NoSQL databases are the new breed of databases developed to overcome the drawbacks of RDBMS. The goal of NoSQL is to provide scalability, availability and meet other requirements of cloud computing. The common motivation of NoSQL design is to meet scalability and fail over. In most of the NoSQL database systems, data is partitioned and replicated across multiple nodes. Inherently, most of them use either Google's MapReduce or Hadoop Distributed File System or Hadoop MapReduce for data collection. Cassandra, HBase and MongoDB are mostly used and they can be termed as the representative of NoSQL world. This tutorial discusses the features of NoSQL databases in the light of CAP theorem.