Hi,
Hyperswap is not correctly implemented by IBM for ESX clustering (so for more 95% of cases), still in 2021. We are trying to use it for our continuity plan but the volumes are swaping from a datacenter to another one without any control. In an ESX cluster, each ESX node realize I/O commands on the mapped volumes/datastores. These I/O commands are not correctly interpreted by the hyperswap algorithm causing the swaps. So with no correct reason an ESX from datacenter A will use a volume on datacenter B. So it implies latency performance problems even without any fault.
Another point not correctly handle by IBM Hyperswap is the split brain. The no-elected IBM array will shutdown, so even local volumes will be unavailable .