Suddenly thought of a problem: about CQRS separation

Most of the system, is the query more, c less, then why we do not put Q in memory, and C directly manipulate the database?

Just as enode,c in memory, and Q in the database, of course q is also easily extended to the memory cache.
Two Enode case demo, can make C performance improved, and let the performance of Q decreased (always read the library, is not used to cache).

I think Enode because the default is equipped with a memory database, Denormalizer directly to the memory, and then directly in the memory database above query.
This memory database to be able to index, if you can participle, like SOLR, it is even better.

Suddenly thought of a problem: about CQRS separation