Posted by kmounkhaty@yahoo.com on 11/30/05 00:06

Hi Guru,

After spening quite sometimes to watch my box, I've seen PAGEIOLATCH is
a lead blocker in my SQL Server 2000 server. Below is the detailed:

SPID lastwaittype waitresource blocked status cmd
57 LCK_M_S KEY: 7:963690681:8 65 sleeping execute
65 PAGEIOLATCH_SH 7:1:217904 0 sleeping select


I thought, latching should be very short-term synchronization. From
systemprocess table, I saw the latch waited in a minute sleeping
without doing any work.

My database is about 23GB and more than 5000 tables. The RAID subsystem
is RAID1 with 1 disk mapped to C and D logically. Data files and tempdb
files are located in one location. Tranlog file and log backup files
are located in the same location with different disk spindle.

Currently, we are experiencing very slowness and IO bound. I'm ready to
rebuild the server by putting the RAID10 and 1 and distributing
multiple data files to different RAID10 and tempdb and log files to
RAID1.

Other than this, how to minimize the IO latch contention?

Thanks so much,

Silaphet,

[Back to original message]