A friend email me asking why ITU-T SG3 is not suitable to handle the issues of Internet Peering. He pointed out that while there are technology differences, the economics and policies should be fairly similar given they are both bilteral arrangement. I beg to differ.
Lets take a detour: When User 1 on carrier A (origination carrier) makes a phone call to User 2 on Carrier B (termination carrier), the usual financial arrangement is the A will pays a termination fee to the B. And in the case where A and B don’t have direct connection but are connected via Carrier C, then A will redirect its call to C (A pays C) and C will reroute the call to B (and then C will pay A B). This is how it works in a nutshell1.
Fairly simple and striaght forward because we are dealing with one and only one application: Voice2. On the Internet, the service providers only see IP packets. They don’t really know what application or content are involved3.
Why not do the same and have termination charges? Lets see: if User 1 on network A send an email to User 2 on network B, the IP packet will flow from A to B which means A pays B…Yes! That works!
Lets see another example: If User 1 on network A request to download a file (e.g. a music or movie) from User 2 on network B, common sense say that A should pays B. However, the IP packets actually flows from B to A (more then A to B)! Erm…doesn’t seem to work here.
Okay how about looking at who is making the “initiating connection”4? Suppose A initiate connection to B, then A pays B. And it would work for both the example above.
Erm, the problem is there is no such concept of “initiating connection” in IP packets. Sure, TCP has it but not all applications uses TCP and it is crazy to think you can capture all the TCP 3-way handshake sequences for accounting purposes. Beside, it will not work for FTP5 and I could cite many examples where it would fail.
No, the economics behind Internet Peering are very different. There are similarities but they are not the same. This is why I feel a new SG, with a different mindset is more suitable. But why ITU? Well, they have a century experience in dealing with difficult bilteral interconnection issues. They just need to bring the right stakeholders into the discussion.
(Okay, I can sense some of you already shaking your head with the mention of ITU… :-)
1 Of course, it is far more complicated. There are issues of circults, interconnection charges etc etc. And differences between terminations charges also creates other interesting business opportunity like call-back and refiling (fraud?) etc etc.
2 So what about SMS? Well, SMS was originally designed as a technical tool so the operators just connect to each without considering the business case. Thus, for a long time, each carrier just charge whatever they like but there are no financial settlements between them.
3 Okay, they can make an intelligent guess by looking at the port but that seldom yields useful information about the nature of the application or content realistically speaking.
4 The first time I see the phrase in an Internet context in the SG-US FTA section on DCMA, my jaw dropped. Obviously the other ISPs are also puzzled which leads me to conclude that whoever craft the phrase is either a bell-head or nut-head. But most likely, a lawyer :-)
5 When you do an FTP to a server, you are only establishing a command TCP session. When you request to download a file, the FTP server will initiate another TCP connection to you (unless you are using passive FTP).