On Sat, 2006-12-02 at 14:31 +0100, Martijn van Oosterhout wrote:
> On Fri, Dec 01, 2006 at 10:58:44PM -0800, Joshua D. Drake wrote:
> > I know this isn't *our* fault :) but I am curious if there is anything
> > we can do about the way postgresql writes files to help limit fragmentation.
> >
> > Essentially, this makes win32 impossible in a 24x7 environment (jokes
> > aside about Win32 in general) because we *have* to defrag on Windows and
> > Windows won't defrag open files (thus anything PostgreSQL is using).
>
> I thought fragmentation was something that disappeared with the FAT
> filesystem. Isn't NTFS smart enought o avoid fragmentation in the first
> place?
Nope... you still have the good old defrag command (well button now).
> BTW, do you know what 11% fragmentation means? Does that mean each file
> is on average split in 9 pieces, because for a 1GB file, 9 pieces isn't
> all that bad.
No, the report had Total Fragmentation, and File Fragmentation. I was
reporting on Total, which I assume is some aggregation.
My concern is that this is over a single bench run. I could imagine that
after a week or two weeks of stead PostgreSQL use, the IO would
gradually get worse and worse.
Joshua D. Drake
>
> Have a nice day,
--
=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 || 24x7/Emergency: +1.800.492.2240
Providing the most comprehensive PostgreSQL solutions since 1997 http://www.commandprompt.com/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate