On Thu, 4 Mar 2004, Bob Hinden wrote: > >coz data from the client may be going thru a different device Y, which is > >being blocked by the fw on that device. fw Y doesn't have the hole > >to let the traffic go through. > > This won't be caused by the load sharing when the data and control are > going to the same destination host. If the data traffic is going to a > different destination host, then it could end up on a different router > because of a more specific route or a redirect. I don't see that load > sharing adds to the complexity.
In cases in which this is mostly important, the hosts have a default route. Then e.g. SIP data and signalling should go through the same path. This is the scenario which fails (in addition to the generic debug etc. problems which happen when something breaks from one first hop router to somewhere in the direction of Internet. Hosts which have more specific routes are set there explicitly, manually or using the more specific route extension. I'd still like to remove the "MUST share load" from the document. Of course, I can see that it's very useful in many scenarios for load distribution. But still not a MUST. What I want to see that the implementations implement this feature, so that it's available for use for those who need it. But I want it disabled by default and a MUST for a toggle to enable/disable it. to implement draft-ietf-ipv6-host-load-sharing-01.txt, otherwise I don't buy your product". Does not help in getting stuff like this deployed. -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
