W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > May to August 1995

caching related topics in the draft and the http-wg discussion

From: Endre Balint Nagy <bne@bne.ind.eunet.hu>
Date: Tue, 15 Aug 1995 00:11:18 +0200
Message-Id: <199508142211.AAA00399@bne.ind.eunet.hu>
To: Internet-Drafts@cnri.reston.va.us
Cc: http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
Hi all,
I try to summarise the impact of the results of current draft and the discussion
on the caching proxies. (O, yeah, caching proxies is my favorite topic for now).
The draft states:
a Pragma: no-cache included in the server resposes simply forbids cacheing of
the response. It's quite straighforward, and makes it possible to mark web
documents, which are in their own nature unappropriate for cacheing.
On the other hand, the draft doesn't state, that absence of this pragma means
that the document contained in the response is appropriate for cahceing.
Unfortunately!
If we used the Pragma: no-cache consistently, e.g. applying it to every
non-cacheable response, the life of cache implementors and administrators would
be much easier. Why not act that way?!
Upgrading a cache software is a costly process - a cache running significant
time can accomodate gigabytes of cached information. If we have no handy way to
convert the whole cache into format of the new software, we will lose those
gigabytes, and will have to pump through the net the whole again.
Any guesswork made on the future internet traffic says, that the dominant part
of the traffic will be HTTP. The bandwith  of the long-distance links won't keep
up with that growth, unless caches will be widely employed. Currently we have
not too much experience with caches, especially with cascaded ones. The future
clearly belongs to cascaded caches. Having good standard, which helps
implementing good and stable caches will decrease the probability of such
situations. The possible ongoing Proxy-Authetication feature will not prevent
running cache software to be useful, and will not force the costly upgrade
process, unless there will be some (wich is not too probable in the foreseeable
future) safe, stable and standard authentication scheme, which will give enough
benefits to compensate the costs. When the standard, safe authentication scheme arrives,
(but recently PGP, the nearly only candidate was removed from shttp proposal)
that time can be a good point to re-standardise the caching again, but that
shouldn't happen too often.

Both the referred proposals, e.g.
Dave Kristols state-ids (URL http://www.research.att.com/~dmk/session.html )
and Koen Holtmans Non-Persistent Cookie proposal want additional rules for
proxies and caches. If this trend continues, cache/proxie implementors
will be lost unless the no-cache pragma is used according to my proposal.

Here comes the proposal itself:
I want to add the following two sentences to section 8.22 of the draft:
All compliant implementations MUST include the no-cache pragma in responses
which are unappropriate to cacheing for any reason. If the reason is political
(non-technical), server administrators should ensure that.

Andrew. (Endre Balint Nagy) <bne@bne.ind.eunet.hu>
Received on Monday, 14 August 1995 15:25:03 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:24 EDT