Traditionally, SRM protected VMs that were on datastores with active/passive array based replication. Therefore a failover meant removing the VMs entirely (shutting them down and unregistering them) and bringing up copies of the datastores on the remote array. In this scenario the source side and the target side had different datastores on different storage devices but the arrays replicated data between them so they could be used to fail VMs back and forth.
In Site Recovery Manager 6.1, VMware introduced support for managing virtual machines that are protected by "stretched" storage. In other words, managing VMs that are on VMFS datastores that are active on two arrays at once. Unlike active/passive, "stretched" storage was active/active. Both sides has the exact same copy of the data at the same time and therefore the same exact datastore was available on both sides at once.
The Everpure FlashArray SRA supports this configuration since the 2.x release of the SRA.
In order to allow for the discovery of stretched devices, you need to configure the SRM array managers in a specific way.
To use the Site Recovery Manager support of stretched storage, you need the enterprise license of SRM. The standard edition does not include this feature.
In the example below, I have a volume hosting a VMFS in a pod called srmPod. this pod exists on two FlashArrays so that the volumes in that pod are stretched. This pod exists on flasharray-m50-1 and flasharray-m50-1.
Login to SRM and create a new array manager. It is important to know which FlashArray is local to which vCenter. In the example environment, flasharray-m50-1 is local to vCenter-01 and flasharray-m50-2 is local to vCenter-02.
In the array manager configuration, verify which vCenter you are looking at first:
I am configuring the array manager on vCenter-01 first, so my local array will be flasharray-m50-1. I will name the array manager m50-1.
I will enter the FlashArray flasharray-m50-1 address and credentials in the local array:
and then flasharray-m50-2 in the peer array:
For the remote array manager in vCenter-02, I will do the opposite. Verify the vCenter and give the manager a name:
Now add the array local to that site:
And then the peer:
The wizard will then show all discovered array pairs.
Enable the pair going from flasharray-m50-1 to flasharray-m50-2 (in this example).
Note pairs with pod names are not relevant to this type of SRM protection. Pod-based pairs are for failing volume from OUT of a pod. This type of failover simply fails VMs from one vCenter to another without moving the data--it stays in the pod. So it will essentially fail over the VM I/O to the pod volumes to go from one FlashArray front-end to the other FlashArray front-end.
If configured right, the array pair will enable and discovered devices will appear.