Does an IRF stack (to use HPE terminology) provide adequate resiliency for a cluster of VMware hosts?

Apologies if this sounds more like a question on an exam. I’ve been in IT for about a decade and exposed to most common infrastructure but more managing and maintaining than designing. I wanted to expand my knowledge a bit.

I’m just curious how the general perception is on core switch redundancy and whether the board feels an IRF stack (or whatever the Cisco equivalent is) provides enough resiliency for running a cluster of hypervisors assuming each host is connected to at least two switches in a stack? Also whether or not using two separate blades in a chassis would be a better or worse option? It’s something I’ve seen a lot in campus type scenarios. What are the chances an entire stack or chassis would fail?

What would people recommend as an alternative? Two completely separate stacks? Stretching the cluster over multiple stacks / chassis? Interested in hearing some opinions / best practices on this. I imagine cost is the limiting factor in a lot of this as well.