i have same config with several Clusters as i create Clusters per CPU Architecture. Westmere,ivy,Sandy,Haswell. i don't use EVC.
Hm wouldn't sense now an Kind of Problem. 1 Point is that theire is additional Heartbeat through storage per Cluster. But on other Side theire is more less overhead for HA Master/Slave per Cluster .
For overally optimize, maybe have a look if you have latest Supported Firmware on Storage Boxes. CHeck if theire is support for Alua/RR and that you use ATS/SCSI Reservation instead of nativ SCSI Reseration. If you fully have ATS Support, you could upgrade to have ATS-Only nativ .
No challenge to share datastores between clusters , No EVC constraint , Host hardware , scsi reservation etc.
I am using it upto some extend in production environment with multiple hosts. cross VCs is also possible (tested in lab).
I have shared datastores between clusters and didn't have any apparent issues in the past, but I wasn't sure if it was best practice.
Thanks for the replies.