Hi, I'm using vSphere 5 and have a requirement to set-up SRM for a handfull of VMs (approx 10). I'm thinking of implementing a single recovery plan, protection group ...etc.
Can anyone detail why > 1 recovery plan will be useful? i.e. cases for different test scenrios or DR.
I'm assuming a single RP can fit all purposes, but have heard that this is not good practice, and would like further detail of types of recovery plans.
Thanks
Hi,
For a small enviromnent like the one you described, I don't think more than one RP is required.
Like Chris said, nn most cases per application RPs are created.
Much less common, but RP can be based on certain piece of the infrastructure (RP per storage array, RP for specific server room).
P.S. good point about real DR / Planned Migration, but it's more difficult to maintain, because you will need to perform every change twice to keep both of RPs in sync. Also, you will need to test both of them to make sure all is working as planned.
Michael.
It depends on your applications being protected. I typically use multiple recovery plans to recover individual applications that may have failed, so that I can granularly recover a subset of the environment, along with a larger recovery plan for a "smoking crater" type disaster.
Just remember: You can use only one recovery plan at a time to recover a protection group. If multiple recovery plans that specify the same protection group are tested or run simultaneously, only one recovery plan can failover the protection group.
http://www.vmware.com/pdf/srm_admin_5_0.pdf
Hi agreed. Basically, the site only has one protection group, but I'm trying to think of scenarios where a number of recovery plans may be required, and one of these being run depending on situation, for example:
(1) RP for a planned migration
(2) RP for a DR event at primary site
Are there any other cases that may need to be addresses?
Thanks
Hi,
For a small enviromnent like the one you described, I don't think more than one RP is required.
Like Chris said, nn most cases per application RPs are created.
Much less common, but RP can be based on certain piece of the infrastructure (RP per storage array, RP for specific server room).
P.S. good point about real DR / Planned Migration, but it's more difficult to maintain, because you will need to perform every change twice to keep both of RPs in sync. Also, you will need to test both of them to make sure all is working as planned.
Michael.
Chris is right.
A single recovery plan for a total disaster is a must, but individual recovery plans for specific applications is a very good practice.
Not only will it allow you to recover a specific app, without having to fail over the entire datacenter, but it will also let you test your proceedures much faster.
If your SRM plan is failing on a specific application, it's nice to not have to wait for all of your other apps to fail over to test a change you've made.