I have read that best practice for failover cluster with SQL 2008 is 4 disk
per instance. Can someone please explain what the purpose of the 4 disks
are? It seems to me that this is a lot of slicing up the disks?

Where ever you read that I would stay away from future readings. There is
no generic best practice to the number of disks that you need since every
database and application is different. And the number of disks has nothing
to do with if it is in a cluster or not.

--

Andrew J. Kelly SQL MVP
Solid Quality Mentors

"Go Oilers" <maxwelli (AT) nospam (DOT) postalias> wrote

The information was from theMSSQL Tips web page. The recommendation was to
use one for quorum, one for msdtc, one for sql server system databases, and
one for user databases.

What I am trying to figure out is that I currently have 2 SQL 2005 servers
with 4 instances of SQL on them. I would like to move them to a two node
SQL 2008 server. Currently the servers each have a 200 GB disk on the san
that is shared by the both instances on the server. The sql program files
are on the hard drive and the database files are on the SAN.

I just want to make it as redundant as possible.

"Andrew J. Kelly" <sqlmvpnooospam (AT) shadhawk (DOT) com> wrote

OK you didn't mean disks you meant arrays or LUNs. While the advice they
gave is great if you have enough LUN's etc as always it depends. But you
need to understand the difference between a disk / disk array and a LUN or
Mount Point. You say that two instances share a 200GB disk on the san so I
assume they are not currently clustered. This is really a LUN which may or
may not be the entire piece of a disk array but not likely. Chances are you
have a much larger array that is broken down into smaller units and
presented as LUN's or even Mount Points. If you move to a cluster the LUN's
cannot be shared between instances if you want them to act independently and
fail over individually. I highly suggest you do a LOT more reading on
exactly what a cluster is and how SQL Server works with a cluster before
even attempting to make any decisions like this. There is a lot of
information to take in and understand that cannot be effectively relayed on
a news group post. Here are just some links related to clustering with SQL
Server and the first one has a white paper you can download on clustering
with SQL 2008 that I highly recommend you read first.

http://www.microsoft.com/sqlserver/2...ilability.aspx

http://msdn.microsoft.com/en-us/libr...4(SQL.90).aspx
http://www.microsoft.com/downloads/d...DisplayLang=en
http://msdn.microsoft.com/en-us/library/ms189910.aspx
http://www.sql-server-performance.co...ing/index.aspx


--

Andrew J. Kelly SQL MVP
Solid Quality Mentors

"Go Oilers" <maxwelli (AT) nospam (DOT) postalias> wrote

Go Oilers,

I support a SQL 2005 / BizTalk environment. There are 2 SQL servers in the
environment. Our cluster groups are basically configured like this:

Core Group:
IP Address
Network Name
Quorum drive

MSDTC Group
IP Address
Network Name
1 Physical Disk
MSDTC resource

Enterprise Single Sign-on Group
IP Address
Network Name
Generic Service for ENTSSO

SQLA (Node-A)
IP Address
Network Name
3 Physical Disks (Date drive, log drive, backup drive)
SQL Server (INST1)
SQL Server Agent (INST1)
SQL Server Fulltext (INST1)

SQLB
Same as SQLA, but with different drive letters.
Same as SQLA, but INST2.

We run SQL Server Analysis Services, so there is a group for that also.
SSASGROUP
IP Address
Network Name
1 Physical Disk

HTH,

DetRich




"Andrew J. Kelly" wrote: