W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1996

[johnh@isi.edu: document about P-HTTP/TCP interactions]

From: Larry Masinter <masinter@parc.xerox.com>
Date: Sun, 18 Aug 1996 00:59:04 PDT
To: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Message-Id: <96Aug18.005904pdt."2757"@golden.parc.xerox.com>
I think this is interesting data for those implementing HTTP/1.1....

------- Start of forwarded message -------
X-Url: <http://www.isi.edu/~johnh/>
To: Larry Masinter <masinter@parc.xerox.com>
Subject: document about P-HTTP/TCP interactions
Date: Tue, 13 Aug 1996 10:31:58 PDT
From: John Heidemann <johnh@isi.edu>


I've been looking at some performance interactions between currently
P-HTTP and TCP implementations and have prepared a brief web page
summarizing my findings:


    title: Performance Interactions Between P-HTTP and TCP Implementations
    url: <http://www.isi.edu/lsam/publications/phttp_tcp_interactions/>

    Abstract:

    This document summarizes several performance problems resulting from
    interactions between the Apache-1.1b4 implementation of
    persistent-HTTP (P-HTTP) and BSD-based implementations of TCP. Two of
    these problems tie P-HTTP performance to TCP delayed-acknowledgments,
    thus adding up to 200ms to each P-HTTP transaction. A third results in
    multiple slow-starts per TCP connection. We discuss each problem and
    summarize solutions.  Unresolved, these problems result in P-HTTP
    transactions which are 10 times slower than standard HTTP and 20 times
    slower than potential P-HTTP for machines connected by a 10 Mbps
    Ethernet. After fixing these problems we observe that P-HTTP
    successfully performs better than HTTP on a local Ethernet.


I have forwarded a pointer to this document to the Apache team.
Because both the techniques used in Apache and BSD TCP implementations
are common, I think that this document will be of interest to the
http-wg mailing list and others doing P-HTTP development.

I would be interested in any feedback you or the working group would
have about this work.

Thanks,
   -John Heidemann
    USC/ISI

------- End of forwarded message -------
Received on Sunday, 18 August 1996 01:01:10 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:08 EDT