On the whole, Internet service providers in the U.S. performed about as well in 2013 as they did in 2012 – largely hitting the same speed and consistency benchmarks. That’s one of the conclusions of the latest FCC report on the performance of consumer-grade fixed broadband services. Diving into the detail, though, shows that DSL-based service is falling further behind the performance levels achieved by cable and fiber technologies.
The FCC puts boxes inside the homes of volunteers across the U.S., who subscribe to a variety of service tiers from major ISPs, and runs standardised tests that mimic typical uses – web browsing and VoIP, for example – and measure peak and sustained speeds, data consumption and other performance related metrics. It then runs a series of calculations that show, among other things, how well ISPs are doing in actually delivering on advertising promises. The published results are national – you won’t be able to find out how your provider does locally.
Comparing sustained upload and download performance against advertised speed shows that Comcast, Cox, Mediacom, Verizon Fiber and ViaSat deliver sustained performance during peak periods that meets or exceeds the levels they sell to subscribers. On the other hand, Verizon DSL, CenturyLink, Frontier DSL and Windstream couldn’t even hit the 90% level. The rest fell somewhere in between.
Other tests showed similar results: cable and fiber are faster and more consistent than DSL. So, as far as the testing went, is satellite, but the one provider measured – ViaSat – was left out of some of the published results, in particular regarding the actual data consumed by its customers. More on that later.
Click for a bigger version, courtesy of the CPUC California broadband availability map.
Verizon says it’s invested more than $500 million in upgrading its broadband infrastructure in southern California and, in contrast to AT&T, it seems to be putting its money into wireline systems, particularly its FiOS fiber-to-the-home offerings. But the company is also making it clear that regulated copper plant belongs to past, and plans for replacing it with unregulated, fiber based Internet protocol service are moving ahead in California and elsewhere.
In its press release the company seems to be claiming to have switched more than 500,000 Californian telephone customers from copper to fiber. I say “seems” because the wording is vague enough to leave plenty of wiggle room. But there’s no doubting Verizon’s intent…
Nationwide, Verizon now serves more landline customers on its all-fiber infrastructure than on its legacy copper network.
For the people and communities that have FiOS upgrades already – 1.4 million households in California, according to Verizon – or are on the road map to get it – it’s a great thing. But there’s two ways to raise the percentage of FTTH subscribers: either build more fiber infrastructure or chop off less profitable DSL service and leave customers dependent on costly mobile data. Verizon is doing both.
For now, the company says its not going to add new markets to its fiber map, just build out what it’s already started. Its systems in California are a patchwork, mostly made up of former GTE territories, where the copper infrastructure was not outstanding to begin with. That leaves future broadband options in doubt for some rural areas of the San Joaquin Valley, the fringes of Silicon Valley and parts of the northern and central coast, along with big swathes of southern California.
Slow residential connections keep DSL speeds down, while cable’s problems are further back in the network. Researchers for the U.S. National Institute of Standards and Technology dug deep into data collected in 2011 by the FCC as part of its Measuring Broadband America program.
…a significant amount of congestion, especially for cable connections, occurs deeper in the network, perhaps, in the “middle mile”…or even farther, where the ISP connects to the “public Internet”. This is somewhat contrary to the popular belief that the edge is more congested than the core.
The “edge”, or last mile, is defined as the final outdoor segment that connects a home to the first electronic box in a provider’s network, for example a cable node in the neighborhood or a DSL port at a telephone exchange.
One way to read the report is that telcos have better core network infrastructure than cable companies, but where there’s the greatest difference between the two – ageing copper wires versus coaxial cable, DOCSIS versus DSL variants – cable comes out ahead. As the authors warn, though…
It is possible that because our method for detecting tight initial segments underestimates their prevalence, they could still be the dominant cause of recurrent congestion while remaining undetected.
Replacing copper lines over the last mile is an all or nothing proposition; it has to be done all at once in a given area, for subscribers and non-customers alike. Upgrading the core network, though, can be done incrementally as traffic and revenue builds. With the caveat that there’s more research to be done, it appears that telephone companies have a more difficult and costly job to do if they want to deliver faster and better broadband service.
DSL broadband provided connections on average delivering download speeds above 80% of the assigned speed tier more than 80% of the time. By contrast, a significant fraction of cable connections received less than 80% of their assigned speed tier more than 20% of the time. One must keep in mind that cable connections typically have higher download speed tiers than DSL connections…the average download speed tier for DSL connections was 5.4 Mbps vs. 13.5 Mbps for cable connections.
In California at least, phone companies tend to be more granular in the broadband availability data they provide to regulators than cable companies. Typically, data for any given city will show a range of advertised DSL speeds across different neighborhoods, while cable companies generally claim to be able to deliver their top tier of service everywhere.
Estimating the actual speeds delivered to customers requires applying a discount to those claims. The NIST data suggests that it is appropriate to apply a steeper discount to cable claims, as I have done in several analyses of the California data. Even so, the resulting estimates still tend to give cable the edge, which is also in line with the NIST findings.
DSL customers are more likely to get the service they think they’re paying for, but with faster speeds, cable customers might not notice the difference.