The Right Questions: FCC Traffic Workshop December 8
Written by Dave Burstein   
Net Neutrality battles in D.C. are getting brutal, so the Bells are bringing out their best. Bill Smith will be on the hotspot, AT&T CEO Whitacre told the Senate they have essentially no congestion on their network. AT&T lobbyists and paid advocates are all over D.C. recently saying the opposite.

The extraordinary panel can shed light on many other crucial technical topics. They've asked us all to send in questions during the webcast http://www.openinternet.gov and I've put some together. Tom Sawanobori is the strategist for Verizon Wireless, where LTE looks to be one of the best networks in the world. Paul Liao is the new head of CableLabs, where they developed DOCSIS 3.0. That's rapidly bringing 50 megabits down to 90% of the U.S., with upstream soon to follow.  Cisco and two academics round out the panel.

kc claffy  has spent a decade analyzing Internet traffic at the University of California. She points out why this session is so important. "our scientific knowledge about the Internet is weak because researchers are typically not allowed access to any data on operational network infrastructure."

Ask her "What are the three or four most important points of data we need?"

Scott Jordan, also University of California, has just written "How to determine whether a traffic management practice is reasonable." It reads like an academic paper, but has sensible conclusions.

  • Traffic management is reasonable if the end users can decide what gets priority. If I want to prioritize my VOIP call over a movie download, that's fine. In particular, I think almost all of us would ask our ISPs to block worms and malware, DDOS, and the like.
  • Traffic management is likely unreasonable if the carrier decides which application or service gets priority. If it's unreasonably anti-competitive, causes undue harm to consumers, or unreasonably impairs free speech, that's not right.

Paul Sanchirico of Cisco runs most of their TV and set top applications. Verizon and AT&T are among his largest customers. Don't put him on the spot where if he tells the truth he'll have angry customers.

  • Cisco has perhaps the best projections of future Internet traffic and costs. Their visual networking index projects future Internet traffic continuing the next five years to grow at about the same rate as it has for the last six. Ask about how VNI is compiled so that we can judge whether it's likely to be accurate.
  • Cisco also has a roadmap of the likely equipment cost/bit of the next five years. I think the decline will be slightly faster, but Paul has much better data. Ask for the actual estimate of equipment cost per bit pricing the next five years. Will they expect 35%/year or 50%/year as one competitor does?
  • Blair Levin and others at the FCC are proposing open set tops, including some built into the TV (Tru2Way.) The cable companies may go along because they want to get rid of cablecards. That requires the long delayed downloadable content protection (DCAS.) I hear Cablevision has it working in the field, but there's nearly no information about what's coming when. What's the timeframe? Will it work with most future TVs (and Verizon FiOS) with an ethernet connection? Will content protection add about $5 or about $50 to the cost of the TV if it has a mass market?

Tom Sawanobori is strategist for Verizon LTE, scheduled to bring megabits to 92% of the U.S. in 2013. Bravo! Verizon estimates the speeds will be 5 to 12 megabits down, half that up, but LTE is a shared system. If many people wanted to watch HDTV over the net, the system couldn't handle the load. So the speed depends on traffic estimates. Verizon has been working hard on the models.Ivan Seidenberg has said they are looking closely at femtocells. Femtocells or WiFi phones move traffic off the wireless network. That frees spectrum, a key resource.

  • What level of traffic per home would stress the 5-12 megabit system?
  • If some people used LTE as their primary home connection, about how many streams pf 3 megabit TV can a tower handle?
  • How many homes might that tower cover?
  • On femtos or WiFi, are the estimates that widespread use can expand spectrum 20-40% realistic?
  • How will Verizon decide on Femtos vs WiFi?

Paul Liao is the new head of CableLabs. Until a few months ago he was CTO of Panasonic, a completely different field. Cable networks are all shared, and there was a problem with congested upstreams that was tricky. Comcast has dramatically reduced the upstream congestion with a modest hardware upgrades,and DOCSIS 3.0 has over ten times the upstream bandwidth. Over 60% of the U.S. will have 3.0 available by the end of next year, and the CITI/Columbia report to the broadband plan expects that to go over 90%. That's advertised as 50 megabits down, with upstream now moving from trials to deployment as well. By early in the plan period, most of the country will have 3.0 so that's what I'd ask my questions about.

  • What percentage of the time will 50 advertised megabit customers actually get 50 megabits? How often will they see 20% less? 50% less?
  • In France and other countries, 50 or 100 megabit is the standard service and of course we'd like similar for U.S. homes. Again, what percentage of the time will 50 be 50, down 20% or down 50%.
  • At some point ?2014 ?2017 the current 160/120 DOCSIS will be stressed. In theory, DOCSIS can go to a (shared) gigabit, but what's a practical expectation?