| From: | "Luke Lonergan" <LLonergan(at)greenplum(dot)com> |
|---|---|
| To: | "Charles Sprickman" <spork(at)bway(dot)net>, "Mikael Carneholm" <Mikael(dot)Carneholm(at)WirelessCar(dot)com> |
| Cc: | "Kjell Tore Fossbakk" <kjelltore(at)gmail(dot)com>, pgsql-performance(at)postgresql(dot)org |
| Subject: | Re: Performance with 2 AMD/Opteron 2.6Ghz and 8gig |
| Date: | 2006-07-29 06:06:27 |
| Message-ID: | 3E37B936B592014B978C4415F90D662D03E6CE74@MI8NYCMAIL06.Mi8.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-performance |
Charles,
> FWIW, once our vendor gets all the pieces (having some issues
> with figuring out which multilane sata cables to get), I'll
> have a dual-core opteron box with a 3Ware 9500SX-12MI and 8
> drives. I need to benchmark to compare this to our
> xeon/adaptec/scsi build we've been using.
Cool! You mean the 9550SX, not the 9500, right?
A trick on the 9550SX with Linux is to set the max readahead to 512KB
and no larger when using RAID10. If you use RAID5, set it to 16MB.
Here is how you set it (put in /etc/rc.d/rc.local) for 512KB on
/dev/sda:
/sbin/blockdev --setra 512 /dev/sda
I was able to go from 310MB/s on 8 drives to 475MB/s that way (using
XFS).
Also, you need to stay away from Linux Volume Manager (lvm and lvm2),
they add a lot of overhead (!!?) to the block access. It took me a long
time to figure that out!
- Luke
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jochem van Dieten | 2006-07-29 15:02:46 | PostgreSQL scalability on Sun UltraSparc T1 |
| Previous Message | Charles Sprickman | 2006-07-29 05:57:19 | Re: Performance with 2 AMD/Opteron 2.6Ghz and 8gig |