Re: WWW performance
marca@ncsa.uiuc.edu (Marc Andreessen)
Date: Sat, 3 Jul 93 00:21:50 -0500
From: marca@ncsa.uiuc.edu (Marc Andreessen)
Message-id: <9307030521.AA03438@wintermute.ncsa.uiuc.edu>
To: Rob Raisch <raisch@ora.com>
Cc: Tony Sanders <sanders@bsdi.com>, www-talk@nxoc01.cern.ch
Subject: Re: WWW performance
In-reply-to: <Pine.3.03.9307021558.H14245-9100000@amber.ora.com>
References: <9307021847.AA16795@austin.BSDI.COM>
<Pine.3.03.9307021558.H14245-9100000@amber.ora.com>
X-Md4-Signature: 779349e685ee03e3c3d222ddea51c93a
Status: RO
Rob Raisch writes:
> Tony, I guess I'm just being dense. What value is there in supporting
>
> <A HREF=... FREQ=44>
> ?
>
> Is this meant to be the number of times a particular link has been
> activated by a client? I'm not sure what value there is here which is not
> already kept in server logs. I'd suppose that this same info could be
> retrieved from logs, via ftp perhaps?
It would be a convenient way for the browser to rank all of the
possible paths from the current document and therefore do prefetching.
In other words, if I (an information provider) know that the link
"General information" on the NCSA home page is the most common link
selected by people who access the home page, then I could use this
FREQ attribute to encode that in the document itself and therefore a
smart browser could prefetch whatever's at the other end of that link
each time (or the first time, or whatever) a user accessed the home
page.
(I think this shouldn't be part of the markup language; however, it's
fun to think about how it could be pulled off otherwise. I'm in favor
of a local agent that watches all dataflow in and out of my system and
can therefore learn to predict what I'm going to do next.)
Marc