From: Alan Cox <alan@lxorguk.ukuu.org.uk>
To: Daniel Phillips <phillips@phunq.net>
Cc: Willy Tarreau <w@1wt.eu>, David Newall <davidn@davidnewall.com>,
linux-kernel@vger.kernel.org
Subject: Re: [ANNOUNCE] Ramback: faster than a speeding bullet
Date: Sun, 16 Mar 2008 21:55:47 +0000 [thread overview]
Message-ID: <20080316215547.07824de7@core> (raw)
In-Reply-To: <200803161457.04580.phillips@phunq.net>
> > That isn't anything to do with what was being proposed. *ORDERING* not
> > flush to media.
>
> This is where you have made a fundamental mistake in your proposal.
> Suppose you have a steady, heavy write load onto ramback. Eventually,
> the entire ramdisk will be dirty and you have to drop back to disk
> speed, right? My design does not suffer from that problem, but your
> proposal does.
In your design the entire ramdisk goes bang and disappears on a crash.
> It gets worse than that. Suppose somebody writes the same region
> twice, how do you order that? Do you try to store that new data
> somewhere, keeping in mind that we are already at terabyte scale? Is
> there a limit on how much overwrite data you may have to store? (No.)
You only have to care about ordering if there is a store barrier between
the two (not usual). You only have to care about filling if you generate
enough dirty blocks at a very high rate (which is unusual for most
workloads). If you don't care about those then we have ramdisk already and
if you want to write a ramdisk driver for external ramdisk great. You'd
also fix the layering violations then by allowing device mapper to
implement things like snapshotting and writeback seperated from your
driver.
Even in the extreme case that you propose there are trivial ways of
getting coherency. Simple example - if you can sweep all the data out in
say 10 minutes then you can buy twice the physical media and ensure that
one of the two sets of disk backups is genuinely store barrier consistent
to some snapshot time (say every 30 minutes but obviously user tunable).
If you at least had some kind of credible snapshotting you'd find people
less hostile to your glorified ramdisk.
> > You have no guarantee of commit to stable storage so your use of the word
> > "transaction" is a bit farcical.
>
> The UPS provides a guarantee of commit to stable storage. No amount of
Stable storage to most people means "won't go away on a bad happening".
Transaction likewise has a specific meaning in terms of an event occuring
once only an either being recorded before or after the transaction
occurred.
Alan
next prev parent reply other threads:[~2008-03-16 22:10 UTC|newest]
Thread overview: 153+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-03-10 6:46 [ANNOUNCE] Ramback: faster than a speeding bullet Daniel Phillips
2008-03-10 7:51 ` Grzegorz Kulewski
2008-03-10 8:23 ` Daniel Phillips
2008-03-10 9:37 ` Alan Cox
2008-03-10 21:03 ` Lars Marowsky-Bree
2008-03-11 11:14 ` Daniel Phillips
2008-03-11 11:23 ` Lars Marowsky-Bree
2008-03-11 11:50 ` Daniel Phillips
2008-03-11 17:26 ` Chris Friesen
2008-03-11 19:56 ` Daniel Phillips
2008-03-11 20:53 ` Willy Tarreau
2008-03-12 8:17 ` Daniel Phillips
2008-03-12 14:41 ` Mike Snitzer
2008-03-13 20:34 ` Rik van Riel
2008-03-14 2:20 ` Daniel Phillips
2008-03-11 21:56 ` Lars Marowsky-Bree
2008-03-11 23:02 ` Daniel Phillips
2008-03-12 13:25 ` Benny Amorsen
2008-03-12 13:30 ` Alan Cox
2008-03-13 15:29 ` Benny Amorsen
2008-03-14 9:30 ` Pavel Machek
2008-03-14 11:07 ` Ric Wheeler
2008-03-14 11:41 ` Benny Amorsen
2008-03-14 12:12 ` Ric Wheeler
2008-03-14 12:56 ` Theodore Tso
2008-03-14 15:47 ` Ric Wheeler
2008-03-14 16:49 ` Theodore Tso
2008-03-14 17:04 ` Ric Wheeler
2008-03-14 18:03 ` david
2008-03-14 19:03 ` writeback cache dangers " Pavel Machek
2008-03-14 19:29 ` Theodore Tso
2008-03-13 9:15 ` Matthias Schniedermeyer
2008-03-11 23:30 ` Daniel Phillips
2008-03-13 13:27 ` Ric Wheeler
2008-03-13 19:02 ` Daniel Phillips
2008-03-13 19:12 ` Ric Wheeler
2008-03-13 19:38 ` Daniel Phillips
2008-03-11 4:23 ` Daniel Phillips
2008-03-10 9:22 ` Alan Cox
2008-03-10 19:01 ` Rik van Riel
2008-03-11 4:28 ` Daniel Phillips
2008-03-11 3:50 ` Daniel Phillips
2008-03-11 13:32 ` Artur Skawina
2008-03-11 14:31 ` Artur Skawina
2008-03-12 13:11 ` Alan Cox
2008-03-12 17:29 ` Daniel Phillips
2008-03-12 18:11 ` Chris Friesen
2008-03-12 22:56 ` Daniel Phillips
2008-03-13 5:45 ` David Newall
2008-03-13 6:17 ` Daniel Phillips
2008-03-13 6:30 ` David Newall
2008-03-13 6:50 ` Daniel Phillips
2008-03-13 7:05 ` David Newall
2008-03-13 7:13 ` Daniel Phillips
2008-03-15 13:32 ` Pavel Machek
2008-03-15 20:22 ` Daniel Phillips
2008-03-15 21:33 ` Pavel Machek
2008-03-15 21:47 ` Daniel Phillips
2008-03-13 6:32 ` david
2008-03-13 7:12 ` Daniel Phillips
2008-03-13 7:55 ` david
2008-03-13 8:06 ` Daniel Phillips
2008-03-13 8:39 ` david
2008-03-13 9:16 ` Daniel Phillips
2008-03-13 16:25 ` david
2008-03-13 19:32 ` Daniel Phillips
2008-03-13 19:50 ` David Newall
2008-03-13 20:03 ` Daniel Phillips
2008-03-14 17:53 ` Jeff Moyer
2008-03-15 20:26 ` Pavel Machek
2008-03-15 20:40 ` Mike Snitzer
2008-03-15 21:05 ` Daniel Phillips
2008-03-15 20:18 ` Pavel Machek
2008-03-15 20:51 ` Daniel Phillips
2008-03-13 9:49 ` Daniel Phillips
2008-03-13 5:39 ` David Newall
2008-03-13 6:14 ` Daniel Phillips
2008-03-13 13:22 ` Alan Cox
2008-03-13 19:14 ` Daniel Phillips
2008-03-13 20:27 ` Rik van Riel
2008-03-14 2:23 ` Daniel Phillips
2008-03-14 5:22 ` David Newall
2008-03-14 5:42 ` Daniel Phillips
2008-03-14 14:00 ` John Stoffel
2008-03-15 20:59 ` Willy Tarreau
2008-03-15 20:56 ` Alan Cox
2008-03-15 21:25 ` Daniel Phillips
2008-03-15 21:08 ` Alan Cox
2008-03-15 21:51 ` Daniel Phillips
2008-03-15 21:17 ` Daniel Phillips
2008-03-15 21:03 ` Alan Cox
2008-03-15 22:00 ` Daniel Phillips
2008-03-15 23:05 ` Alan Cox
2008-03-16 21:57 ` Daniel Phillips
2008-03-16 21:55 ` Alan Cox [this message]
2008-03-16 22:36 ` Daniel Phillips
2008-03-16 22:46 ` Alan Cox
2008-03-16 23:39 ` Daniel Phillips
2008-03-17 11:53 ` Alan Cox
2008-03-17 1:31 ` David Newall
2008-03-17 2:42 ` Daniel Phillips
2008-03-17 3:59 ` david
2008-03-17 5:52 ` Daniel Phillips
2008-03-17 6:49 ` david
2008-03-17 8:16 ` Daniel Phillips
2008-03-17 10:39 ` Alan Cox
2008-03-17 13:52 ` Ric Wheeler
2008-03-17 14:42 ` david
2008-03-17 17:23 ` david
2008-03-17 17:30 ` Willy Tarreau
[not found] ` <200803180233.10156.phillips@phunq.net>
2008-03-18 13:03 ` David Newall
2008-03-18 16:36 ` david
2008-03-31 11:40 ` Daniel Phillips
2008-04-01 0:28 ` david
2008-04-01 4:07 ` Daniel Phillips
2008-04-01 4:23 ` david
2008-04-01 6:08 ` Daniel Phillips
2008-03-18 13:57 ` Alan Cox
2008-03-31 11:39 ` Daniel Phillips
2008-03-17 7:14 ` David Newall
2008-03-17 8:25 ` Daniel Phillips
2008-03-17 18:56 ` David Newall
2008-03-23 9:33 ` Pavel Machek
2008-03-23 20:44 ` Daniel Phillips
2008-03-15 21:54 ` Willy Tarreau
2008-03-15 22:33 ` Daniel Phillips
2008-03-15 23:22 ` david
2008-03-15 23:57 ` Krzysztof Halasa
2008-03-15 23:22 ` Willy Tarreau
2008-03-16 3:33 ` Daniel Phillips
2008-03-16 5:24 ` David Newall
2008-03-16 12:49 ` Ingo Oeser
2008-03-16 6:56 ` Willy Tarreau
2008-03-16 22:12 ` Krzysztof Halasa
2008-03-16 13:14 ` Alan Cox
2008-03-16 19:04 ` Theodore Tso
2008-03-16 22:02 ` Krzysztof Halasa
2008-03-15 23:18 ` Bernd Eckenfels
2008-03-16 5:42 ` David Newall
2008-03-16 20:48 ` Daniel Phillips
2008-03-16 22:15 ` Krzysztof Halasa
2008-03-16 22:38 ` Daniel Phillips
2008-03-16 23:08 ` Krzysztof Halasa
2008-03-16 23:43 ` Daniel Phillips
2008-03-10 14:51 ` Artur Skawina
2008-03-10 18:49 ` Chris Snook
2008-03-11 5:06 ` Greg KH
2008-03-11 5:22 ` Daniel Phillips
2008-03-11 5:48 ` david
2008-03-11 6:27 ` Greg KH
2008-03-12 12:01 ` tvrtko.ursulin
2008-03-12 17:27 ` Daniel Phillips
[not found] <OFA00954A4.45F32CA2-ON8025740B.005D7B40-8025740B.005EECA6@sophos.com>
2008-03-13 19:34 ` Daniel Phillips
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080316215547.07824de7@core \
--to=alan@lxorguk.ukuu.org.uk \
--cc=davidn@davidnewall.com \
--cc=linux-kernel@vger.kernel.org \
--cc=phillips@phunq.net \
--cc=w@1wt.eu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox