Re: full-scan vs index for "small" tables
- From: Steve Perry <sperry@xxxxxxxxxxx>
- To: cary.millsap@xxxxxxxxxx
- Date: Tue, 27 Jun 2006 21:46:52 -0500
On Jun 27, 2006, at 09:52 AM, Cary Millsap wrote:
"Most people guess “It’s got to be I/O.” But “it” is I/O in fewer
than 5% of cases I’ve witnessed since about 1995"
What have been the majority of problems you've run into?
For me, IO used to be the problem ( after 2GB drives went away and
before SANs showed up). Mainly because the sysadmins treated
database file systems like "file servers". they'd put everything on a
few spindles and fill them up. then mgt. would say the servers were
under utilized if the average utilization was below 60% so they would
load up the servers with applications to max out the memory.
After I changed companies and started using SANs, most problems seem
to be cpu bound on the server. I've seen a few really good nested
loops queries on small tables that would take hours to complete.
I'm curious what others have run into.
Other related posts: