ACID is hard to scale. But isn't NoSQL/NoACID the lazy way around the problem? Why not solve the ACID scalability problem instead of just going NoSQL? It's a hard problem to solve, but here are a few ideas on how to solve it.  Alexander Thomson andDaniel Abadi have some ideas.

 

http://dbmsmusings.blogspot.com/2010/08/problems-with-acid-and-how-to-fix-them.html

http://db.cs.yale.edu/determinism-vldb10.pdf