Monday, January 31, 2011

How Does Your Broadband Provider's Performance Rank

One of the great frustrations in getting broadband service is that what you're sold and what you pay for is not always equal to what you're being delivered. While the carriers are quick to promote download speeds, the actual bandwidth performance that you get is often tucked neatly under the umbrella of caveats like "speeds up-to". Meanwhile, using bandwidth management and oversubscription, they deliver connected bandwidth far below their published maximum values.

While most of us are aware of these carrier practices (or rather, we know that sometimes our broadband doesn't seem as fast, even if we don't know what the cause is), the sad truth is that we haven't really had access to tools that enable us to accurately measure there performance beyond basic speed test sites and anecdotal tales on bulletin boards like you find on DSLReports. As a consumer, this makes it hard for you to competitively shop. Add to that the hassles of switching, and you often find yourself at the mercy of a very limited market. But what if you knew which ISPs delivered the best, most reliable bandwidth?

Fallout in the war for Net Neutrality
Late last year you may have picked up on a story about a disagreement between Comcast and Level3 -- or Comcast and Netflix, depending upon how you look at it. Essentially, Comcast noticed that every evening their network traffic ramps up as many of their customers grab their computers, their iPads or their Netflix-enabled Blue Ray players and begin streaming movies from Netflix. Since Comcast would prefer to stream their own digital content that they can charge for, they decided that Level3/Netflix should be forced to pay additional charges to deliver all of that Netflix content. And another battle over the issues of Net Neutrality started.

Recently, Netflix fired back. In this post on the Netflix tech blog, Netflix Performance on Top ISP Networks, Netflix documents the streaming performance of many US and Canadian ISPs. With the volume of streaming media that Netflix has been serving, they have been able to learn a lot about network performance and sustained content throughput. According to Netflix, when they stream an HD video, the ideal amount of bandwidth needed for optimal quality would be 4800 Kbits/second (4.8 Mbits /second). Based on their graphs, none of the ISPs are delivering sustained throughput at that level.

Finding the Take-away
You can look at this post from a lot of different angles and find a lot of meaning. If you're looking to argue from the side of an ISP or a carrier, you'll find comments on the lack of differentiation between services (or potentially levels of service) in averaging provider performance. You'll also find infrastructure costing and service level analogies that point to why Netflix streaming is such a bandwidth hog and how that hurts other users on the ISPs network. And if you're looking for some clear examples that rebut those arguments, you can also find that in the comments.

To me, there are a couple of clear issues here. The first has to do with availability and demand. Part of what makes the infrastructure argument so frustrating is disconnect between the message being used to promote the product being sold and the actual business practices. Consider, in the past ten years, I have personally made five or six infrastructure upgrades to my own home network. I've upgraded from no ethernet through 10Mbit and 100Mbit to 1Gbit ethernet. I've added wireless and improved my wireless network twice. Data on my home network can move anywhere from 64Mbps to 1000Mbps. During that time, I've been through two DSL providers and one cable provider, three DSL modems and two cable modems. And while my cable bandwidth far outperforms my old DSL bandwidth, it ranges from 2Mbps to bursts of 15Mbps. And yet my entire home network infrastructure probably cost less than the cost of one year of access through either DSL or Cable.

As we envision the infrastructure that we need, the information superhighway (to bring back an archaic term), we have turned the responsibility for managing our roads to a private system of toll booth operators that profit simply from having you sign up to access their system of roads -- regardless of whether you are driving. One early commenter to the Netflix post made an analogy to airplanes and airports. Perhaps a more accurate analogy might be that you pay for access to an airline that promises to fly you anywhere you want for a flat monthly fee. At the same time, for every 100 seats in the plane, they have sold 1000 seats to people, then offer those seats on a first come, first serve basis.

If we really want to see our online world transform, we need to see our infrastructure transform. While the world is building out faster and faster networks, the US continues to be held captive to an infrastructure that is limited an industry focused farming profits from legacy systems. Like the entertainment industry with physical media versus digital content, they're trying to prop up a fading industry by stifling innovation and evolution. Unless we change our direction on network infrastructure and network management at a broader policy level, we will increasingly see access and content decisions shaped entirely by the profit interests of a handful of content-carrier monopolies ala Comcast-NBC.

Think back to the promised Google experiment delivering Gigabit broadband service and how many cities were jumping through hoops trying to be part of the trial. This is something that the country needs, that the people want. Unfortunately, we aren't going to get there when the debate over whether 1.5, 2.0, 3.0 or even 4.8 Mbps is good enough.

No comments: