mon, 06 Feb 2006 07:50:51 -0500 >>>> From: George Swallow <swallow@cisco.com> >>>> Subject: [mpls] LDP Multicast to WG Document >>>> To: mpls@ietf.org >>>> Message-ID: <20060206125051.3E3BD17B472@SwallowPB.local> >>>> >>>> It has been requested that >>>> >>>> Label Distribution Protocol Extensions for Point-to-Multipoint and >>>> Multipoint-to-Multipoint Label Switched Paths >>>> draft-minei-wijnands-mpls-ldp-p2mp-00 >>>> >>>> become a MPLS WG document. >>>> >>>> Please indicate your support or objections by the end of this week. >>>> >>>> Thanks, >>>> >>>> ...George >>>> >>>> ==================================================================== >>>> == >>>> == >>>> George Swallow Cisco Systems (978) >>>> 936-1398 >>>> 1414 Massachusetts Avenue >>>> Boxborough, MA 01719 >>>> >>>> >>>> >>>> >>>> >>>> ------------------------------ >>>> >>>> _______________________________________________ >>>> mpls mailing list >>>> mpls@lists.ietf.org >>>> https://www1.ietf.org/mailman/listinfo/mpls >>>> >> >> >> _______________________________________________ >> mpls mailing list >> mpls@lists.ietf.org >> https://www1.ietf.org/mailman/listinfo/mpls > > > _______________________________________________ > mpls mailing list > mpls@lists.ietf.org > https://www1.ietf.org/mailman/listinfo/mpls _______________________________________________ mpls mailing list mpls@lists.ietf.org https://www1.ietf.org/mailman/listinfo/mpls
|
|