On May 9, 2008, at 3:46 AM, Juha Heinanen wrote: > Dean Willis writes: > >> One nifty technique is to pair up the proxies, such that there are >> N/2 >> pairs, then define a hash across the namespace that maps to each user >> ID to one of those pairs, Your UA will then end up registering to and >> forming a flow with both proxies in its assigned pair. Other proxies >> in the pool that receive a request for you can use the hash to find >> the right pair, and if either proxy in the pair has lost the flow, it >> knows to consult the other member of the pair. > > dean, > > i would really love to implement something like this, but the > problem is > that because of ob draft you cannot assume that "UA will then end up > registering to and forming a flow with both proxies in its assigned > pair". >
Then you're reading a different outbound draft than I am, because as far as I know, enabling the UA to bind multiple proxies is what the outbound draft does. Without it, you have no real way to do this. Yes, binding multiple proxies is optional with respect to the protocol. In general, UAs will be configured (and this draft doesn't specify configuration) to bind multiple proxies, but it's true that UAs do not absolutely have to bind multiple proxies in order to place and receive calls. Deployments just scale the number of proxies based on the desired reliability. Since the mechanism works with a single binding, there's no rationale under RFC 2119 for making usage of multiple bindings a MUST. If we did that , then we'd have to make it a MUST for proxies to always be deployed in pairs, and that clearly isn't going to happen in every deployment scenario. I can only afford one proxy! > in order cope with proxy failure, you must then play with all kinds > tricks to make the still working proxy in the pair to assume the ip > address and (in case of tcp) the flows of the dead proxy, which is not > easy at all. No, in order to deal with proxy failure in a single proxy case, the easiest thing to do is wait for the UA to note that its flow has dropped (because its keepalive failed) and rebind. You could certainly do a state-transfer to the standby-router thing, but that probably costs more than just deploying dual proxies would have. But if the UA used the optional multi-proxy binding then there would be no need to wait. This provides operational advantages, at cost, so it's up to each deployment to decide whether the advantages outweigh the costs. -- Dean _______________________________________________ Sip mailing list https://www.ietf.org/mailman/listinfo/sip This list is for NEW development of the core SIP Protocol Use [EMAIL PROTECTED] for questions on current sip Use [EMAIL PROTECTED] for new developments on the application of sip
