There’s an interesting interview up at The Register:
So on January 1, 1983 when TCP/IP was deployed, it all worked fine. Primarily the net was used for email. Then there were more FTP sessions, and it began to melt down.
So people were writing a lot of papers in mid-1984 about what was then called “congestion collapse” Some of the design features of TCP windowing actually made congestion worse; so protocol engineers went to work. They made enhancements to TCP such as Exponential Backoff – another thing stolen directly from old Ethernet and Slow Start – where the initial window size is small. They re-engineered TCP to solve IP’s congestion problem.
Today, the internet is only stable to the extent people are using TCP over it. People also tend to miss that you can defeat TCP’s attempt to limit traffic over something less than congestion of the backbone if you simply have multiple instances of TCP.
Some guy with strong opinions.
Richard, you have argued here and elsewhere that the Internet should not be treated like the telephone network. I think you mean that the economic hypothesis that was used to justify common carrier regulation (and later the Computer Inquiries for DSL) does not apply to the Internet for some technical reasons. What are those technical reasons? (The strength of economics is that it applies across all industries. Its application generally is not limited because of specific, technical reasons.)
The economic hypothesis, as applied to the Internet, is that the last-mile wireline owners will use things like QoS to give their content an advantage in transmission over rival content providers, which will stifle competition and therefore innovation at the edges.
I know you probbaly do not believe in that theory because it has not come true yet. Or perhaps you doubt that it hurts social welfare even if it happens. Nevertheless, if we assume that wireline owners will exclude rival content providers, and that such exclusion will have a negative impact on social welfare, is there any way to prevent the wireline owners from excluding rival content providers without interfering with justifiable network management?
My point is technical, not economic.
The Internet is a packet network, so regulation should address its behavior as a packet switcher, not require it to behave in the first-in, first-out manner at every switching point. In order for a packet network to operate correctly, it has to prioritize packets such that network management packets have priority over user data packets and so forth.
Telephone networks allocate bandwidth according to a strict formula that doesn’t allow unused network resources to speed-up existing sessions. Packet networks allocate resources on demand, by their very nature. That isn’t to say that network resources have to be allocated on demand to anybody and everybody.
Some folks want IP to become the “converged network” that handles TV and Voice as well as web traffic and e-mail. That can’t happen without priority treatment of real-time traffic, and that means QoS.
Ok, but do you have an answer for the second question? One solution is price regulations (like we had in telephone) on various elements of broadband service (e.g. QoS). Even people who like price controls recognize that they are not ideal. Another is “no QoS fees,” but that could deter the wireline owners from providing QoS or, if they do offer it, the “no QoS fees” policy could lead to an overload in the QoS queues and render QoS ineffective. Is there some other solution that would not interfere with Internet management?
The only way I know to limit access to a scarce resource is to charge for it. You could establish a quota for QoS as part of every broadband bill, but that’s the same as charging for it, only you force everybody to pay for QoS whether they want it or not.
So you charge for it, one way or another.