Akamai had to ration their customers' bandwidth and turn away some; the North American content networks were close to saturation, with heavy traffic in Europe and Asia as well.While the content delivery market as a whole has historically used marketing phrases to imply that they have "unlimited capacity", that's simply not reality. This is the way the Internet works and it puts things into perspective when many want to say that the Internet will drive broadcast TV out of business, that HD quality video will be delivered to millions anytime soon or that online video will get the same sized audiences that TV broadcasts get. The Internet has major limitations for delivering video to large-scale audiences and while many want to complain about the problem being at the ISP end, or with the end users broadband speed, the bottleneck is also on the CDNs end.
Template Tools
| Obama Proves 2M to 10M Net Viewers Practical. 7-8M Watched |
| Monday, 16 February 2009 20:14 |
|
Obama's inauguration reached a peak of 7-8M simultaneous viewers, Dan Rayburn calculates, including 3.8M at Akamai alone. Adding a few assumptions to Dan's figure, I'd estimate CDNs can currently deliver 20M streams at low bandwidth and perhaps 2M at digital cable quality. "TV" can range from the 100-200K of that day's inauguration to 400K YouTube lousy to 6 megabit good HD, so there's no single figure possible for how many viewers can be supported at peak. Working with some estimates of Akamai's market share, etc, content delivery capacity for video over the net in the West is currently perhaps 4 terabits, excluding Google/YouTube. Add Google and Asian networks and you easily triple that figure, but I don't have hard data. The 4 terabits on the Western market at different speeds yields:
Those numbers are for fairly demanding programs with action, using 2008 state of the art compression (h.264 or similar.) The SD and HD figures are based on the actual compression used on U.S. IPTV and cable. Meet The Press style talking heads need far less bandwidth. Compression gets more efficient every year, although experts tell me the progress these days is much slower than a few years ago. A conservative estimate is you'll need about a third less bandwidth in 4-5 years, although at least one senior engineer expects better. The limit on video delivery is the not the total capacity of the Internet, which by all accounts can grow an order of magnitude fairly quickly. The content delivery networks have a limited number of servers and router, and purchase a limited amount of bandwidth from the Internet core providers. There are very few shortages of fiber that could be lit or manufacturing capacity of servers and routers, so with enough time the video capacity can expand to meet all but extreme peak demands. I'd guess those numbers could easily go up 5 or 10 times in three years or so if they had the customers, and possibly more. Level 3 has an extraordinary amuont of dark fiber, and switch/router companies like Cisco and Juniper continue leapfrogging each other with capacity. Mark Cuban came to a similar conclusion, that two 500K TV streams would stretch today's capabilities. If Mark is thinking 2 megabit digital quality, he's on track with this real world example. Akamai, the largest, possibly could handle Mark's two programs, but they wouldn't have much reserve. Including other programming, Akamai delivered 7.7M streams at peak. At 12:15 p.m. ET, Akamai was running more than 2 terabits per second. That included 800 gigabits of Flash streaming, other video formats, and Akamai's ordinary traffic of Microsoft updates, etc. Akamai's customers include CNET, NBC, Thomson Reuters, USA Today, Yahoo, The New York Times, Ustream, Viacom, and the Wall Street Journal. (Wade Roush) http://blog.streamingmedia.com/the_business_of_online_vi/2009/01/akamai-and-numbers.html |