A bane of our existence, when working with dbs, is that we always conflate storage with query optimization. When we talk about databases, we unconsciously think

A bane of our existence, when working with dbs, is that we always conflate storage with query optimization. When we talk about databases, we unconsciously think of both concerns. This is traditionally instilled in us as that makes the most sense starting out, but almost always what ends up happening is your dataset needs to be shared among multiple applications, each with its own optimization requirements. And that's when things start to go haywire. At some definition of scale, a better design is to have a storage layer that is not really concerned with query optimization, and allow different applications to design their own query systems. Said another way, even when you are working in RDBMS's, it's good to acknowledge storage and query as two separate concerns.