Consul was the original recommended “first class” storage backend for Vault.
However, it was complex for admins, and the Consul “all data in RAM at runtime” model wasn’t great for Vault either.
So Integrated Storage was created. For a while, Integrated Storage was still under development, and a bit too fragile to depend on for real. But now it has matured to the point where HashiCorp are positioning it as the first choice.
I don’t think anyone should be creating new Vault clusters using Consul anymore.
You can happily have a 5 node Integrated Storage cluster using Vault OSS, but I’m not sure there’s much point in doing so unless you are running on very unreliable machines.
In a Vault OSS cluster, only one node is ever actually doing useful work - the rest are just waiting to take over for it, and proxying client requests to the one active node whilst doing so.
Enterprise adds the concept of “performance standby”. It is not multimaster. Rather, it means that the standby nodes are able to satisfy read-only client requests themselves, without needing to send the requests to the active node.