The survey of nosql technologies sounds more like a rationale for another O'Reilly bookshelf!

I mean, there are definitely arguments that can be made for alternatives to relational databases, but without those special cases, in such an early stage of maturation and without standards, it's worthy for pursuing by the hardy cowboy or blissful novice.

As the nosql technologies do mature and standards emerge, I think it should be expected that they will be subsumed into existing database products as new features.