HBase is the open-source implementation of Google's BigTable architecture. It strives to scale to billions of rows and millions of columns, all distributed across an inexpensive cluster of commodity servers. HBase's primary storage system is HDFS, Hadoop's distributed and replicated file system. In addition HBase has a native interface to Hadoop's Map Reduce framework which allows for easy development and subsequent execution of batch jobs that can scan entire tables. HBase fits into a new category of database systems often summarized under the term "NoSQL" - or non-relational databases. While there are quite a few system that comprise the term they often only overlap marginally or have nothing in common but the fact that they abandon traditional models found in relational database system (RDBMS) for specialized architectures that are built to solve a specific problem. For HBase this is storing a virtually endless amount of data while ensuring that write and read performance stays constant, no matter how large the dataset is.