Platform: SQL 2000 Enterprise with SP4, running on Windows 2003 Enterprise, 8 cores, 8GB RAM, and internal (!) storage. Databases are roughly 100GB in total.
Scenario: The waitstats are heavily skewed to hardware:
CXPACKET - 35.9%
WRITELOG - 18.4%
LCK_M_S - 13.2%
PAGEIOLATCH_SH - 10.9%
NETWORKIO - 10.4%
LATCH_EX - 5.6%
PAGEIOLATCH_EX - 2.6%
LCK_M_IX - 1.5%
LCK_M_IS - 1.2%
PAGELATCH_UP - 0.1%
I am ignoring the LATCH and LCK types for now, because we're moving to SAN shortly during a SQL 2005 upgrade, and the internal drives are heavily fragmented anyway.
However, I'm concerned about the CXPACKET type. In some cases I've had 80% wait times relating to this. Does the parallel execution plan setting need to be reviewed? (it is set to 5 at the moment). Will moving to SAN fix this as well? Should I even worry in the first place?
Thanks in advance!