This goes along with full text search, the ability to reduce entity sets based on a textual LINQ query server side would speed up apps and reduce costs for customers.

I was interested in researching table storage because of the scalability, but no text search means I can not even run basic text search? Accessing entities based purely on a PK/RK pair is a handicap. Search should be an integral or even extensible part of this platform. It is kind of ironic that moving "up" the storage ecosystem to a relatively expensive hosted platform means I give up something like this. Pulling down whole datasets just to filter them locally is a win for MS with more transactional fees and bandwidth charges along with reduced computational overhead, but is a drawback for customers for the very same reasons.