In some specific scenarios such capabilities are required:
- e.g. in development a single (and costly) service instance may need to be shared by multiple apps consuming it for the purpose of testing
- in production, when defined by architecture, a few MTA solutions are sharing consumption of a service instance but have otherwise independent lifecycle
In MTA modelling one can achieve this by:
A: ignoring the validations for ownership: each MTA's operations would "manage" (create/update/...) the service instance
B: Determining a single MTA to own the service instance, and all others would expect it to be already existing when deployed