OK, after several trial installations and reading many
white papers and newsgroup posts, I have become completely
confused on the proper use of Default vs. Named Instances
vs. Virtual Servers. I am looking for the CORRECT METHOD
to implement.

This is what we have:
[2] server nodes, QAT1 and QAT2 running on Windows 2003 EE
SQL Server 2000 Developer's Edition

This is what we NEED:
We want to have 2 "instances" of SQL Server running on
EACH node of the cluster. So, QAT1 will have
two "instances", and QAT2 will have two "instances". This
means that there will be [4] "instances" total, with each
one having its own databases. I originally thought I
could install two virtual servers on each node, using
the "default instance" rather than a named instance. But
when I go to install the first instance on the second
node, QAT2, after the default instance has been installed
on QAT1, it forces a "named instance". We want to
avoid "named instances" if possible for simplification of
front-end client connectivity, but that may not be
possible. I thought the documentation said that each node
could have up to 16 "virtual servers". Can someone

I forgot to mention that each node has its own DATA and LOG
drive. So, on the first node, QAT1, there is drive I: and
K: respectively. We would need the two "instances" on QAT1
to both get installed on the same I: and K: drive.

Would this be two virtual servers on QAT1 each with a
default instance, or a single virtual server on QAT1, with

Thank you very much Chandrasekar and Uttam for clarifying
this situation more for me. So, it looks like I want to
use all named instances for consistency, with a virtual
server required for each named instance. If I understand
you correctly, a separate virtual server is REQUIRED for
each named instance: 1 to 1. I can not have one virtual
server for QAT1, with two named instances, and one virtual
server for QAT2, with two named instances. Also, if I
understand you correctly, each named instance must have
its own set of drives. So, I would need the following
configuration:

[VIRTUAL SERVER NAME]\[NAMED INSTANCE]@[DATA DRIVE]
SQLCLV-QAT1-1\INS1 @ D: [LOG H:]
SQLCLV-QAT1-2\INS2 @ E: [LOG I:]
SQLCLV-QAT2-1\INS1 @ F: [LOG J:]
SQLCLV-QAT2-2\INS2 @ G: [LOG K:]

I would have liked to have only two virtual servers, with
two named instances each, but you are telling me that this
is not possible. I would also have liked to have only one
data and log drive for each node, QAT1 and QAT2, four
drives total, but this also does not seem possible.
Please confirm that I am understanding this correctly.

Hi -
For our discussion, first let us take default and named instance in a
cluster. Later we will take virtual server into consideration.

+SQL server in a 2 node cluster supports 16 instances (this is what you
read).
+In your case, QAT1 and QAT2 together supports 16 instances. Note that the
support of 16 instances is not for a single node but both nodes put
together supports a maximum of 16 instances.
+The value 16 is true even for a non-clustered server. If you have a
windows server standalone machine, it supports a maximum of 16 SQL server
instances. So a clustered node is no different from the non-clustered node
in terms of the supporting the maximum number of instances.
+ There can be only ONE default instance in a cluster. So (1 default + 15
named instances = 16 instances) in a cluster. This is true for a
stand-alone windows server as well. Note that there is no difference
between clustered and non-clustered machines in this point as well.
+ In your case, you wanted to have 4 instances, 2 running on each node.
So QAT1 will have 2 instances ( let us assume 1 default and 1 named)
QAT2 will have 2 instances ( both of this should be named
instances)
This is because only one default is allowed for the entire cluster.

If the above is clear, then we take virtual server into this discussion:

In your case, as mentioned, you will have 4 instances of sql server running.

QAT1 --> 1 default and 1 named
QAT2 --> both named.

Here, there are totally 4 virtual servers. Each instance of SQL server
needs a virtual server name. All your clients will use only these virtual
server names to connect to the database. If you want to have 16 instances
in a cluster, then you need 16 virtual servers. So the number of instances
and number of virtual servers are 1 to 1 related.

Hope this helps.

Regards,
Chandra




--------------------
||Content-Class: urn:content-classes:message
||From: "Michael [multnomah]" <anonymous@discussions.microsoft.com>
||Sender: "Michael [multnomah]" <anonymous@discussions.microsoft.com>
||Subject: Default vs. Named Instance vs. Virtual Server
||Date: Tue, 13 Jan 2004 14:38:55 -0800
||Lines: 26
||Message-ID: <0b9601c3da26$06fafbb0$a001280a@phx.gbl>
||MIME-Version: 1.0
||Content-Type: text/plain;
|| charset="iso-8859-1"
||Content-Transfer-Encoding: 7bit
||X-Newsreader: Microsoft CDO for Windows 2000
||Thread-Index: AcPaJgb4Hk5og5nORFWMthfgNkLpTA==
||X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300
||Newsgroups: microsoft.public.sqlserver.clustering
||Path: cpmsftngxa07.phx.gbl
||Xref: cpmsftngxa07.phx.gbl microsoft.public.sqlserver.clustering:12485
||NNTP-Posting-Host: tk2msftngxa08.phx.gbl 10.40.1.160
||X-Tomcat-NG: microsoft.public.sqlserver.clustering
||
||OK, after several trial installations and reading many
||white papers and newsgroup posts, I have become completely
||confused on the proper use of Default vs. Named Instances
||vs. Virtual Servers. I am looking for the CORRECT METHOD
||to implement.
||
||This is what we have:
||[2] server nodes, QAT1 and QAT2 running on Windows 2003 EE
||SQL Server 2000 Developer's Edition
||
||This is what we NEED:
||We want to have 2 "instances" of SQL Server running on
||EACH node of the cluster. So, QAT1 will have
||two "instances", and QAT2 will have two "instances". This
||means that there will be [4] "instances" total, with each
||one having its own databases. I originally thought I
||could install two virtual servers on each node, using
||the "default instance" rather than a named instance. But
||when I go to install the first instance on the second
||node, QAT2, after the default instance has been installed
||on QAT1, it forces a "named instance". We want to
||avoid "named instances" if possible for simplification of
||front-end client connectivity, but that may not be
||possible. I thought the documentation said that each node
||could have up to 16 "virtual servers". Can someone
||explain this CLEARLY???
||

Hello,

Don't feel bad. It is confusing at the beginning.

It appears that you have a two node Win2K3 cluster and want to install four
virtual instances of SQL Server 2000. At any time you want to run two
virtual instances of SQL Server 2000 on each node.

You can have upto 16 virtual SQL Server 2000. This is correct. But out
these 16, only one can be default and other 15 will be named instances. Or
you can have 16 named instances. You cannot have more than one default
instance.

To avoid confusion, you can have all the virtual instances of SQL Server
2000 as named instances. (If you want default instance then you can have
one default and three named instance).

Also, remember that each SQL Server 2000 virtual server requires its own
disk resources that cannot be used by other instances. So, both your
instances cannot use the the same drive I: and K:.

HTH,

Uttam Parui
SQL Server Developer Support Engineer, MCDBA, MCSE, MCT
Product Support Services
Microsoft Corporation

This posting is provided "AS IS" with no warranties, and confers no rights.

Are you secure? For information about the Strategic Technology Protection
Program and to order your FREE Security Tool Kit, please visit
http://www.microsoft.com/security.

Hi - Yes, your understanding is correct.
As mentioned by Uttam, you need to have dedicated drives for each virtual
sql server. Since you will be having 4 named instances (4 virtual sql
servers), you need 4 different disk drives for storing log and data
folders.

Regards,
Chandra

--------------------
||Content-Class: urn:content-classes:message
||From: "Michael [multnomah]" <anonymous@discussions.microsoft.com>
||Sender: "Michael [multnomah]" <anonymous@discussions.microsoft.com>
||References: <0b9601c3da26$06fafbb0$a001280a@phx.gbl>
<6f2Kc5i2DHA.3088@cpmsftngxa07.phx.gbl>
||Subject: RE: Default vs. Named Instance vs. Virtual Server
||Date: Tue, 13 Jan 2004 16:15:17 -0800
||Lines: 26
||Message-ID: <03a801c3da33$7d2b8fe0$a601280a@phx.gbl>
||MIME-Version: 1.0
||Content-Type: text/plain;
|| charset="iso-8859-1"
||Content-Transfer-Encoding: 7bit
||X-Newsreader: Microsoft CDO for Windows 2000
||Thread-Index: AcPaM30pX5ZNX+UkQOeDjaFQ+ug9oQ==
||X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300
||Newsgroups: microsoft.public.sqlserver.clustering
||Path: cpmsftngxa07.phx.gbl
||Xref: cpmsftngxa07.phx.gbl microsoft.public.sqlserver.clustering:12494
||NNTP-Posting-Host: tk2msftngxa14.phx.gbl 10.40.1.166
||X-Tomcat-NG: microsoft.public.sqlserver.clustering
||
||Thank you very much Chandrasekar and Uttam for clarifying
||this situation more for me. So, it looks like I want to
||use all named instances for consistency, with a virtual
||server required for each named instance. If I understand
||you correctly, a separate virtual server is REQUIRED for
||each named instance: 1 to 1. I can not have one virtual
||server for QAT1, with two named instances, and one virtual
||server for QAT2, with two named instances. Also, if I
||understand you correctly, each named instance must have
||its own set of drives. So, I would need the following
||configuration:
||
||[VIRTUAL SERVER NAME]\[NAMED INSTANCE]@[DATA DRIVE]
||SQLCLV-QAT1-1\INS1 @ D: [LOG H:]
||SQLCLV-QAT1-2\INS2 @ E: [LOG I:]
||SQLCLV-QAT2-1\INS1 @ F: [LOG J:]
||SQLCLV-QAT2-2\INS2 @ G: [LOG K:]
||
||I would have liked to have only two virtual servers, with
||two named instances each, but you are telling me that this
||is not possible. I would also have liked to have only one
||data and log drive for each node, QAT1 and QAT2, four
||drives total, but this also does not seem possible.
||Please confirm that I am understanding this correctly.
||
||
||

Another option that you have is to use Polyserve MxDb for sql server
based on a shared file system. All the instances could share the same
drive, or two drives for that matter (db and log) and it will enable
you to scale to more than 4 nodes clusters.

Tomer Meshorer
Database Engineering
Polyserve Inc (http://www.polyserve.com/sql)


[quoted text, click to view]