On Thu, 2009-11-05 at 13:54 -0600, Josh Patten wrote: > OK I just had a thought about location based DNS SRV that I wish to > share with the members of this list in hopes of sparking discussion. A > while back, when I actually implemented a redundant setup, I set the DNS > server to have "views" ( > http://www.oreillynet.com/pub/a/oreilly/networking/news/views_0501.html > ) in that certain subnets would see the same DNS records but with > different SRV priorities for different servers. > > The big advantage to this is that I could give the local sipX redundant > sipX proxy a higher priority on that subnet than the core sipX server so > in the event of a severed connection, extensions could still call each > other and make emergency calls without needing to use a separate > survivability proxy and without a drop in service. This is also useful > for locations that have a large distance between them so call signaling > stays local whenever possible. The disadvantage is that manual > configuration is necessary and DNS changes to one location's zone view > require changes in all other zone views as well.
What you describe is a perfectly kosher thing to do. Congratulations on working it out (documentation on the wiki would be good to help others follow in your footsteps). This is exactly what the priority field of the SRV record is meant to be used for. > My question is has anyone (other than me) ever attempted this (it > worked, BTW), and what the developers think of implementing a frontend > into sipXconfig to make implementing this type of setup easy, or if it > would be a waste of time/resources? There are a good many things that could be done better if we had a real DNS manager built into sipXecs. Volunteers to build one would be more than welcome (here's a chance to be a hero). The big deployment challenge with that is that many users already have some other DNS server (and some of those are M$), so it can be difficult to provide the configuration in a useful way (a chance to be a really Big Hero). > Also, is there a limit to the number of redundant proxies that can be > configured? Last I heard it was 3 but I don't remember if that was a > technical limitation or just an artificial one. At present it is 3, but it is an artificial limit. At some point, the synchronization of registrations in an N-way configuration would begin to be a problem. We've tested 5 in the lab, and it seemed to be ok, but it wasn't for a long time or under very heavy load. _______________________________________________ sipx-users mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipx-users Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users sipXecs IP PBX -- http://www.sipfoundry.org/
