An Entity System involves having an entity table. An entity table simply stores all of the entity ids. These ids have the properties that they are auto-generated and unique. They identify a specific entity in the system, which could be anything. If you have an entity id, then all you do is search that entity table to see if it exists. Now the application for this is for distributed systems where you are looking for data and you contact the many servers' entity table for what you are looking for.
It would be sweet to have an entity id associated with Actions, Resource States, Accounts, and etc., so you can look for what you want in one place and associate things with that one entity for flexibility, sort of like an Entity Component System (ECS). I'm not completely sure what defines an ECS, so maybe this is all wishful thinking and I created a variant of an ECS already. Anyway, it would be nice to just be able to query many servers with the entity you are looking for if the original server you were assigned to doesn't have it or even be able to find what server/cluster the entity may be located. I'm just getting started with knowing what the ECS is, so this may already be done, but I've been thinking of maintaining a statistical data structure called a Bloom Filter on entity ids, so that it would near constant time access to looking for whether an entity id exists in the server instead of doing a query for one of those guys, which takes time even if it doesn't exist. It seems that Google BigTable already does something like this with a Bloom Filter according to the Wikipedia page on Bloom Filters, so it may be possible for me to do this.