"But losing the SAN is just as likely as losing any normal server."
Patently false. Statistically, SAN's that employ 2 or more controllers have significantly lower failure rates (by at least an order of magnitude). SAN's are designed with no single point of failure (SPOF). Dual power supplies, dual (or more) controllers, passive disk back-planes, etc.
[ ... ]
Virtual SANs (good ones) after controller failure (in a Virtual SAN concept that's a host failure) will move VMs to another host and re-balance storage after some time. Technically it's self-healing. What SAN with a baked controller does in such a situation? It texts a message to admin to replace physically damaged controller... and sits for hours waiting for double failure to put SAN down completely :)