W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1997

Now available: extensive HTTP proxy traces from DECWRL

From: Jeffrey Mogul <mogul@pa.dec.com>
Date: Fri, 28 Feb 97 18:50:50 PST
Message-Id: <9703010250.AA28348@acetes.pa.dec.com>
To: http-wg@cuckoo.hpl.hp.com
During the summer of 1996, my intern (Tom Kroeger, of UC Santa Cruz) at
Digital's Western Research Lab, working together with an intern at
Digital's Network Systems Lab (Carlos Maltzahn, U of Colorado) obtained
lengthy traces from Digital Equipment Corporation's main HTTP proxies.
Because of Digital's firewall, a majority of the external HTTP
references from our corporate network flow through these proxies.

These traces cover the period between 29 August 1996 and 22 September
96.  This is a total of approximately 24,477,674 references.  On
mid-week days during this period, we usually saw about 1.3 million
references per day.

The traces are sanitized, so that it is simple to tell whether two
references are to the same server, URL, query, or from the same client
... but it should not be possible to discover the actual identity of
any host or URL in these traces.

After some false steps (including the beta release of some buggy
traces, and some procrastination on my part), I'm pleased announce the
availability of these proxy traces to a larger audience.

The main page for these traces is
    ftp://ftp.digital.com/pub/DEC/traces/proxy/webtraces.html
Users are cautioned to read the conditions attached to the traces,
and especially to note that we have already detected (and fixed)
one serious bug with the traces ... so they are not guaranteed to
be correct.

Please read the documentation there before sending any email
regarding these traces.

At some point in the near future, a link will be made from the
Internet Traffic Archive to these traces.  Stay tuned.

-Jeff
Received on Friday, 28 February 1997 19:01:04 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:30 EDT