SM: Let me ask you a couple of detailed questions on that framework. You have servers inside of various ISPs. The intelligence in that server is your technology and that is how you route traffic. Is that fair to say?
TL: At a very high level that is correct. There is a lot of detail and things we do that go into making that happen. At a very high level, we are not an ISP so we do not route traffic in the traditional sense. How we move packets from one server to another and how we direct end users to get to a machine that is very close to them, how we handle load balancing and fail over and what functionality we get into servers near end users … these are the key problems we solve. Originally it was static content but now full applications can run there. That is what makes us unique.
SM: It has been almost 10 years now that you have been in business, and of course the volume of traffic and the level of richness in the web has expanded significantly. So has the level of technology understanding and capacity inside of these ISPs and telecom operators. How do they view you today?
TL: The telecom operators and ISPs view us as a very valued partner. There is along list of ISPs who want Akamai servers in their network. That is because we save them money, and we improve their performance to their end users. When our servers come in, they arrive with the 20,000 most popular domains pre-loaded. We give their users a faster experience.
We help them in times when there are internet events. If there are attacks, or there is a big media event such as a big software download or a big virus patch, that could swamp out the ISP if it does not have Akamai servers. ISPs love having Akamai servers. It is a very symbiotic relationship, and beneficial to both parties. We want to be there because we get the greatest performance and scalability, and they want us there because of what we do for them.
That has not really changed, in fact, it has gotten stronger as we have grown and we have more of the web on our servers. We have more value for our ISP partners than ever today.
You talk about how the Internet has scaled. The Internet has grown in capacity but in disproportionate ways. In the last mile you find the edge ISPs which is where we have the majority of our deployment. The content delivery capacity has gone up by a factor of 50 to end users. I had a 256K modem, I now have a 15 Meg service and my father has a 40 Meg service. You have a big pipe going into his home. On average, over the last five years, you have had a factor of 50 increase in the last mile of bandwidth.
The first mile bandwidth, where the data centers are located at the core of the Internet, has grown by a factor of 20. You think about 5 years ago, a typical customer might have a T3, 45 MB for the bandwidth. Today that same customer has a GiG-E connectivity. More or less a factor of 20 there and it is not keeping up with the last mile.
The big problem with the internet has been the middle mile. The peering points between the big networks and the smaller local networks, the big hauls, are the main problem area. That area has, more or less, grown by a factor of 6. The reason is that money is flowing into the internet in the first mile, because you pay for bandwidth when you use it. Money flows in at the last. As a customer, I pay a lot less today for a high speed connection than I used to for a 256K connection. As a data center, you pay a lot less for a GiG-E connection than you did for a T3 connection 5 or so years ago. The money only comes in at the beginning and the end.