A few weeks back, Matt Brunk posted an interesting entry about his experience with an outage at his ITSP, which took his business off line for over two hours. His experience is not unique and should send shivers down that back of IT managers and CIOs. Not picking on any one service provider, but many of the ITSPs are pretty new and some are still trying to work out the "what if" scenarios. Based on my discussions with end users, reliability and survivability are key objections to exploring SIP Trunking. They like the cost savings of SIP trunking, but desperately want to avoid the risk of a complete service outage (also known as a "resume generating event"). At the same time, I've seen some very creative solutions deployed with SIP trunks that reduce the risk of outages, and it seems like the time is right to share what I've learned.An increasingly common approach I'm seeing is "walk before you fly" with SIP trunking, using a "Hybrid TDM/SIP Trunking" architecture. This architecture keeps some/all of the existing TDM trunks in place, but adds SIP Trunks to handle a portion of the outbound traffic. The cost saving of this architecture is realized by moving the billable long distance traffic over to the SIP trunks while the TDM trunks provide inbound, local access and survivability. Frankly, it's a rare organization that does a full "rip and replace" on their TDM trunks, completely cutting over to SIP trunks for 100% of their telephony services.
The most common situation has end-customers with an existing TDM PBX in place that doesn't directly support SIP trunks. In this case, the SIP trunks would have to be terminated on a media gateway which would convert the trunks back to TDM for connection to available trunk ports on the PBX. Once the SIP trunks are converted back to TDM and connected to the PBX, the routing logic in the PBX can direct traffic to either trunk type, based on the service status of the trunks. However, this model does require that the media gateway can detect that the SIP trunk is down and busy-out or pass an alarm indication to the PBX, preventing the PBX from routing calls to a dead circuit.
But what if there are no available TDM trunk ports, or if expanding the TDM PBX is expensive or difficult? This requires use of the "drop and insert" installation, where a media gateway is "inserted" into the TDM trunk circuit(s)--one side of the gateway faces the PBX and the other faces the existing TDM trunk circuit--both sides using TDM interfaces. By default, the gateway would transparently route traffic from the TDM trunk to the PBX and visa-versa.
Still following me? Okay, now the fun begins by bringing in one or more SIP trunks and terminating them on the media gateway. With both TDM and SIP trunks terminating on the media gateway, routing tables within the gateway would be used to intelligently direct the traffic between the circuits and the SIP trunks, using dialed number and/or status of the trunks to control the route of the call. If the SIP trunks or the TDM trunks go out of service, the traffic would be re-routed to which ever facility is still operating, to ensure the call goes through.
Cost savings and survivability--all this without upgrading the PBX. Pretty cool huh?
