Have recently watched Matt Jordan‘s session on Kamailio World 2014
On slides 26-29 of his presentation
he speaks about a (completely new, for me at least) approach to build scalable telephony systems, using N instances of Kamailio and N
instances of Asterisk
Are there any whitepapers, howtos, “implementation experience reports”, whatever, available, that would describe such an approach in details and help some not-so-advanced admins to at least understand “if is it what they need, or not exactly, or not at all” ?
We are planning to look closer at Kamailio (or any other proxy, like OpenSip) as a way to do both load-balancing and failover solutions, so that refusal of any Asterisk instance should have minimal possible effect on the overall system availability.
A lot of questions howevere arise, like: what if one SIP user got REGISTERed at Server 1, and the other on Server 3, so how can they call one another ?
Also, outbound registrations can be done from one instance at a time, say it’s done from Server1 for Trunk1, so how can users, that got authenticated at Server2, call thru that registration (Trunk1) ?
Also, Kamailio itself has to be protected from failing, and probably even from overload…
Would be great to read something in-depth about that