Click here to monitor SSC
SQLServerCentral is supported by Red Gate Software Ltd.
 
Log in  ::  Register  ::  Not logged in
 
 
 
        
Home       Members    Calendar    Who's On


Add to briefcase 12»»

multiple instance sql cluster with different windows configuration Expand / Collapse
Author
Message
Posted Monday, April 1, 2013 1:46 PM
SSC-Enthusiastic

SSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-Enthusiastic

Group: General Forum Members
Last Login: Tuesday, July 15, 2014 8:13 PM
Points: 124, Visits: 750
We have a two node cluster. In order to utilize the max resources we are planning to add more instances to this cluster by adding few more nodes. Let us assume--

server A - cluster node--100 GB ram and 24 core cpu (active)
server B - cluster node--100 GB ram and 24 core cpu( active)
server C - new node -- 50 GB ram and 12 core cpu( passive)
server D - new node -- 50 GB ram and 12 core cpu (passive)

What we are planning to do is to run sql instance of server C and D on server B and make use of the resources. When server A fails we need it to failover to server B and if this happens we can manually failover the two instances running on server B to node C and node D respective. Is this possible? I know in cluster all the nodes should have same resources. Need suggestions?
Post #1437595
Posted Tuesday, April 2, 2013 11:11 AM
SSC-Addicted

SSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-Addicted

Group: General Forum Members
Last Login: Yesterday @ 10:27 AM
Points: 401, Visits: 2,004
Not sure. Wait for Perry's reply.
Post #1438015
Posted Wednesday, April 3, 2013 2:32 AM


Grasshopper

GrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopperGrasshopper

Group: General Forum Members
Last Login: Tuesday, September 30, 2014 10:10 AM
Points: 15, Visits: 126
SQL Cluster 2008 need not the exact same hardware but rather compatible hardware. I'm currently running a 2 node cluster with Node A (64proc, 256GB) the primary and Node B (24proc, 128GB) as the passive. Now where this becomes a problem is if Node A crashes can Node B handle the load? In our environment it can but queries do become slower and its okay for us because it is only temporary until we get Node A back up.

It all depends on your load and can the weakest link (Node in this case) hold all your resources?

-king
Post #1438207
Posted Thursday, April 4, 2013 3:48 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Today @ 8:47 AM
Points: 6,488, Visits: 13,962
deep_kkumar (4/1/2013)
We have a two node cluster. In order to utilize the max resources we are planning to add more instances to this cluster by adding few more nodes. Let us assume--

server A - cluster node--100 GB ram and 24 core cpu (active)
server B - cluster node--100 GB ram and 24 core cpu( active)
server C - new node -- 50 GB ram and 12 core cpu( passive)
server D - new node -- 50 GB ram and 12 core cpu (passive)

What we are planning to do is to run sql instance of server C and D on server B and make use of the resources. When server A fails we need it to failover to server B and if this happens we can manually failover the two instances running on server B to node C and node D respective. Is this possible? I know in cluster all the nodes should have same resources. Need suggestions?

In theory, yes all nodes should have the same configuration. This is especially important to the following

CPU family
Base hardware (i.e. server model)
Patch level
Network configuration (including the NIC type)

Now, in my current role we actually have a 6 node cluster that has 2 "high spec" nodes and 4 lower spec nodes (but all 4 the same). The key to remember here is how the clustered instances will be distributed, just because you install a clustered instance into a multi node cluster it doesn't mean you should then install it across all the cluster nodes.
Only install the instance across nodes where you want the resources to be highly available.

So, having 2 nodes that are slightly lower spec does not present an issue but you should abide by the above regarding CPU family, network configuration and more importantly the patch level. The software version and level are the most important here.


muthyala_51 (4/2/2013)
Not sure. Wait for Perry's reply.

Lol, thank you


-----------------------------------------------------------------------------------------------------------

"Ya can't make an omelette without breaking just a few eggs"
Post #1438703
Posted Thursday, April 4, 2013 7:16 AM
SSC-Enthusiastic

SSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-EnthusiasticSSC-Enthusiastic

Group: General Forum Members
Last Login: Tuesday, July 15, 2014 8:13 PM
Points: 124, Visits: 750
Thanks Sailorking and Perry!

Post #1438782
Posted Thursday, April 4, 2013 2:38 PM
SSC-Addicted

SSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-Addicted

Group: General Forum Members
Last Login: 2 days ago @ 8:53 AM
Points: 411, Visits: 1,310
Dropping my 2 cents here.

This may not be true for 2+ node Cluster, but I always try to put more RAM on server than the total amount of SQL instances I have if, hypothetical scenario, I need to run those on one node only.

For instance:

RAM on Server A: X GB
RAM on Server B: Y GB

So... RAM Instance C + RAM Instance D < X + Y

That's in order to avoid the situation that two instances have more allocated RAM than actual server's RAM.
Post #1439016
Posted Thursday, April 4, 2013 6:26 PM
SSC-Addicted

SSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-AddictedSSC-Addicted

Group: General Forum Members
Last Login: Yesterday @ 10:27 AM
Points: 401, Visits: 2,004
Perry,

I was not able to follow this statement, can you please elaborate--

The key to remember here is how the clustered instances will be distributed, just because you install a clustered instance into a multi node cluster it doesn't mean you should then install it across all the cluster nodes.
Only install the instance across nodes where you want the resources to be highly available.
Post #1439063
Posted Friday, April 5, 2013 3:54 PM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Today @ 8:47 AM
Points: 6,488, Visits: 13,962
muthyala_51 (4/4/2013)
Perry,

I was not able to follow this statement, can you please elaborate--

The key to remember here is how the clustered instances will be distributed, just because you install a clustered instance into a multi node cluster it doesn't mean you should then install it across all the cluster nodes.
Only install the instance across nodes where you want the resources to be highly available.

Say you have a cluster with 4 nodes ServerA, ServerB, ServerC and ServerD. You install a clustered instance to ServerA, you don't necessarily install it to the remaining 3 nodes.


-----------------------------------------------------------------------------------------------------------

"Ya can't make an omelette without breaking just a few eggs"
Post #1439517
Posted Wednesday, April 10, 2013 9:38 AM
Mr or Mrs. 500

Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500Mr or Mrs. 500

Group: General Forum Members
Last Login: Tuesday, October 21, 2014 11:20 AM
Points: 554, Visits: 1,624
What is the purpose of additional server nodes if you're not going to extend the SQL cluster there? If the servers with SQL go down, you're down. (Unless the server is being used for other things, like a network fileshare.)

Post #1440902
Posted Wednesday, April 10, 2013 10:06 AM


SSCertifiable

SSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiableSSCertifiable

Group: General Forum Members
Last Login: Today @ 8:47 AM
Points: 6,488, Visits: 13,962
dan-572483 (4/10/2013)
What is the purpose of additional server nodes if you're not going to extend the SQL cluster there?

The whole idea of creating a multi node cluster is to make cluster administration easier. One 6 node cluster is easier to manage and uses less network name and IP resources than 3 x 2 node clusters.

Just because you have 6 nodes you do not necessarily want to extend your instance across all of the nodes, you'd generally leave at least one or as passive failover partners on hot standby for key instances.


dan-572483 (4/10/2013)
If the servers with SQL go down, you're down. (Unless the server is being used for other things, like a network fileshare.)


If i have 6 nodes with standard quorum setup of node majority with witness and 4 nodes fail then the cluster will be offline anyway. Whats the likelihood of 4 servers failing all at once, are they on a separate geographic site. you have to factor all this into your cluster design. Also, install 12 instances across 6 nodes and see how long the SP installer takes to run


The SQL Server installation quotes the following

Installing a SQL Server 2008 R2 Failover Cluster
Then, you run Setup again for each node you want to add to the cluster.


-----------------------------------------------------------------------------------------------------------

"Ya can't make an omelette without breaking just a few eggs"
Post #1440920
« Prev Topic | Next Topic »

Add to briefcase 12»»

Permissions Expand / Collapse