We recently pushed out some firmware upgrades to one of our customer networks. Their This ADTRAN gem--VQM that I previously wrote about--provides ample call quality statistics and reporting on "routed IP/SIP" calls. My answer was an emphatic yes and the feature--Voice Quality Monitoring (VQM) is already embedded in the ADTRAN routers.
For our customer, all voice calls include interoffice intercom and fax calling between customer sites; there are no SIP or IP trunking services and BGP (Border Gateway Protocol) is deployed in this hosted MPLS network. There are three sites and we gathered data on at least 1,000 calls for each location and both configurations.
I've pondered the traffic and the IP-PBX first and second choice of protocols. I considered all traffic and whether or not it makes sense to route calls out using G.729a to use less bandwidth or to stick with G.711u first choice for higher chances of success to ensure voice quality. Then, I also considered redials because of potentially poor audio quality and the extra traffic burden it carries. So those are my musings and these are only two compression rates. Equipment configurations, network topologies and providers may yield different results.
When we initially setup the gateways between sites we intentionally made the first choice G.729a followed by G.711u. We did this to conserve bandwidth since the customer sites use secure web portals for a good deal of their business. Again, no known complaints. After some more pondering we changed all the gateways and switched first choice to G.711u followed by G.729a. No complaints followed. (See Table below)
So with MPLS services the question is the same as for SIP trunking providers--what percentage of calls are guaranteed to carry acceptable MOS scores at each compression scheme, or are these metrics simply not considered? My buddy Samir Kakkar over at ADTRAN reminded me that, "in a typical VoIP call, any end can request (for any reason) a change in the codec midstream during a call." Which configuration is best? Higher compression rates may mean more calls on the pipe gaining higher utilization while better quality may mean potentially fewer calls for help and not packing as many calls over the pipe. Yet still I find myself instinctively wanting it all (High utilization, better quality, less calls for help); is this old dog chasing his tail?
