RE: Hard Disk selections
- From: "Eric Buddelmeijer" <Eric.Buddelmeijer@xxxxxxxxxx>
- To: <bdbafh@xxxxxxxxx>, <bunjibry@xxxxxxxxx>
- Date: Thu, 22 Sep 2005 17:44:37 +0200
_____
Van: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]
Namens Paul Drake
Verzonden: donderdag 22 september 2005 17:20
Aan: bunjibry@xxxxxxxxx
CC: Oracle-L
Onderwerp: Re: Hard Disk selections
Bryan,
It depends.
What you have going for you is:
CPU bogomips (500 MHz to likely 3.4 GHz)
Memory bandwidth (100 MHz FSB to 800 MHz effective)
Newer disks tend to have faster interfaces, larger cache, higher RPMs and
shorter seek times.
The SAN will likely have some form of cache.
>This triggered me. We had some issues with a SAN cache not being configured
at all. Result was the throughput became >lousy although the disks were fast
enough. What I learned from it is that you have to check if this is
configured correctly >before looking at your database.
If you're only loading up data into the buffer cache at instance startup and
only writing to disk for redo, controlfile and archived redo logs, you may
not notice the fewer disks.
If you find that the CBO wants to do all table scans, you're in trouble.
The only way to know for sure is to test.
Get as much memory in the box as your budget allows and only use half of it.
Use w2k advanced server instead of w2k server, as it has the /PAE and /3GB
options.
W2k3 server standard edition supports the /3GB option.
hth.
Paul
Other related posts:
