- From: Jeffrey Mogul <mogul@pa.dec.com>
- Date: Fri, 16 Dec 94 13:52:35 PST
- To: Simon E Spero <ses@tipper.oit.unc.edu>
- Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
One other thing that was discussed was the relative advantages of using
a session method vs. an ignorable header. It turns out that there is a
problem with using ignorable headers when proxies are used - if a proxy
which doesn't interpret the header is used to talk to a server which
does handle the header, the connection can become deadlocked (the end
server things that the proxy doesn't want it to drop the connection,
whilst the proxy is sitting there waiting for the connection to drop).
This is a good point. However, the situation is not actually a "deadlock",
since the end server will presumably time out the connection after a
while.
This could lead to some longish delays, of course, but if the end server
varies the idle timeout according to the observed number of reuses
of a connection (i.e., start with a short timeout, and increase it if
if the client is observed to reused the connection) then the situation
will probably converge nicely.
Even with a 1-second idle timer, my simulations show about half of
the TCP connections are avoided. I have not simulated an adaptive
timer, but I'll put that on my list of things to do. I would hazard
a guess that over a reasonable interval, this would do almost as well
as the optimal case.
-Jeff
Received on Friday, 16 December 1994 14:06:16 UTC