When a client registers to PBXes extension, the url of the SIP server is always pbxes.org. Is connection gets redirected by geographic location to the closest server, or it should be explicitly defined to minimize latency?
What can be done for latency if clients registered to extensions of the same pbxes account located one in US and other is in Europe?
In order to help you in this specific situation, please provide us with the geographic locations of the actual SIP UAs, and the server on which the account is currently hosted.
To partially answer your question though, we should define the two kinds of latency involved in every call. First, is the signaling latency of the SIP packets, incurred when setting up or tearing down a call. Second, is the media latency of the RTP packets which carry the voice, which is controlled via SIP Re-Invites (the Audio Bypass setting found on the setup of Extensions & Trunks) which can affect the quality of the call.
I will concentrate on media latency. The media connections are made from the datacenter that you choose in 'Personal Data'. This can only be chosen once for an account.
So if you have endpoints near two datacenters media will either go thru one or the other, but not thru both.
If you want to go across two POPs you need to create two accounts.
If you want the audio packets go directly bypassing the POP enable 'audio bypass'.
Can 'audio bypass' work if SIP user agent is behind NAT?
The simple answer is "No", but I read that SIP providers with session border controllers are able to establish RTP 'bypass' if NAT setup correctly. How true is that?
