Harddisk performance degrading over time?

Actually, the culprit is probably PostgreSQL and not
the ext2 file system. Have you tried vacuuming you
database and running the benchmarks again? PostgreSQL
uses a non-overwriting storage manager, and so vacuum
needs to be run periodically to clear out expired
tuples.

Jason

--- Daniel_���kerud <zilch@home.se> wrote:

Do the hard disk degrade in performance over time on
a Linux ext2 filesystem? All my performance test
programs show 20% worse performance and I have done
NADA! I just ran them again :(

Do you have to defrag?

---
Daniel ���kerud

[ Don't underestimate the power of stupid people in
large groups]

__________________________________________________
Do You Yahoo!?
Get personalized email addresses from Yahoo! Mail
http://personal.mail.yahoo.com/

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering the
most. The non-indexed searches is less affected. This is not only PostgreSQL
but also MySQL.

The database is totally restored each test run.

Daniel �kerud

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering the
most. The non-indexed searches is less affected. This is not only PostgreSQL
but also MySQL.

I know the BSD filesystems are self-defragmenting. I don't know if the
ext2 filesystems are the same. Surely someone must know.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

On Sun, 24 Jun 2001, Bruce Momjian wrote:

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering the
most. The non-indexed searches is less affected. This is not only PostgreSQL
but also MySQL.

I know the BSD filesystems are self-defragmenting. I don't know if the
ext2 filesystems are the same. Surely someone must know.

Not exactly, BSD ffs (fast filesystem) isn't self-defragmenting, it just
fragments differently [less than] ext2 :)

Yes, BSD is more "doesn't fragment much" rather than "self
defragmenting".

See this for paper and some tools to get ffs fragmentation stats:
http://www.eecs.harvard.edu/~margo/usenix.195
(it has links to critique of the paper as well)

There's a tool ftp://ftp.uk.linux.org/pub/linux/sct/defrag/ to defrag ext2
filesystem, but its considered to be alpha quality and not really
maintained.

Yes, thanks for the clarification.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

On Sun, 24 Jun 2001, Bruce Momjian wrote:

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering the
most. The non-indexed searches is less affected. This is not only PostgreSQL
but also MySQL.

I know the BSD filesystems are self-defragmenting. I don't know if the
ext2 filesystems are the same. Surely someone must know.

Not exactly, BSD ffs (fast filesystem) isn't self-defragmenting, it just
fragments differently [less than] ext2 :)

See this for paper and some tools to get ffs fragmentation stats:
http://www.eecs.harvard.edu/~margo/usenix.195
(it has links to critique of the paper as well)

There's a tool ftp://ftp.uk.linux.org/pub/linux/sct/defrag/ to defrag ext2
filesystem, but its considered to be alpha quality and not really
maintained.

ext2 doesn't need to be defrag'ed either. You CAN, but it shouldn't need it
unless you are doing something very strange.

Ole Gjerde

----- Original Message -----
From: "Bruce Momjian" <pgman@candle.pha.pa.us>
To: "Daniel �kerud" <zilch@home.se>
Cc: "Jason Earl" <jdearl@yahoo.com>; "PostgreSQL-general"
<pgsql-general@postgresql.org>
Sent: Sunday, June 24, 2001 10:57 PM
Subject: Re: Harddisk performance degrading over time?

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering the
most. The non-indexed searches is less affected. This is not only

PostgreSQL

OK,
so now I know it wasn't that.

The strange thing is just that, that looking at all the test values
collected, it seems like only indexed inserts and indexed selects has gotten
slower for BOTH PostgreSQL and MySQL,

Never mind... as long as it is maximally slow now ;) *kidding*

Daniel �kerud

ext2 doesn't need to be defrag'ed either. You CAN, but it shouldn't need

it

unless you are doing something very strange.

Ole Gjerde

----- Original Message -----
From: "Bruce Momjian" <pgman@candle.pha.pa.us>
To: "Daniel �kerud" <zilch@home.se>
Cc: "Jason Earl" <jdearl@yahoo.com>; "PostgreSQL-general"
<pgsql-general@postgresql.org>
Sent: Sunday, June 24, 2001 10:57 PM
Subject: Re: Harddisk performance degrading over time?

Oh yeah,
vacuuming is not the problem here.

Thanks anyway... :)

And also, it seems that it is the indexed searches that is suffering

the