SRM 5.8 Array Based Replication

SRM 5.8 Array Based Replication

January 5, 2015 4 By Eric Shanks

If you plan to use Array Based Replication for your SRM implementation, you’ll need to install and configure your Storage Replication Adapter on your SRM Servers.  The SRA is used for SRM to communicate with the array to do things like snapshots, and mounting of datastores.

Pair the Arrays

Once your SRAs have been installed in both your sites and you’ve gotten the arrays replicating, you’ll want to pair the arrays in SRM so that they can be used for protection Groups.  Open the “Array Based Replication” tab in the “Site Recovery” menu of the web client.  Click the Add button.

Here, we’ll want to add a pair of array managers (one for each site).

arraypair1

Select the sites that you’ll be workign with.

arraypair2

Choose the SRA type.  If you only have one SRA installed, only one option should be available.  In this case, we’re using EMC RecoverPoint.

arraypair3

Now, we need to configure the manager with the IP Addresses, and names as well as an account that has enough privileges to create snapshots, as well as mount and unmount LUNs.

arraypair4

 

Now we’ll configure the opposite site’s array manager as well.  Same rules apply.arraypair5

Once we’ve configured the array managers we can enable them which will make them a pair that replicate to each other.

arraypair6

Finish the wizard.

 

arraypair7 - Copy

 

Protection Groups for Arrays

When you create a protection group for virtual machines being replicated by array based replication, you will give it a name as usual, and a site pair.

PG1

Choose the Protected site and the that it’s an Array Based Pair.  Select the pair.

PG2

 

Select the datastores that contain the virtual machines.  All of the virtual machines on this array pair will be protected.

PG3

 

Summary

Array based replication does not take much additional effort for VMware Site Recovery Manager, but may take some additional planning to make sure your protection groups are in the right datastores.  Remember that all VMs in a datastore will be failed over together.