On Fri, Feb 7, 2014 at 2:50 PM, Vasileios Vlachos <vasileiosvlachos@gmail.com> wrote:
So basically all is down to which keys the node will end up being responsible for. It could end up holding 100% of the data (that is including replicas if I read it right). Looks like ensuring that a node is capable of holding 100% of your data is necessary...

Right, which is not very common. Also man that one node ends up with a lot of data on it, that you have to cleanup.

This is why people tend to use the multi-DC approach.