pg.ddx.io  pgsql-performance@postgresql.org mailing list archive  
help / color / mirror / Atom feed
Update on putting WAL on ramdisk/
3+ messages / 2 participants
[nested] [flat]

* Update on putting WAL on ramdisk/
@ 2003-12-12 22:02  William Yu <wyu@talisys.com>
  0 siblings, 1 reply; 3+ messages in thread

From: William Yu @ 2003-12-12 22:02 UTC (permalink / raw)
  To: pgsql-performance

Some arbitrary data processing job

WAL on single drive: 7.990 rec/s
WAL on 2nd IDE drive: 8.329 rec/s
WAL on tmpfs: 13.172 rec/s

A huge jump in performance but a bit scary having a WAL that can 
disappear at any time. I'm gonna workup a rsync script and do some 
power-off experiments to see how badly it gets mangled.

This could be good method though when you're dumping and restore an 
entire DB. Make a tmpfs mount, restore, shutdown DB and then copy the 
WAL back to the HD.

I checked out the SanDisk IDE FlashDrives. They have a write cycle life 
of 2 million. I'll explore more expensive solid state drives.




^ permalink  raw  reply  [nested|flat] 3+ messages in thread

* Re: Update on putting WAL on ramdisk/
@ 2003-12-12 22:36  Russell Garrett <rg@tcslon.com>
  parent: William Yu <wyu@talisys.com>
  0 siblings, 1 reply; 3+ messages in thread

From: Russell Garrett @ 2003-12-12 22:36 UTC (permalink / raw)
  To: William Yu <wyu@talisys.com>; pgsql-performance

> WAL on single drive: 7.990 rec/s
> WAL on 2nd IDE drive: 8.329 rec/s
> WAL on tmpfs: 13.172 rec/s
>
> A huge jump in performance but a bit scary having a WAL that can
> disappear at any time. I'm gonna workup a rsync script and do some
> power-off experiments to see how badly it gets mangled.

Surely this is just equivalent to disabling fsync? If you put a WAL on a
volatile file system, there's not a whole lot of point in having one at all.

--------------------------------------------------------------------
Russ Garrett                                            russ@last.fm
                                                      http://last.fm




^ permalink  raw  reply  [nested|flat] 3+ messages in thread

* Re: Update on putting WAL on ramdisk/
@ 2003-12-12 22:45  William Yu <wyu@talisys.com>
  parent: Russell Garrett <rg@tcslon.com>
  0 siblings, 0 replies; 3+ messages in thread

From: William Yu @ 2003-12-12 22:45 UTC (permalink / raw)
  To: pgsql-performance

Russell Garrett wrote:
>>WAL on single drive: 7.990 rec/s
>>WAL on 2nd IDE drive: 8.329 rec/s
>>WAL on tmpfs: 13.172 rec/s
>>
>>A huge jump in performance but a bit scary having a WAL that can
>>disappear at any time. I'm gonna workup a rsync script and do some
>>power-off experiments to see how badly it gets mangled.
> 
> 
> Surely this is just equivalent to disabling fsync? If you put a WAL on a
> volatile file system, there's not a whole lot of point in having one at all.

These tests were all with fsync off.

And no, it's not equivalent to fsync off since the WAL is always written 
immediately regardless of fsync setting.




^ permalink  raw  reply  [nested|flat] 3+ messages in thread


end of thread, other threads:[~2003-12-12 22:45 UTC | newest]

Thread overview: 3+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2003-12-12 22:02 Update on putting WAL on ramdisk/ William Yu <wyu@talisys.com>
2003-12-12 22:36 ` Russell Garrett <rg@tcslon.com>
2003-12-12 22:45   ` William Yu <wyu@talisys.com>

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox