If you are considering becoming a 1M/1M premium member and would like to join our mailing list to receive ongoing information, please sign up here.

Subscribe to our Feed

Speeding Up the Internet: Algorithms Guru and Akamai Founder Tom Leighton (Part 4)

Posted on Saturday, Oct 20th 2007

SM: You are saying that the middle does not have a business model supporting it. The first mile and last mile have business models, but the middle does not.

TL: Right. The middle mile is where you get into a lot of problems with the Internet. It is because there is no money flowing in there. You have the little ISPs paying big ISPs for transit. The big ISPs get into fights whether or not they should be paying each other. You see the incidents where level 3 ISPs will fight over deep peering. Because they are arguing over who should pay whom and what should be paid. They are all under financial pressure, and the result is that those connections are not upgraded as fast.

A little ISP, in order to save money, is willing to be congested for an hour a day during peak traffic. You have this bottleneck which is sort of the big middle mile, the long haul, going from network to network and through peering points which are congested, and the Internet routing protocols ignore congestions. There is not an ability to route around congestion areas using normal protocols.

That is why Akamai places its servers in the last mile. That gets us beyond the bottleneck. All of our competitors place their servers in a few core data centers. It is a lot easier to think of managing 20 locations, but it does not work nearly as well and inherently that will limit the scalability and it limits the performance. That is why Akamai is so much faster and has so much better quality in the delivery of content like video to end users.

SM: That is a good transition into the question about what the impact of video is on the Internet. It is a different type of traffic which has really ballooned in the last few years. What does that do to Akamai, what does it do to the Internet, and what are we going to see?

TL: Video increases our traffic which is good for our business. It increases the burden on the Internet which is bad for the internet. The good news and the bad news is that we are just at the tip of the iceberg on video.

Downloading a video is about 1000 times the traffic of an iTunes song. If we really get TV over IP into the home, or DVD or HD quality video into the home over IP, that is going to explode the traffic on the internet. We have not even seen the beginning of the impact on the Internet of that yet.

We are supporting HD video with our product, but it is small scale relatively speaking to audio or static objects. The Internet is going to be put under a lot of stress. The content delivery network and services which are organized at the core and the data centers are going to hit a wall. They will buy bigger and bigger pipes at the data centers, but the peering points will remains congested.

Akamai is in a great position there. As we think about growing our capacity over the next few years we were thinking on the order of a couple hundred terabytes per second to support a large audience for TV or better quality of video. The only way you can do that is to be serving that content locally. You are never going to get that capacity out of data centers.

This segment is part 4 in the series : Speeding Up the Internet: Algorithms Guru and Akamai Founder Tom Leighton
1 2 3 4 5 6 7 8 9 10

Hacker News
() Comments

Featured Videos


I agree that video broadcasting and transmiting has become quit popular and effective last year.

It’s popular on Youtube and it remains question to me how Youtube deals with such burden of traffic every day when millions of people watch there video clips at the same time.

Video broadcasting became very effective in online businesses with increasing convertions and customers as well as growing ROI.

Boris C. Saturday, October 20, 2007 at 3:27 PM PT

I agree with Boris. I think that video on the Internet is the next untapped resource of capital…if…someone can figure out how to make money LEGALLY by stealing other people’s material. Becausr that’s what more people are searching for.

Big Daddy Bootstrapper Saturday, October 20, 2007 at 7:16 PM PT

“we were thinking on the order of a couple hundred terabytes per second to support a large audience for TV or better quality of video.”

This is a ridiculous statement. It’s pretty sad that someone who passes himself off as a scientist would make a comment like this. Why? Well let’s look at the numbers. 200 terabytes per second is equal to 1600 teraBITs per second since there are 8 bits in a byte. If you’re streaming HD video encoded at 1mbps (which is great quality), in order to generate 1600 terabits/sec of traffic you’d have to have 1.6 billion people watching at the same time.

Come on!

Brett Sunday, October 21, 2007 at 8:24 PM PT

Brett: 1 Mbps isn’t great quality… at least not in the near future…

Suppose you want to watch a video with the following characteristics : 2048×1024@72 fps and H.264 video compression… you’ll need more than 100 Mbps in this case
And 10 million people isn’t that huge if you assume that people will be watching TV over IP (which I don’t believe but it’s reasonable to consider this case)

Rafael Monday, October 22, 2007 at 4:49 AM PT


You must be smoking some good stuff. There’s no reason to run video on the internet at 72fps when you don’t need to and 30fps looks perfectly good.

Also this idea of the “middle mile” is a fabrication. There’s no “middle”. The major ISPs like Comcast and Cox buy transit from backbone providers. ISPs that are also carriers, like AT&T and Verizon, connect to their own backbones and then to other backbone networks. The “middle mile” has no reality to it.

Brett Monday, October 22, 2007 at 10:53 PM PT