*BSD News Article 71863


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!news.unimelb.EDU.AU!munnari.OZ.AU!news.ecn.uoknor.edu!news.cis.okstate.edu!newsfeed.ksu.ksu.edu!news.physics.uiowa.edu!math.ohio-state.edu!uwm.edu!news-res.gsl.net!news.gsl.net!nntp.coast.net!dispatch.news.demon.net!demon!news.ktgroup.co.uk!mike
From: mike@lurch.ktgroup.co.uk (Mike Pellatt)
Newsgroups: demon.ip.support,demon.tech.unix,comp.unix.bsd.freebsd.misc
Subject: Re: Batch FTP and Web Pages
Date: 23 Jun 1996 16:48:04 GMT
Organization: Knowledge Technology Group
Lines: 47
Message-ID: <slrn4sqta3.hin.mike@lurch.ktgroup.co.uk>
References: <31c2e7bd.14691630@news.demon.co.uk> <834878464snz@pair.com> <834921960snz@michaels.demon.co.uk> <835206024.5881.2@diltd.demon.co.uk> <bKyuJQj030n@sktb.demon.co.uk> <835386370.22158.1@diltd.demon.co.uk> <slrn4snqn3.dmo.mike@lurch.ktgroup.co.uk> <4qht30$15c@illuin.demon.co.uk>
NNTP-Posting-Host: lurch.ktgroup.co.uk
X-NNTP-Posting-Host: www.ktgroup.co.uk
X-Newsreader: slrn (0.8.6.1)

On 22 Jun 1996 22:45:52 GMT, Richard Letts <richard@illuin.demon.co.uk> wrote:
>Mike Pellatt (mike@lurch.ktgroup.co.uk) wrote:
>) On Fri, 21 Jun 1996 19:46:09 GMT, Ric Harwood <ric@diltd.demon.co.uk> wrote:
>) >In demon.ip.support, "Paul L. Allen" <pla@sktb.demon.co.uk>, "Paul
>) >L. Allen" wrote:
>) >
>) >>It already is, if you but have the wits to use it.  Tell your browser to use
>) >>Demon's proxy server.  If the page is in the proxy server's cache, you will
>) >>get it quite quickly.  If it isn't there, the proxy server will go and get
>) >>it.  Now, here's the clever part: if it's taking too long to turn up, bugger
>) >>off and try again later, when it will be in the proxy server's cache waiting
>) >>for you...
>) >
>) >So once I have asked for it the proxy get's it even though I have
>) >have dropped the session, the soccet and the line?
>) >Funny, it didn't seem to work that way, but I admit that I wasn't
>) >watching for it in that way.
>
>) Certainly the Harvest cache works that way. Demon ran Harvest
>) for a (short) while, but seem to have reverted to the CERN caching
>) server.  Dunno why - I know CERN is slower, but I beleieve there are some
>) stability porblems with Harvest under heavy load. I don't know
>) if CERN behaves that way - it's process model is very different from
>) Harvest.
>
>both *can* behave that way, and both can be configured *not* to.
>
>CERN required OODLES of RAM on the server as each 'document'
>[text, picture, gif, etc] requires a separate PROCESS.
>Harvest uses a non-blocking design with a single process, and is possibly
>better for small servers.

I should have said that about Harvest - indeed, it's more versatile
than that. The maximum size of document to be cached is one of the
configurable settings (per protocol), and documents larger than
the maximum cacheable size are proxied without being cached. For
such documents, disconnecting and coming back later would gain you
nothing.

I seem to remember I set the maximum cacheable size up to 4MB on
our server to help with the slow US links.

-- 
Mike Pellatt, VCS Limited (A Knowledge Group company)
Tel: (+44) 117 9007500     Fax: (+44) 117 9007501     Mobile: (+44) 468 192021
Home Page: http://www.ktgroup.co.uk/~mike/