Tuesday, March 27, 2012

A DB with the same name om two nodes

I'm not completly sure how to phrase this question.
In the 2 node cluster I'm working with, I noticed that there is a database
named "Utility" on both nodes of the cluster, which confuses me. I thought an
instance was clustered, not a database, so finding the duplicate DB doesn't
make sense.
Now I understand that master, model, msdb & tempdb will be duplicated across
nodes, but these are system DBs and (I would have assumed) special. The
Utility DB is serving the function for local stuff similar to a combination
of msdb & master, but for local admin applications.
I was of the impression that an instance was clustered, not a database. So,
I don't understand how the duplicate could (or should) be there.
OK, it is clear that you "can", but what happens during a failover? It would
seem that the copy on the failed instance would be unavailable and could
cause issues.
Also, does that mean that you can have a db on a clustered instance that is
not available after a failover?
"Edwin vMierlo [MVP]" wrote:

> "JayKon" <JayKon@.discussions.microsoft.com> wrote in message
> news:9467D8C2-0970-485E-9ADF-D3DA2E4E314A@.microsoft.com...
> an
> doesn't
> across
> combination
> So,
> Yes you can,
> so "Instance1" has a database called "data1" in a cluster group called
> "Group1"
> and "Instance2" has a database called "data1" in a cluster group called
> "Group2"
> Maybe I misunderstand your post, but I do not see a problem
> rgds,
> Edwin.
>
>
|||Two different instances on the same cluster act completely independently of
each other, so the database names can all be the same or different.
No different really than if you installed two instances on a stand alone and
created a database called "myDB" on each, other than the fact that in your
cluster, each instance has its own drive resources.
Kevin Hill
3NF Consulting
http://www.3nf-inc.com/NewsGroups.htm
Real-world stuff I run across with SQL Server:
http://kevin3nf.blogspot.com
"JayKon" <JayKon@.discussions.microsoft.com> wrote in message
news:12B664B7-DF3F-4B5D-8B0F-4AA9F1B2B4E3@.microsoft.com...[vbcol=seagreen]
> OK, it is clear that you "can", but what happens during a failover? It
> would
> seem that the copy on the failed instance would be unavailable and could
> cause issues.
> Also, does that mean that you can have a db on a clustered instance that
> is
> not available after a failover?
> "Edwin vMierlo [MVP]" wrote:
|||How many instances do you have installed?
There should only be one set of system databases per instance. And, they
should be located on a "shared" physical disk. Only 1 node should have
ownership of this disk at a time; therefore, you shouldn't have multiple
copies unless you had multiple instances.
Sincerely,
Anthony Thomas

"JayKon" <JayKon@.discussions.microsoft.com> wrote in message
news:9467D8C2-0970-485E-9ADF-D3DA2E4E314A@.microsoft.com...
> I'm not completly sure how to phrase this question.
> In the 2 node cluster I'm working with, I noticed that there is a database
> named "Utility" on both nodes of the cluster, which confuses me. I thought
an
> instance was clustered, not a database, so finding the duplicate DB
doesn't
> make sense.
> Now I understand that master, model, msdb & tempdb will be duplicated
across
> nodes, but these are system DBs and (I would have assumed) special. The
> Utility DB is serving the function for local stuff similar to a
combination
> of msdb & master, but for local admin applications.
> I was of the impression that an instance was clustered, not a database.
So,
> I don't understand how the duplicate could (or should) be there.
|||It's less of a concern, than a comprension issue.
So, if I have one database per instance (what I would have expected) during
a failover, that DB will be seen by the other instance. However, if I have
the same database name on both instances, then during a failover, the local
copy will be the only visable one?
Like I sad, a compresion issue.
"Edwin vMierlo [MVP]" wrote:

> As Kevin already mentioned, after failover, there should be no difference,
> other than the two instances are online on the same physical node.
> An instance "lives" in a cluster group. The cluster group "acts" like an
> completely independent server, with it own Network Name, Ip address, disks,
> databases.
> Hope this helps to take your concerns away,
> Best Regards,
> Edwin.
> MVP - Windows Server - Clustering
>
>
> "JayKon" <JayKon@.discussions.microsoft.com> wrote in message
> news:12B664B7-DF3F-4B5D-8B0F-4AA9F1B2B4E3@.microsoft.com...
> would
> is
> database
> thought
> The
> database.
>
>
|||SQL Server instances failover, not databases. It is no difference than
running multiple instances on a stand-alone server, except instances in a
clustered configuration also differ by virtual server network name, but they
are still independent, ISOLATED binaries and databases. Nothing is "shared"
between the instances except for the cluster nodes that can potentially host
the resources.
The databases failover from one node to the other because the SQL Server
instance, network name, IP address, and disk change ownership between the
nodes. When SQL Server starts on the new host, it recovers the databases
just as if you had just restarted the services.
Sincerely,
Anthony Thomas

"JayKon" <JayKon@.discussions.microsoft.com> wrote in message
news:5EDABA97-5CC3-42BF-9EEA-A051C2A92FC7@.microsoft.com...
> It's less of a concern, than a comprension issue.
> So, if I have one database per instance (what I would have expected)
during
> a failover, that DB will be seen by the other instance. However, if I have
> the same database name on both instances, then during a failover, the
local[vbcol=seagreen]
> copy will be the only visable one?
> Like I sad, a compresion issue.
>
> "Edwin vMierlo [MVP]" wrote:
difference,[vbcol=seagreen]
disks,[vbcol=seagreen]
could[vbcol=seagreen]
that[vbcol=seagreen]
DB[vbcol=seagreen]
duplicated[vbcol=seagreen]
special.[vbcol=seagreen]
called[vbcol=seagreen]
called[vbcol=seagreen]

No comments:

Post a Comment