MSCS Cluster

[Former Member]

Hi,

We are planning to build the following system - and i have couple of questions

Node A - ASCS, Db Instance, ERS installed locally

Node B - ASCS. Db Instance, ERS Installed locally

What could be the time taken for failover if Node A - Db fails? Will I loose the updates to Db?

What could be the time taken to failover if Node A - ASCS fails?

Thank you!

[Former Member]

Ok, let me clarify the issue.

NodeA:

Has two main cluster resources. First one is Database and second one is SAP. Before the NodeA fail, all the database and its all resources are running on NodeA, as well as SAP application server and SAP ASCS resources.

Please find the MSCS components on NodeA, below;

• SAP application server

• Replication server

• Enqueue server + message server (belongs to virtual host)

• Database resources

and NodeB:

• SAP application server

• Replication server

Let's assume that NodeA is failed. Then all the resources move to the NodeB;

NodeB:

• SAP application server

• Replication server

• Enqueue server + message server (belongs to virtual host)

• Database resources

And all the DML transactions will be lost, if you face with a failure on NodeA.

I hope that I clarified the issue.

Best regards,

Orkun Gedik

[Former Member]

>>Node A - ASCS, Db Instance, ERS installed locally

>>Node B - ASCS. Db Instance, ERS Installed locally

Both nodes has same applications? something is wrong. In windows MSCS you need to configure cluster group for each application. ie. ASCS group, DB group you can move this group one server to other or both can run on any server. The scenario is very much clearly mentioned in the installation guide.

If the cluster fails over during business hours and the updates what are in process will terminate and roll back. About the time to take to failover is managed by cluster admin, as you can define the cluster failover time and attempts before moving cluster group to other node.

[Former Member]

Hi,

Firstly, it depends to what db system is you are running on. I am assuming that you are running on Oracle.

What could be the time taken for failover if Node A - Db fails? Will I loose the updates to Db?

If you are talking about, updates under DB13, you will not lose anything. But, if you are talking about some uncompleted transactions in memory, at Oracle level, you may lose some updates.

What could be the time taken to failover if Node A - ASCS fails?

It depends to the OS you are running on. For example, HACMP on AIX is moving resources to NodeB and continue to run. Means that database and application server resources are starting up on NodeB. On Windows 2008 cluster, only database resources are moving to NodeB and continue to run on NodeB with one application server.

Best regards,

Orkun Gedik