Thanks for your feedback, I appreciate it.
Check what I have to say at http://www.powerpostgresql.com/PerfList
Because the proprietary application running the business has not
certified on it. Unfortunately, I am at the mercy of their support in
case something goes wrong.
They're currently on a two-disk Adaptec RAID1 with Postgresql 7.4.2.
And you've not upgraded to 7.4.6 because .... ?
Okay, thanks. Even with 7-disks? I trust that. So, RAID 1+0 (sw) is
probably the best option. I've run sw RAID personally for years without
issue. I am a bit hesitant in doing sw RAID for a production server for
a database --- probably because its not my server. Any thoughts on sw
RAID for Postgresql?
The drive array is a 7-disk fibre channel on a Qlogic 2100 controller. I
am currently testing RAID5 (sw).
In general, RAID 5 is not so great for databases. See the article for more.
Okay. Darn. While I don't write the queries for the application, I do
interact with the company frequently. Their considering moving the
queries into the database with PL/pgSQL. Currently their queries are
done through ProvIV development using ODBC. Will context switching be
minimized here by using PL/pgSQL?
The main reason of moving to a drive array is the high level of context
switches we get during the day (>30K for 20 mins per hour). The OS and
database exist on the same disk but seperate parition (which probably
makes little difference)
Unfortunately, the context switches are probably due to a known issue in
PostgreSQL, and changing the drive array won't help this issue (it may help
other issues though). Search the archives of this list, and pgsql-hackers,
for "Context Switch Bug".
For the CS bug, the only workaround right now is to avoid the query structures
that trigger it.
Centos 3.3 (RHEL 3.x equivelent)
Adaptec 2100S RAID
Qlogic QLA2100 Fibre
Dual Xeon 2.8 CPUs with HT turned off.
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match