In a computing world that moves towards a more dynamic approach to servers and services, having VSM server as a stateful service is a roadblock to being able to provide high availability and redundancy for a server cluster. The modern approach to high availability is to have stateless services that read configuration and state from external services. If a service fails, trash the entire VM/container the service runs in and start a new VM from a known state.
Additional reading/inspiration: https://12factor.net/
Related bugs: Bug 6127 - Support HA over more than two vsmservers