I understand the benefits of using a VM swapfile datastore for VMs that will be replicated. However, if I have a large cluster (500+ VMs), is it OK to have all vSphere hosts in the cluster use the same VMFS volume for 500+ swap files? This is assuming that memory would never be so constrained that the swap files would be used. If it is OK to do this, would it be a concern if the VMFS volume was on a host/datastore that didn't support ATS Locking via VAAI?
It would be a good alternative if you could create full memory reservations on VMs to be used by SRM. However, it appears that vSphere still creates a 0K .vswp file when using a full memory reservation, which will still need to be deleted by SRM on failover which will slow the process down.
Thanks,
-MattG