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

RE: New document on "Simple hit-metering for HTTP"

From: Paul Leach <paulle@microsoft.com>
Date: Fri, 16 Aug 1996 16:54:33 -0700
Message-Id: <c=US%a=_%p=msft%l=RED-77-MSG-960816235433Z-42939@tide21.microsoft.com>
To: "'hallam@ai.mit.edu'" <hallam@ai.mit.edu>
Cc: 'http-wg' <http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com>
X-Mailing-List: <http-wg@cuckoo.hpl.hp.com> archive/latest/1385
At the most gross and simplistic way of measuring complexity (weight),
your proposal was three I-Ds, totalling 26 pages; ours is one draft of
11 pages -- factoring out bolierplate, that's about 1/3 as much real

The minimum implementation is not much more than to attach a counter to
entries in the cache and report them every time the entry is
revalidated. Adding an implementation of max-uses to the minimum
implementation adds a new way of controlling cache expiration that may
be useful in contexts other than hot counting. The next level of
implementation is to initiate an asynchronous HEAD when purging an
entry. If you're really masochistic, you can defer sending the HEAD
hoping to batch a bunch of them together -- but that's completely
optional (and to my mind of dubious value).

>From: 	hallam@ai.mit.edu[SMTP:hallam@ai.mit.edu]
>Sent: 	Friday, August 16, 1996 2:17 PM
>To: 	cuckoo.hpl.hp.com@http-wg.uucp
>Cc: 	hallam@ai.mit.edu
>Subject: 	Re: New document on "Simple hit-metering for HTTP" 
>With the latest suggested addition to the "simple" hit metering
>draft it has become as complex as my original proposal. The
>only difference being that in my proxy notification draft I 
>was also considering a possible generalisation of the
>mechanism to cover other types of notification such as updates.
>Do people really believe that using the HEAD (or whatever) method
>to communicate hit counts is really much simpler than the server
>periodically requesting log files? I originally started with a 
>scheme very close to the "simple" proposal. I had to expand the
>scheme after talking to people from Nielssen, Gallup and co.
>In the notification draft defered loading of logfiles was handled
>via a 3 byte, uuencoded string of a 24 bit mask corresponding
>to prefered download periods (in GMT).
>Note that on the compresssion side log file exchange is a lot 
>better than "simple" hit count. Each logfile entry is a lot more 
>compact than a simulated hit.
>	Phill
Received on Friday, 16 August 1996 16:56:00 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:40:17 UTC