> From [email protected] Thu Aug 7 17:26:33 1997
> From: "Farley, Tim" <[email protected]>
> To: "'[email protected]'" <[email protected]>,
> "'TCP over Satellite'"
> <[email protected]>
> Subject: RE: Problematic Approaches
> Date: Thu, 7 Aug 1997 18:47:46 -0400
>
> >> The problem is that the client asks for a page, then has to receive
> >> it before it can be parsed and subsequent requests pipelined.
>
> Perhaps I wasn't clear.
>
> The client doesn't have to *completely* receive an HTML page to pipeline
> subsequent requests. It is certainly feasible to build a browser which
How big is this page?
If the window isn't big, and the HTML dribbles in, then this all works
for the sum of requests for a single page, but you're back where you
started for the 'user read time' gap.
If the window is big, then the entire file is in transit. By the time the
client gets the beginning of the file, the end is already in transit,
and the window is liable to be slamming shut.
> even hit the </HTML> tag of the initial GET. It might require a
> multi-threaded design, but it could be done.
It also requires:
kernel with very small timeslice
otherwise the application isn't awake fast enough
to eat the beginning of the file quickly enough
sockets with very small buffers (on some OS's)
some OS may delay waking an application waiting
for input until the socket is more than 2/3 full,
or the timeslice ends (as above)
> Depending on how large the initial page is versus the round trip time,
> this could in many cases keep the window from slamming completely shut.
> It would be interesting to prototype such a design and see how much of
> an effect it would have.
Given the fact that we've seen the slamming occur on
networks with as few as 4-8 packets in the RTT pipe,
I'm not optimistic that a satellite emulation will find better :-(
Joe
----------------------------------------------------------------------
Joe Touch - [email protected] http://www.isi.edu/~touch/
ISI / Project Leader, ATOMIC-2, LSAM http://www.isi.edu/atomic2/
USC / Research Assistant Prof. http://www.isi.edu/lsam/
This archive was generated by hypermail 2b29 : Mon Feb 14 2000 - 16:14:29 EST