| Cable Demands Accurate National Speed Tests |
| Monday, 01 March 2010 17:01 |
Cable visited the FCC and suggested the U.S. build a system like Britain's OFCOM/SamKnows. The U.K. study that found typical cable speeds within 20% of the advertised speed. The broadband planners stated "Actual median speeds lag advertised by ~50%." That's been widely quoted but seemed unlikelty to me based on what I know about U.S. networks. I've seen actual logs of major networks. The problem turned out to be bad data from Comscore. The carriers shouldn't have been surprised people have this wrong. They have been refusing for a decade to give an honest answer on speeds. They also fell about 10% in the evening, an indication that Virgin was overselling. With DOCSIS 3.0 across Virgin today,repeating the testing should show the problem almost elimanated. The big U.S. cablecos have been more aggressive than Virgin at node-splitting and backhaul, so they should be well under the 20% discrepancu. My Time Warner cable modem is solid at close to 10 meg down and 600K up, without throttling, despite the very heavy volume we use for jennie's video. Well-maintained cable networks should have minimal congestion problems, especially as they resolve the p2p upgrade issues. http://www.ofcom.org.uk/research/telecoms/reports/broadband_speeds/broadband_speeds/broadbandspeeds.pdf DSL "Up to 8 megabit"delivered less than half that speed. The best DSL results came from Carphone Warehouse and Sky, which have unbundled abnd developed their own backhaul network in most of the country. They had almost no evidence of congestion problems. Conversely, the ISPs reliant on BT Wholesale/Openreach had obvious congestion problems in the evening. This corresponds to my previous report that BT based networks - including BT Retail and most small ISPs - are the primary problem because the wholesale divisions price too high. Tony Werner of Comcast and his peers shouldn't get too cocky because not all the problems are resolved. At GigaOm's San Francisco event, nearly a third of the Comcast customers in the room had speed complaints about Comcast. This was a technically sophisticated audience, so they probably could tell whether the problem was Comcast or elsewhere. Clearly, some areas remain underprovisioned even if the overall statistics are looking good. Jitter, latency, and packet loss measured in the OFCOM study was minor except for gamers. "The average performance of web browsing, latency, packet loss, DNS Domain Name Service) resolution and failure rates and jitter, for all access technologies at all headline speeds, were sufficient to have no significant detrimental impact on the overall consumer experience of using most internet applications." A lot of folks like to prove their expertise in broadband by pointing to possible problems of jitter/latency, but they very rarely affect users of streaming video, VOIP, and even ordinary videoconferencing. I hope Comscore refunded the FCC any fees paid.
Here's the NCTA filing about their visit. Dear Ms. Dortch: On February 25, 2010, David Don and Jason Livingood of Comcast, Terri Natoli of Time Warner Cable, Grace Koh of Cox Communications, and Jim Partridge and the undersigned of the National Cable & Telecommunications Association (collectively the “Cable Representatives”) met with the following FCC staff members to discuss issues related to measurement and disclosure of broadband performance: Joel Gurin, Chief, Consumer and Governmental Affairs Bureau Julie Saulnier, Consumer and Governmental Affairs Bureau Patrick Webre, Consumer and Governmental Affairs Bureau Michael Jacobs, Consumer and Governmental Affairs Bureau Monica Desai, Wireless Telecommunications Bureau Jeff Tignor, Wireless Telecommunications Bureau Bill Dever, Wireline Competition Bureau James Miller, Office of Engineering and Technology Peter Bowen, Omnibus Broadband Initiative Byron Neal, Omnibus Broadband Initiative Consistent with comments filed by NCTA and the individual companies in the abovereferenced dockets, the Cable Representatives explained that cable operators provide customers with a significant amount of information regarding their broadband services and that they continually strive to improve the quality and quantity of communications to better meet customer needs. We also explained that the cable industry is committed to working with the Commission and with other parties to develop additional measurement, disclosure, and educational tools that could benefit consumers. During the discussion, we addressed the following topics: Ms. Marlene H. Dortch February 26, 2010 Page 2 Measurement of broadband performance. The Cable Representatives explained that online speed tests potentially have some value if they are well designed, but that a poorly designed speed test provides no useful information to consumers. The Cable Representatives expressed general support for the approach taken by Ofcom in the United Kingdom. A number of cable operators have held discussions with SamKnows, the contractor used by Ofcom, and the Ofcom/SamKnows approach is a useful reference point for the development of a common set of metrics for U.S. broadband providers. The purported gap between “actual” and “advertised” broadband speeds. The Cable Representatives explained that a variety of factors affect broadband performance, many of which are beyond the control of the broadband provider. In addition, we raised questions regarding the methodology for quantifying this gap that has been cited in presentations by the OBI team. Specifically, in the absence of publicly available information describing the methodology, we identified potential concerns in the manner in which the speeds were measured and in the assumptions that were made. We encouraged the OBI team to make sure that the description of this gap in the Broadband Plan, as well as any recommendations for closing it, clearly address the multiple other factors that contributed to the results, separate and apart from the broadband network provider. Consumer education. The Cable Representatives encouraged the Commission staff to consider developing programs that would educate consumers about the variety of factors that affect their Internet experience. In addition to understanding more about the capabilities and performance of the services provided by broadband Internet access providers, we explained that consumers also would benefit from additional information regarding their home computer equipment and networks, as well as information regarding the performance of web sites and transport networks. Please do not hesitate to call should you have any questions regarding this filing. Respectfully submitted, /s/ Steven F. Morris Steven F. Morris cc: J. Gurin J. Saulnier P. Webre M. Jacobs M. Desai J. Tignor B. Dever J. Miller P. Bowen B. Neal http://bit.ly/cYu6Uv |
Britain's OFCOM hired SamKnowsand GfK NOP Ltd, to put monitoring devices at 1,600 carefully selected homes. Overall, the average speed was 4.1Mbit/s or 57% of the average advertised headline speed. They fell to around 3.7 megabits in the evening peak.Cable did much better, with average speeds over 8 megabits on a line advertised as 10 megabits.