Fixing the Internet Traffic Jam, Part II: Vint Cerf

the thing that is missed in data rate specifications is whether the rate is continuous or peak. these kinds of ratings have to be considered on many types of equipment and i don't see any reason we should think the internet is any different.

if i buy 5 mbs service for browsing that is one thing: i only need a burst of data here and there, most of the time my connection is idle and others can make use of the channel. but if I want CONTINUOUS data flow that is a horse of another color.

the obvious thing to do is to set a maximum transfer limit in frames per unit of time as well as the rate. for example, for browsing i might like a 5 mb/sec line with a limit of (e.g.) 5m packets in any 15 min period. I've been on for an hour this morning, browsing, and have needed to receive 4,636 packets so far. But I've been drinking coffee, reading the baseball scores and talking to the wife.

My personal feeling is those needing CONTINUOUS rated data service will need to lease private connections; they will not be allowed on the community system cables.

But that Bell split didn't last enough. It was proven that small companies can't make it by themselves, the investments needed are great as well as the worldwide competition.

Not even a company with 52 million users can make it. Sprint/Nextel reported a loss of almost $30 billion and with no light at the end of the tunnel, but maybe WiMax overtakes the market at a surprising rate (and that's a very BIG maybe).

Much as I hate the idea of paying more for more bandwidth, I certainly see the rationale here. However I suggest that the level of performance I pay for should be the minimum level, not a fixed capacity. If at a given moment there is more network available, why not use it? The specified access rate should apply only when the network is busy.

I'm probably the type of user who would 'shoot crap' by paying for a lower level of performance in the hope that, most of the time, I would actually receive the higher level. But this model would self-adjust quite nicely to demand, and those willing to pay more could have the higher level guaranteed.

(reply to Mr. Roques)

Quite frankly, I don't know why the Cable TV company wants to be the phone company, nor do I understand why the phone company wants to light up my idiot box.  Neither decision seems to go with the inherent capabilities of those types of companies.  But they're certainly allowed to choose to do these things. 

With that in mind, one political solution I've often heard suggested is the forced separation between transit and content. This solution assumes an inherent conflict of interest between being in the business of delivering video entertainment (Cable TV) and owning the very transit lines that would link people to competitors.  

That forced breakup idea makes my Libertarian rash break out all over. It seems like a shift in the Bell Breakup magnitude.  But my own experience tells me that we can't count on network operators to behave themselves when their loyalties are split between allowing users to access the content, services, and destinations that they want and locking them in to xtuple-play packages and "monetizing" their surfing behavior.

Robb Topolski

I don't see any ISP following this business model. They are seeing how their revenues are diminishing because offering the access is not where the money is, but in the applications and services.

Just like baseball games, selling the ticket but then letting anyone sell food, drink, souvenirs and whatever they can come up with, would cut a huge chunk of their revenues.

Both solutions (bits or flow control) have some flaws, with bits: what will happen with retransmitted bits (because of congestion)? and with flow, it's very hard to guarantee certain amount, because no network is planned to handle the peak traffic.