Way back at the dawn of the Voice over IP era, IP mavens used to dismiss voice as "just another application on the network." Then they started running that application on the network and discovered it was anything but "just another" application. It took several years to get voice quality on IP telephony networks anywhere close to the quality that end users had come to expect from TDM systems.The catch is that even the current voice over IP environment is only a piece of the convergence picture. Most enterprises still run VOIP exclusively within sites/campuses, and connect these "islands" of VOIP over private networks.
The goal is to run VOIP end to end, and the method for doing so is SIP trunking. This is also perceived to be a less expensive option, with the systems integrators at Dimension Data estimating that they save 40-70% on the WAN cost of their IP telephony by transitioning from PRIs to SIP trunks. And nowadays, because everyone is lunging at anything that promises cost savings, SIP trunking is *the* hot topic.
All of this you knew, because we've been talking about it pretty much nonstop on No Jitter and in our various VoiceCon outlets for the better part of this year. But the more we talk about SIP trunks, the more it sinks in, just how big a commitment SIP trunking is, how revolutionary it could be as the foundation of a true UC enterprise deployment-and how much work it'll take to make SIP trunks really deliver the quality of service we've come to expect. In short, it's the "application on the network" problem taken to the next level- the WAN level.
I bring this up because longtime friend and veteran analyst Tom Nolle of CIMI Corp. recently mentioned a statistic, in passing, that blew me away. Tom wrote to me that, "Worldwide, packet voice generates 8.3 times the number of support incidents as circuit voice." When I asked Tom for a bit more specifics, he pointed me to his company's surveys of global carriers, which showed the aforementioned average of 8.3 times more support incidents for all VOIP services, though SIP trunking services specifically showed a slightly smaller rate-7.5 times the number of incidents.
Tom's data showed these figures declining from 2003 to 2009, though they weren't falling nearly fast enough; in 2003, the overall incident rate was 8.8 times higher for VOIP services overall, falling to this year's 8.3 times. Since SIP trunking services are newer, Tom's data only goes back to 2007, when he found the identical rate of 7.5 times more incidents associated with SIP trunking services than circuit-switched services.
As we go forward with our coverage of SIP trunking, we'll continue to look at the potential savings, but what's become clear is that this is very likely not something you can just go out and order to today, then sit back and watch the savings pile up. You'll have to wrestle with issues of service availability, inter-carrier and carrier-to-enterprise interoperability, quality of service, security, and lots more.
And of course, just as we learned in the LAN/campus that the first step is a network assessment, that's very clearly going to also be the prerequisite for your SIP trunk planning. Except that your network assessment will be much more complex, because you'll be examining not just CPE, as you did with the LAN, but service provider contracts and installed services.
The upside is that you may be able to get some quick wins just by doing the WAN assessment; you may find unused circuits you can de-commission or contracts you can re-negotiate. But for a large enterprise, the SIP trunking road may well be a long one.
