How to keep Linux and SQL DB VM's replicas in sync

[Alan_ORiordan]

Hi

I have a Windows SQL Server which I want to Re-IP to the DR site.

I also have a Linux, Oracle DB server which cannot be Re-IP'd natively other than by a boot script

My problem lies with how to get both these replication jobs to run continuously and run and finished so that they are both complete at around the same time or with the shortest possible time difference.

Currently I have a replication job for the SQL server and a separate job for the Linux machine. I also have a slow link (10Mbit) and a trial of enterprise plus edition. Should I have a WAN accelerator for each VM? If I can't have them both in the same job due to Re-IP settings how can I get them as close as possible to being in sync at the DR site at the same time?

Thanks for the advice

[veremin]

Why not to create one job selecting two VMs as it source, let your proxy to simultaneously process them, use Re-IP settings for SQL VM and pre-freeze script for Oracle VM?

[Shestakov]

Should I have a WAN accelerator for each VM?

WAN Acceleration is highly recommended for slow links like yours. And once you are using Enterprise Plus Edition, it`s available for you.
Thanks.

[Alan_ORiordan]

Vladimir

If I create a new replication job for both VM's:

1) For Network Mapping: If I add both mapping rules as the source VM's are on different hosts
2) For Re-IP: If I only create a RE-IP rule with the IP details of the Windows SQL VM, will Veeam detect the source IP address and apply the rule to the correct VM
3) Job Settings: Storage optimization: The replication job is created on the DR Veeam console and the target is the local DR Host so I assume LAN target is correct?
Guest quiescence: Select tick box (For Linux VM) - Have I understood this correctly, see final note at the bottom
4) I have mapped the VM's to the existing replicas on the DR host and Veeam has detected them properly
5) Guest processing: Enable application-aware processing (For Windows SQL). In the applications options for the Linux VM I have selected to disable application processing

The benefit of this single job is that it will have use of the WAN accelerator and I won't need a second one.

I don't support linux unfortunately a 3rd party does that side of things for us. Do you have an example of a Linux script and I could give it to them and what it will do and what happens if it isn't implemented. What is the difference between using that Vs native VMWare quiescence?

Thanks very much

[veremin]

After having thought about that for a while, I start to believe that changing IP via pre-free/post-thaw script is not a good idea, as it's likely to break communication with this VM during backup or, at least, until post-thaw script takes place.

[Alan_ORiordan]

Could someone confirm my settings for the mixed DB job please and specifically the VMware quiescence options.

I will leave the Re-IP operation for the linux machine handled by the boot up script that the 3rd party created then which will ping the Default Gateway, if it's not there then it will re-configure it's IP settings for the DR site

Thanks

[veremin]

If you're going to handle Re-IP for that VM via 3-party script, then, the described settings look solid.

[Alan_ORiordan]

OK thanks, I decided to turn off VMWare native quiescence as I don't think that is required for Linux and SQL is using MS application aware VSS.

It's convenient having the job that will do both VM's and the other two original non scheduled replication jobs that can do them each individually for troubleshooting.

Thanks

[veremin]

OK thanks, I decided to turn off VMWare native quiescence as I don't think that is required for Linux and SQL is using MS application aware VSS.

Yep, for Oracle VM it doesn't seem to be required. Thanks.

[dellock6]

Well, it's a long story to explain in a post, but for Oracle quiescence is not required if you just need to restore the database to the time of the backup.
Otherwise, enabling archive logs and placing the database in backup mode is preferred, so Oracle can properly write all transactions in archive logs and so offer point in time restores. But this can be done with scripts stored in Veeam without VMware native quiescence.

[Alan_ORiordan]

This new double VM job I created seems to be stuck on the Discovering replica VM step. It is the first time I have run this double VM job as I had to first reset the CBT information on the linux server. It has been on this for 25 mins so far. I will leave it running until it either produces and error or starts the sync.

Any other guidance in this scenario?

[Alan_ORiordan]

Nevermind it is continuing now, just surprised at the length of time for that step given other jobs

[Alan_ORiordan]

I have seen the job take a long time - 48 mins on the following step:

31/03/2015 07:17:35 :: Required backup infrastructure resources have been assigned

What happens here that could take so long or be delayed?

[veremin]

As it stands from the label, during that step the backup resources (such as proxy, etc.) are being assigned. Typically, it should just a few seconds. So, the experienced behaviour looks unexpected.

If the issue persists, it might be worth opening a ticket with our support team and letting them investigate it directly.

Thanks.

[Alan_ORiordan]

I just realised the delay was the time taken by the first VM in the job which is why this one had to wait for the same resources - proxy/WAN accelerator to become available.