Looking for physical design suggestions...

I got a server that has a RAID-5 array partitioned into C: and D:
drives (OS Win2K Adv. Server installed on C:). The server also has a
mapping to a NAS device using the latest protocols that trick the
system into thinking the map is actually a local SCSII drive. That's
drive X:.
This server is used only for SQL, and contains an OLTP database that
sees a lot of use and is pretty heavily indexed.
I am toying with the idea of centralizing my data storage on the NAS
(data center network segment is 1-gigabit ethernet). So I was
thinking about putting my primary data file on the NAS (drive X:) and
keeping all tables there, creating a secondary data file on local
RAID-5 (drive D:) and putting all non-clustered indexes there, as well
as keeping the tempdb there and specifying the sort in tempdb option.
Log files would also remain on D:.

If anyone can suggest a better scenario given the above setup - I'd
love to hear it. Much appreciated.

Alexey AksyonenkoAlthough it is technically possible to put SQLServer database files on a NAS
drive, this is not supported by Microsoft and they strongly recommended that
you don't do it. With a NAS drive you lose many of the benefits of a
client-server database, you will get poor performance and you risk
corrupting your data.

Databases belong on direct-attached or SAN storage.

--
David Portas
----
Please reply only to the newsgroup
--

"Alexey Aksyonenko" <Alexey.Aksyonenko@.coanetwork.com> wrote in message
news:1449e414.0309260612.7558f05f@.posting.google.c om...
> I got a server that has a RAID-5 array partitioned into C: and D:
> drives (OS Win2K Adv. Server installed on C:). The server also has a
> mapping to a NAS device using the latest protocols that trick the
> system into thinking the map is actually a local SCSII drive. That's
> drive X:.
> This server is used only for SQL, and contains an OLTP database that
> sees a lot of use and is pretty heavily indexed.
> I am toying with the idea of centralizing my data storage on the NAS
> (data center network segment is 1-gigabit ethernet). So I was
> thinking about putting my primary data file on the NAS (drive X:) and
> keeping all tables there, creating a secondary data file on local
> RAID-5 (drive D:) and putting all non-clustered indexes there, as well
> as keeping the tempdb there and specifying the sort in tempdb option.
> Log files would also remain on D:.
> If anyone can suggest a better scenario given the above setup - I'd
> love to hear it. Much appreciated.
> Alexey Aksyonenko|||Thanks David. Actually, they came out with this new protocol that as I
said makes the system think that the mapping is a local drive. But I
will definitely keep this in mind. Any suggestions on the actual
physical design portion, NAS aside?

*** Sent via Developersdex http://www.developersdex.com ***
Don't just participate in USENET...get rewarded for it!|||"Alexey Aksyonenko" <alexey.aksyonenko@.coanetwork.com> wrote in message
news:3f7469fb$0$62077$75868355@.news.frii.net...
> Thanks David. Actually, they came out with this new protocol that as I
> said makes the system think that the mapping is a local drive. But I
> will definitely keep this in mind. Any suggestions on the actual
> physical design portion, NAS aside?

Ignoring the NAS the answer is: Doesn't really matter. Since your C and D
drives are logical partitions of a single physical RAID 5 disk, it won't
make much of a difference how you lay out the files.

>
> *** Sent via Developersdex http://www.developersdex.com ***
> Don't just participate in USENET...get rewarded for it!