Kind of depends, storage DRS is for IO and space. So if you have multiple datastores, it will balance the space usage between them and the IO. You will also no longer need to pick a datastore to put your VMs in (if you are using multiple).
It really depends on if you find that you move VMs about manually as free space is running low in the datastores, if not, then not really much point. Or if you use automated provisioning of VMs then you do not need to think / implement something to pick the correct datastore. The downside to the DRS moving of VMs between datastores is if you have snapshots on your arrays, you will start using up a bit of space if it moves VMs around.
With the active / active, is it true active / active (read and writes to both sides) like vplex / metro cluster (Dell) GAC (Hitachi) or whats just come out in the primeras (HPE)? Or dual active / passive (volumes replicated between each array with different active on each side)?
For IO, If the former, no need, if the latter, possibly.