NoSQL technologies often use different models for failover, high availability, and multi-data center support than traditional databases. In this session we will review approaches for different products such as Dynamo, BigTable, Yahoo PNUTS, CouchDB, and MongoDB. Discussion will include topics such as eventual consistency and CAP theorem.

The CAP Theorem states that we cannot achieve consistency, availability, and tolerance to network partitions in a distributed data system. In this session we will review how traditional database deployments have traded off these items, and how NoSQL sometimes offers a different approach to the problem. Different non-relational database products take different approaches: for this reason we will examine how different products deal with replication, multiple copies of information, consistency, failover, high availability. We will examine the implications of different approaches when working with multiple data centers and also in terms of performance.