According to VMWare, the issue is in Storage vMotion on a slowpath. The downtime, while acceptable for one VM, is unlikely to be acceptable for every VM on the host, especially if they are all sending traffic through the slowpath.
As long as the slowpath is programmed to reconnect when disconnected, there is no need to reboot DSVA. You just have to set the retry in the slowpath reconnection logic. There is no guide for doing this in the VMware documentation, but it is good practice to have such a loop in case slowpath gets disconnected for other reasons, such as a crash.
VMware has no specific plans to support Storage vMotion for slowpaths at this time.
The suggested workaround is to disable Storage VMotion on the DSVAs.