Re: Using alternate equal cost routes effectively


Casey Leedom (admin.cognet.ucla.edu!casey@CS.UCLA.EDU)
16 Sep 88 08:08:13 GMT


> From: dagg@lace.lbl.gov (Darren Griffiths)
>
> In many cases there are different paths to the same host, either via
> backup redundant links or because of longer hops through the network.
> Usually the gateways know about these different paths and take the
> fastest (determined by the metric) and leave the other path untouched.
> This other path may be idle and quite useful, there are two ways of
> making use of this path.

> From: kwe@bu-cs.BU.EDU (kwe@bu-it.bu.edu (Kent W. England))
>
> When you say "different paths" do you mean alternate routes to
> the same address or alternate addresses to the same host (which is the
> subject of the thread)? In the first case, you as a user don't care what
> paths your datagrams take so long as they get there quickly. However,
> load sharing is done by ciscoIGRP and is proposed in Moy's Open SPF IGP
> protocol draft for equal-cost routes.
>
> Presumably, the untouched routes are being used by other
> datagram streams.

  Where can I get a copy of Moy's protocol draft? I'd be interested to
know if it's anything like a scheme that Mike Karels described to me a
few months back. (Please forgive me Mike if I butcher this in trying to
explain it.)

  Mike suggested that:

          When a host H gets a RIP broadcast from gateway G2 advertizing
        a route to a destination D with cost C(G2, D) and H has already
        got a route to D through gateway G1 with cost C(G1, D) == C(G2,
        D), then H should ``throw a die'' to decide whether to switch
        gateways for D.

          The probability of switching should be very low to minimize
        routing table thrashing and allow timing statistics to develop so
        the new congestion controls wouldn't be defeated.

          The effect of such a modification of a host's response to RIP
        broadcasts would be to have hosts their routing choices out among
        a set of gateways offering equal cost routes to destinations. If
        a gateway goes down, all hosts currently using that gateway would
        immediately switch gateways. If a gateway comes up, hosts will
        slowly start using it. If a gateway is flakey and doesn't ever
        stay up reliably, almost no hosts will ever try to use it.

  Mike may well have been quoting Moy's idea, but I have no idea since we
were just talking informally. (Again Mike, please forgive me if I've
butchered this.)

Casey



This archive was generated by hypermail 2.0b3 on Thu Mar 09 2000 - 14:43:30 GMT