This is a must-read for dealing with large, realtime datasets with event sourcing, from the creator of Storm, a distributed realtime computation system.
"What makes scalable data systems difficult isn't the CAP theorem. It's a reliance on incremental algorithms and mutable state that leads to complexity in our systems."
"I turned CRUD into CR, split persistence into separate batch and realtime systems, and obsessed over the importance of human fault-tolerance."
"Mutability is really just an inflexible form of garbage collection (that also interacts poorly with the CAP theorem)."
 https://github.com/nathanmarz/storm/wiki #cap #eventsourcing #bigdata