Hello Richard,

This approach is definitely not a one size fits all, and exchanges flexibility for the speed, performance and less heavy resource usage. But like I said to Michael, some work arounds are still possible.

You could use docker persistent storage behind the containers and have them share a database, assuming the database allows multiple applications to use it.

You could also write a microservice to manage the data and maybe provide an rpc interface for other applications to connect to. But the simplest workaround could simply be to not use an embedded store in this scenario, and just use a distributed store like redis, or even a nosql database like mongodb.

What do you think? Do you think any other approach could work?