> > Rewriting Keith's statement to say "Using one TCP stream per client means
> > each client gets at least a little bit of bandwidth in edgewise when there
> > is a hog", may better.
>
> But still wrong. A single fast stream on an ADSL link will completely lock
> out mostly idle streams. You would get far better results doing fair
> share multiplexing of one TCP stream in your compressing proxy.
>
Yes, exactly correct. Funny you should say that: I worked with a few
others on designing exactly that kind of encapsulation protocol (though
the allocation of bandwidth policy was left to the user of the protocol).
See http://www.w3.org/Protocols/MUX/WD-mux-980708.html (now a dead project).
For such a simple thing, it turned out to be more subtle than I originally
thought, to avoid potential deadlock.
- Jim
--
Jim Gettys
Cambridge Research Laboratory
HP Labs, Hewlett-Packard Company
[email protected]
lmpx.com only provides a reader for public news (NNTP) servers. It is not
affiliated with the servers or forums shown here and is not responsible for
the content of articles, which is written by their respective authors.