mmol_6453's Journal: Queues in future versions of HTTP?

I'd read the recent article on a book about HTTP, and it ocurred to me that HTTP implementations could learn from P2P networks.

For example, Gnutella clients have queues where a client will connect periodically, and check its place in line. That way, a connection doesn't need to be maintained in order for a consumer to hold his place in line. This lets a popular server to transmit information in a true first-come-first-serve manner, rather than a manner where whoever retries.

This could seriously improve performance under times of heavy server load.

Of course, I'm no expert.

well, to some degree

[intermodal]

but i think it's more likely a feasable feature of FTP than of HTTP. I think for HTTP it would just turn into a bottleneck. Too many administrators (at the demand of managers lacking in technical knowledge) would set it too small, or rely on server gear inadequate for their real needs. The current way of doing things is one of the checks and balances of the internet...the more traffic you have, the more server/bandwith you need. See the parallels between rush-hour traffic and this system?

sorry for the