Congestion on the Internet is real and in some cases very serious, according to FCC and academic studies measuring broadband speeds at various points in the U.S. Both studies found backups at some points connecting Internet service providers and the network's backbone -- the equivalent of on-ramps -- that affected quality of service, particularly for high-demand uses like Netflix video streaming.
"During our testing we noticed that at certain points through the test network we saw some very serious congestion," said a Federal Communications Commission official in a conference call with news media. "We aren't prepared to make conclusions right now, but we will be looking at how video service providers are affected by this congestion."
David Clark, a senior research scientist at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory, said Wednesday that his study's preliminary data shows congestion seems to be mostly sporadic and temporary, lasting two to three hours at a time. But in some cases, he told a congressional briefing, the slowdowns at some interconnection points have lasted as long as 18 hours a day over several months.
FCC Chairman Tom Wheeler said last week that his agency was looking into complaints by Netflix and backbone providers Cogent and Level 3 that ISPs were purposely allowing congestion, to pressure streaming video services into contracting with them for special treatment. The allegations appear against the backdrop of Wheeler's controversial proposal to allow ISPs to give some Internet traffic priority over other traffic, abandoning the FCC's previous doctrine of so-called Net neutrality.
A federal appeals court threw out the Net neutrality regulations in January. The FCC chairman then proposed replacement rules in May that critics maintain would curb freedom of speech on the Internet. Many of the biggest companies on the Internet -- including Google, Facebook, Amazon, Microsoft, Netflix, Dropbox and Yahoo -- oppose Wheeler's proposal to allow paid "fast lanes," saying it would threaten innovation.
Clark's ongoing congestion study is being conducted by MIT and the Cooperative Association for Internet Data Analysis at the University of California San Diego Supercomputer Center. The data is preliminary, and the study does not yet assign fault for the congestion, Clark said at the briefing with the Congressional Internet Caucus Advisory Committee.
He said the FCC's probe was warranted. "It may be appropriate for the FCC to clear its throat and say, 'What's going on here?' " Clark said. In some cases, "this has been going on for months."
Measuring Broadband America
The FCC's congestion observations were preliminary, based on raw data, and came incidental to its fourth annual Measuring Broadband America report, which looks at how actual service provided by broadband providers measures up to advertised claims.
That report found that in general, cable- or fiber-based ISPs deliver or exceed their advertised network speeds, with average download speeds running 97 percent of advertised speeds during peak usage hours.
Not so much for DSL-based providers, which face technological hurdles. DSL providers Qwest/Centurylink, Verizon, Frontier and Windstream all failed to provide at least 90 percent of advertised speeds during peak usage hours, the FCC found.