Containers are great for ephemeral applications where you do not need to save state. Databases, as a general rule, are not useful if ephemeral. So try to set up persistent storage for these databases, where you can 'claim' disk storage that is hopefully somewhat independent of the pod lifecycle.
But what about those cases where you are starting with no active database instance? Or do you need to bring in a new database server into a database cluster and make sure it is populated with the latest and greatest data? How do you accelerate these new systems into being productive? There are ways to seed caches so that there is a minimal warm-up period that can greatly reduce the time needed to have a productive system. And there are tools like MySQL's InnoDB Clone plug-in that can rapidly populate a barebones system.
And you need a way to observe the performance of these database instances. What factors do you need to watch such as cache hits and disk throughput are obvious performance factors. But there are other factors such as queries not using indexes that are performance killers or redundant data structures that consume memory excessively.
This is a database in the cloud talk for folks who are not database folks. You will discover the wild and wooly world of databases where they can be tamed to work in the cloud. There are a lot of factors that need to be considered for efficient database operations if you want smooth and fast performance from your database instance