Brix Networks — developer of service-assurance solutions, including monitoring tools, for telecommunications carriers, cable MSOs, and Internet service providers — has released a questionable, decidedly non-scientific study suggesting that the quality of VoIP calls is deteriorating.
The company analyzed data it gathered from its site called TestYourVoip.com, which was designed to allow consumers to test the quality of VoIP services. In its study, published Monday and covered today on CNET Networks’ News.com site, Brix alleged that call quality had declined five percent in the past 18 months. With almost one million VoIP connections tested through its website, Brix determined that 20 percent of all calls suffered from what it deemed unacceptable quality, an increase from the 15 percent of calls that it categorized as unacceptable last year.
Kaynam Hedayat, chief technology officer for Brix, attributed the decline in voice quality to the increasing contention for bandwidth on IP networks among various services, including voice, video, music downloads, and interactive gaming. His reasoning is that consumer consumption of Internet bandwidth is outstripping the availability of broadband bandwidth.
“The network is ready for VoIP. But now that there are more services running over the same pipe, carriers need to differentiate packets and prioritize service.”
Well, not so fast. They are several problems with the Brix VoIP call-quality study, and I’ll try to address a few of them.
First of all, since Brix is sampling only self-directed queries to its TestYourVoip.com site, the sample methodology likely is skewed statistically. There’s a strong likelihood, in other words, that people already concerned about, or experiencing trouble with, their VoIP call quality would visit Brix’s site to corroborate their concerns.
Second, individual IP-based voice sessions do not consume excessive bandwidth. Voice sessions are sensitive to delay, but they are not bandwidth hogs, certainly nowhere near as voracious as streaming video or P2P file sharing.
Simply put, voice calls across IP networks don’t require prodigious bandwidth, though, yes, their quality could be affected by downstream contention for bandwidth against more demanding applications. Still, we’d have to possess a lot more data about access, metropolitan-area and core-backbone congestion before we could reasonably reach that conclusion. The Brix study provides no such data, and its mechanism for collecting data is more of an advertisement for its commercial services than an objective means of gathering relevant data.
Furthermore, we are inclined to wonder about Brix’s probable bias and about what motivated it to collect and publish this data.
From where, for instance, does Brix derive its revenues? If you examine its website, you’ll see that its customer references represent a veritable who’s who of the telecommunications and cable industries — including AT&T, Comcast, Verizon, Cox Communications, Telus, Cable & Wireless, Qwest, Sprint, Rogers Communications, Bell Canada, Telstra — plus notable Internet service providers such as EarthLink and AOL, which, of course, belongs to the not-so-peaceable kingdom of Time Warner.
There’s nothing like currying favor with your customer base, currently fighting net neutrality and campaigning for toll-based quality of service, to ensure that more business comes your way.
You can check out other reactions to the Brix pseudo-study at GigaOM and Techdirt. The comments at both sites are quite good and cover some of the same ground — and a whole lot more — that I addressed here.