Re: Performance problems with Postgresql/ZFS/Non-global zones on Solaris?

[Mayuresh Nirhali <Mayuresh(dot)Nirhali(at)Sun(dot)COM>]

Bob,

First, Which exact version of S10 are you using ?

In general, I have seen having logs and data on different pools helps.
Also, let us know your wal tunning parameters like commit_delay, fsync. 
If you are using S10u4, setting commit_delay to a non zero number should 
help get better performance.

Rgds
Mayuresh


jiniusatwork-postgresql(at)yahoo(dot)com wrote:
> In the hopes that someone has already blazed this trail ...
>
>
> I'm running Postgresql (v8.1.10) on Solaris 10 (Sparc) from within a non-global zone. I originally had the database "storage" in the non-global zone (e.g. /var/local/pgsql/data on a UFS filesystem) and was getting performance of "X" (e.g. from a TPC-like application: http://www.tpc.org). I then wanted to try relocating the database storage from the zone (UFS filesystem) over to a ZFS-based filesystem (where I could do things like set quotas, etc.). When I do this, I get roughly half the performance (X/2) I did on the UFS system. I did try to run some low level I/O tests (i.e. http://www.iozone.org/) to see if there was a filesystem performance difference, but there doesn't seem to be any between the UFS and ZFS numbers I'm seeing. 
>
> So, I was hoping someone might have already tried this type of setup and can provide some suggestions for helping boost the ZFS performance numbers (and save me a truss debugging session). 
>
> Are there any known issues with using Postgresql and ZFS?
>
> Things I have already tried:
>
> - setting the record size of the ZFS filesystem to be 8K (vs 128K ) default -- no noticeable performance difference
>
> - addind the ZFS filesystem as a loopback (i.e. lofs) filesystem in the non-global zone -- no noticeable difference
>
> - adding the ZFS filesystem as a dataset filesystem in the non-global zone -- no noticeable difference
>
> Bob
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 9: In versions below 8.0, the planner will ignore your desire to
>        choose an index scan if your joining column's datatypes do not
>        match
>