[quoted text, click to view] On May 13, 3:28 am, cricket...@gmail.com wrote:
> A new server has been configured where I work and the OS, SQL Server
> Data and Logs have all been put on a Raid 5 array. I challenged this,
> and got told this "is the best way of doing it in regard to
> performance" by a senior DBA. This server will contain over 40 fairly
> large databases, and is a pretty powerful server spec wise with 12GB
> RAM, 4 Cpus.
>
> >From my understanding, you should have the OS and SQL log files on
>
> RAID 1 and the SQL data files on Raid 5. Is that correct?
>
> >From searching through Google, and EE, it appears as though I'm
>
> correct, but I've been been attempting to find something "official" on
> Microsoft that I use to back my opipion up but I can't seem to find
> anything. Does anyone know of anything I can use? Or am I wrong with
> what I've said?
How about citing Books On Line to support your case?
[quoted text, click to view] >From Optimizing Database Performance | Database Design | Physical
Database Design | RAID
"If data must be quickly recoverable, consider mirroring the
transaction log and placing the database on a RAID 5 disk. RAID 5
provides redundancy of all data on the array, allowing a single disk
to fail and be replaced in most cases without system downtime. RAID 5
offers lower performance than RAID 0 or RAID 1 but higher reliability
and faster recovery."
There is a nice diagram/explanation of RAID 5 at
http://www.pcguide.com/ref/hdd/perf/raid/levels/singleLevel5-c.html. And RAID 1, a.k.a. mirroring, at:
http://www.pcguide.com/ref/hdd/perf/raid/levels/singleLevel1-c.html. (These take a while to load.)
The "Recommended Uses" sections bolster your case.
Now for the politics: how do you go back to the senior DBA?
The extract from BOL, if read quickly, could be construed as
supporting him.
Maybe rewriting that as:
If data must be quickly recoverable, consider:
- mirroring, i.e. using RAID 1, for the transaction log
and
- placing the database on a RAID 5 disk.
Roy