Cluster (VCS) - replication of SQL2005 between nodes with independent data bases

[DiegoT]

Hi,
I have formed a Cluster with 2 nodes. Each Node has installed SQL 2005. I need to replicate the data bases between both nodes. Can make this replication without adding a shared disc. What is the way but simple to realise this configuration? Thanks,

[rhanley]

Hi Deigo,

This can be accomplished by using either VVR (Veritas Volume Replicator) or 3rd party hardware replication (i.e. EMC Mirrorview, HTC TrueCopy, etc....).

In Summary, you will create what is called a Replicated Data cluster. This means that the nodes will be located in the same cluster, but they will not share storage and a cluster-aware replication solution will be put in place to handle that aspect.

You did not state which version of SFW-HA you have installed, so I will assume the latest version (5.1). I have included a link to a SQL Solutions pdf document which includes detailed steps for configuring SQL 2005 in a Replicated Data Cluster (Section 4 - Replicated Data Clusters).

If you have any specific questions after reading this, please don't hesitate to post them to the Forum, otherwise, it may be best to open a Support Case if you have more than just a few questions or need more indepth knowledge than can be provided in the forum.

exftpp.symantec.com/pub/support/products/Storage_Foundation_for_Windows/306326.pdf

I hope this helps,
rhanley

[DiegoT]

Hello rhanley,
Thanks for your answer.
I confirm to you that if I am using Haa for Windows 5.1.
I am following the points mentioned in docuemto of the Link, but at the time of creating the Dinamic Disk Group it does not accept the disc to me of my server for being an IDE.
My case: I have 2 server, that do not share disc, only have its disc IDE formatted with a single primary unit (E:).
I have 1 SQL 2005 in the node 1 and another SQL 2005 in node 2, and I only need that the data base of node 1 is talked back towards node 2 at the time of failover. Everything what talks about Shared Disk does not apply in my servers configuration, but requests me the documentation to realise the configuration of the Replicated Data Cluster. You can say to me like itself with the estrucura that I have Navy.
I appreciate your attention, Greetings

[rhanley]

Hi DiegoT,
    I'm not sure if I understand this exactly, but I know there is one issue where it will fail to add a dynamic disk to a disk group if it is on the same BUS as your system disk. This will report the error:

V-76-58645-525 - The disk can not be imported as a member of a cluster dynamic disk group because it is not on a shared bus.

If that's the case, you can check out the following Technote: http://support.veritas.com/docs/266015

To use a disk that is on the same BUS as the sytem disk, you will need to run a single command: vxclus UseSystemBus ON

After that, you should be able to add the disk to a Clustered Disk Group.

If this is not your issue, can you provide the exact error you get when trying to add the disk to a Disk Group?

Thank you,
rjhanley

[DiegoT]

Hi rjhanley,
The error is: "Warning V-47-1-216 . List of selected disks contain IDE disk. Cluster disk group cannot by created."
this message appear when add the system disk to a new cluster disk group.
I read your message and investigate to create cluster disk group and dont found any option to do this whit the disk IDE that has my server.

Perhaps the best thing is to replicate the bases of another way, (ex. with Replicate SQL) and maintain them in the local disc of each SQL, and setting une gruop for cluster SQL without mount disks, only controlate events from SQL service to mantein only one instance UP of SQL server.

That is correct?

Thank you,
DiegoT

[rhanley]

Hi DeigoT,
  Sorry about that, I missed the comment that these were IDE disks.

The issue is that clustered Disk Groups use SCSI reservations which of course are not supported with IDE disks. I suspect that you don't have any issues creating a standard secondary Disk Group (non-clustered)?

If so, you could always configure secondary disk groups and configure replication, but like you stated, you will not be able to control these in the cluster which means that any failover would require manual intervention (you would need to perform the replication migration from with VEA or the command line before you could bring up SQL on the other node). If these servers have SCSI adapters, it may be best to simply add a new disk to each system as this will be a far better solution.

Configuring this inside and outside of the cluster would require a pretty strong knowledge of VCS and VVR and furthermore, all failovers would have to occurr manually or there would need to be extensive scripting in place to automate certain actions.

rjhanley