Hi all,

Does anyone know if it possible to have active/passive
instances of SQL on a single virtual server? I'm
thinking not but if anyone has any info it would be much
appreciated.

Cheers,

Dave.

Are you asking if you can have an Active / Passive (now referred to as
Multi-Instance) cluster on a single physical server? If so since Clustering
is a hardware fail over solution so it would be mostly pointless. With that
said I have seen people use VMWare to simulate multiple servers for testing
purposes but there is no reason you would want to do this in a production
scenario.

--
Andrew J. Kelly SQL MVP


"Dave N" <anonymous (AT) discussions (DOT) microsoft.com> wrote

I posted more out of curiosity to be honest. A colleague
asked if he could have a more than one SQL instance on a
single virtual server (i.e not create a new SQL instance
on a new virtual server on the same cluster). I'm not
talking at the hardware level no, at a SQL level. My
colleague states that you could then run the 2 SQL
instances in an active/passive mode. This would be in a
testing (dev) environment yes - purely theory work at the
minute.

Dave,

You can install SQL Server on a single-node cluster. However, each clustered
instance of SQL Server on that one machine must be installed with a new
virtual server name and IP.

Unless there's some subtlety I'm missing, I believe your colleague is wrong.
On a single node cluster, there is no other node of the cluster to fail over
to. In other words, the end result is Active, period, and no passive
whatsoever, because there's no second node. Or a better way to put it,
you're suggesting multi-instance SQL Server virtual servers on a single
node.

Also, as Andrew points out, there's no failover on a single-node cluster,
and so single-node clusters are normally meant only for testing and
learning.

Hope this helps,
Ron
--
Ron Talmage
SQL Server MVP


"Dave N" <anonymous (AT) discussions (DOT) microsoft.com> wrote

Ron,
Thanks for this. I am Dave's colleague who originally
posed the question. This was derived from the principle
that if clustering fails over pre-defined (cluster-aware)
resources why not failover between resource defined on one
node only. Since Dave posted this I have acrried out some
more investigations and beleive the SQL setup wizard is
limiting the installation to select resources on diffrent
nodes. If indeed it cannot be done it would be nice to
know if this is a limitation of MSCS or SQL Setup.

NB Although there is a BOL entry headed 'How to install a
one-node failover cluster' it just install a single
instance that doens't failover.

Many thanks,
Tim