Re: [9fans] network storage quips

[Brantley Coile <brantley@coraid.com>]

[-- Attachment #1: Type: text/plain, Size: 258 bytes --]

we've heard these objections for a couple of years,
but have yet to see real problems.  in most cases
people accept a local swap disk, which fixes the
whole problem and still avoid captured data on the server.
you can even have two swap disks mirrored.

[-- Attachment #2: Type: message/rfc822, Size: 2691 bytes --]

From: Charles Forsyth <forsyth@terzarima.net>
To: 9fans@cse.psu.edu
Subject: [9fans] network storage quips
Date: Thu, 27 Oct 2005 23:29:27 +0100
Message-ID: <cb471810f88f8aa05ec1f0682d6f2dd1@terzarima.net>

today i noticed in a recent ;login: a conference summary of the 2005 Kernel Developers Summit,
which observed that the convergence of networking and storage subsystems
``can cause problems when memory gets tight; the block layer needs to write out pages
to free memory, but a network-based storage layer must allocate memory to accomplish
those writes.  There are things that can be done to address this problem,
but Linus Torvalds also wants to push back on the manufacturers of these systems.
Rather than go through all this trouble to make network-based storage work,
wouldn't it be better to just install a local disk?''