Hi All,

My company is looking to build a scalable, high availability SQL Server 2008
instance. I've been searching online to find definitive answers for a couple
of questions, namely:

If I setup Fail-over Clustering on 2k8 with 3 nodes, am I able to split the
workload across all 3 nodes? I've read that I can do this with a shared copy
of a read-only version of the database, but I haven't seen anything that says
'Yes' all three nodes can write to the same database. Obviously the benefit
to this is that I can scale out and purchase more nodes to improve
performance. Am I off base here or does SQL 2008 not support this? If SQL
2008 does not support multiple writes, what is the benefit to having more
than two nodes for an organization?

Also, in terms of Mirroring vs. Clustering, its clear that the big
difference is instance level copies vs. database level copies. Just how
difficult is it (approximately) to move the logins and jobs from one server
to the other?

Generally at what point would you recommend clustering over mirroring?

Thanks for the help!

Only one node can write to an individual database. failover clustering is a
high-availability technology, not a scale-out technology.

Multiple nodes allows for multiple SQL Instances. Many organizations use
more than one SQL Server

Moving instance objects across to mirror partners is totally dependent on
the number of objects and the frequency of changes of those objects.

--
Geoff N. Hiten
Principal SQL Infrastructure Consultant
Microsoft SQL Server MVP


"jimjawn" <jimjawn (AT) discussions (DOT) microsoft.com> wrote

Clustering is a High Availability (HA) solution - not a scale-out solution.
Extra nodes are there in case the node on which the instance is running,
fails. Mirroring is also an HA solution. However, you can create a
snapshot off of the mirror and run read-only queries against the snapshot.

You can script out logins and jobs from one instance and create them on
another instance.

Clustering has the advantage that everything is done at the instance level.
Everything fails over together. Mirroring has the advantage that you can
select individual DB's, as well as having the ability to create snapshots
for read-only queries, thus providing a form of scaling out.

--
Tom

----------------------------------------------------
Thomas A. Moreau, BSc, PhD, MCSE, MCDBA, MCITP, MCTS
SQL Server MVP
Toronto, ON Canada
https://mvp.support.microsoft.com/profile/Tom.Moreau


"jimjawn" <jimjawn (AT) discussions (DOT) microsoft.com> wrote

Hi All,

My company is looking to build a scalable, high availability SQL Server 2008
instance. I've been searching online to find definitive answers for a
couple
of questions, namely:

If I setup Fail-over Clustering on 2k8 with 3 nodes, am I able to split the
workload across all 3 nodes? I've read that I can do this with a shared
copy
of a read-only version of the database, but I haven't seen anything that
says
'Yes' all three nodes can write to the same database. Obviously the benefit
to this is that I can scale out and purchase more nodes to improve
performance. Am I off base here or does SQL 2008 not support this? If SQL
2008 does not support multiple writes, what is the benefit to having more
than two nodes for an organization?

Also, in terms of Mirroring vs. Clustering, its clear that the big
difference is instance level copies vs. database level copies. Just how
difficult is it (approximately) to move the logins and jobs from one server
to the other?

Generally at what point would you recommend clustering over mirroring?

Thanks for the help!