"The shared-nothing architecture allows the system to work with very inexpensive hardware, and without any specific requirements on hardware or software. It also does not have any single point of failure because each component has its own memory and disk."
If it's a shared-nothing architecture and a node goes down, then isn't all the data that was on that node down as well? Isn't that like a single-point of failure, to be running with an inconsistent set of data because all the data on the downed node is unavailable?
Also, a question about shared-nothing architectures: If you cluster your data across several nodes, isn't that data considered shared? e.g. if you have a record on node 1, but a record on node 2 has a foreign key pointing to the record on node 1.