Serious SRM Confusion!!!

[adamhodgkins83]

Hi all,

I have a question about SRM 4 to do with a project that I am working on at the moment.  I've come into the project half way through and don't know much about SRM so not the ideal start, but anyway....!

The first point of confusion is trying to find information on the replication side (EMC VMAX - SRDF), there are 2 sites, and there are 2 domain controllers on the protected site, 2 at the recovery site, in the same domain and they are Active/Active.  Along with the DC's, there are also SRM and vCenter servers at each site that are up and running.  Should these 'Protected Site' servers reside on datastores that are being replicated with SRDF or does it not matter?

Some ideas are that as everything is on the same domain so will have different names, IP's etc. that as long as they are not in the protection groups to get restarted should a failover be initiated, that this will work fine...  Other ideas are that this setup is a complete no-go and these servers should be stored on datastores that are not being SRDF'd...

Any help/advice would be greatly appreciated as always!!!

Thank you

Adam

[weinstein5]

The first point of confusion is trying to find information on the replication side (EMC VMAX - SRDF), there are 2 sites, and there are 2 domain controllers on the protected site, 2 at the recovery site, in the same domain and they are Active/Active.  Along with the DC's, there are also SRM and vCenter servers at each site that are up and running.  Should these 'Protected Site' servers reside on datastores that are being replicated with SRDF or does it not matter?

Each site will need there own SRM and vCenter server - so there is no need to replicate those machines - in vCenetr as part of the recovery the VMs will be part of the recovery site vCenter sever -

[adamhodgkins83]

Hi David,

Thanks for the info!  The next thing is that the client had previously been told that the VM's that do not need to be replicated (DC's, vCenter, SRM server etc...) must be on a datastore that is not being replicated with SRDF, and that if they are then this will not work properly, and may cause corruption...  Do you know if this is the case or is it OK to have everything on the SRDF datastores and just not map them to anything at the recovery site to be restarted?

Thank you

Adam

[hosterman_cody]

You can choose to replicate any VMs that you want and any or all of them (or none) can be managed by SRM. It depends on how you set up your SRM protection groups. There are some RDF limitations where your configuration might require failing over devices that arent protected by SRM, but that can be easily avoided by changing the configuration of RDF to increase the allowed granularity of failover--but this scenario is rare anyways.

For more information on SRM 4 and SRDF, please refer to this techbook:

http://www.emc.com/collateral/software/technical-documentation/h7061-srdf-adapter-vcenter-srm.pdf

Hope it helps

Cody

[mal_michael]

You can have unprotected VMs on replicated datastores.

This is not recommended with SRM5, since this will cause issues with failback.

Michael.

[adamhodgkins83]

Thanks for the info!

We are using 4.1, and I don't think that they are planning on upgrading to 5 any time soon but I will make them aware of this!

Cheers

Adam

[hosterman_cody]

What Michael said is definitely true. But if the VMs are on different replicated datastores entirely from VMs that are protected by SRM failback wont be affected (assuming they aren't grouped into the same datastore group).

[adamhodgkins83]

That was my thinking, I'm trying to figure it out so that the VM's that don't need to be protected and won't be covered in the protection groups are on completely separate datastore(s) than the machines that do need to be protected.

This of course depends on how generous the storage team want to be but it's good to know the pros and cons of both methods!

[AnthBro]

The most important part of the SRM design phase when using storage based replication is ensuring you plan your storage inline with your recovery goals while continuing to provide the performance your machines required.  I would try and involve the storage team as a key stakeholder in any SRM project.  The majority of SRM projects I do have actually been generated from EMC Recoverpoint sales, or vica versa.  Shuffling machines around datastores are almost always required for optimum configuration, there are always exceptions but these are few.