Very, very interesting topic!
Lync sends OPTIONS sip request every minute or so. If the gateway/sip trunk does not reply in timely manner, the destination is marked “internally” as Down. Now, even thou marked, if we have only one gateway, the next outbound call will be send to the gateway anyway and it will either be established (if the issue was resolved), or fail (if has not been). In case of two or more gateway, the call will be offered to the next gateway that properly responded to OPTIONS and meanwhile, Lync will still try to verify the first gateway availability by periodically sending OPTIONS request. Once promptly replied with 200 OK, this gateway will be marked as “available”.
To achieve a real resiliency, one must have two independent (in case of sip trunking) ITSP because having two or more gateways connected to the same or different provider(s) does not mean anything if your WAN goes down (happened to me last week).
On provider’s side, there are few options; just… they vary by provider. Mine for example, would accept inbound from two different ordinate IP but send to one only. I know another provider in Florida who’s soft switch monitors the gateway availability in the same manner as Lync does, and route the call to alternative destination if the primary does not respond, while accepts from at least two ordinate IP’s. This is, more or less, full resiliency on provider’s side.
Drago
http://www.lynclog.com
