Huge data production has opened a platform for proliferation of various database solutions. In turn, it is challenging to identify the appropriate task specific database. The paper comprises of: 1) giving a clear picture of various categories of NOSQL database, features of each database 2) Selection of database store 3) Characteristics comparison of NoSQL over SQL. 4) Experiment is conducted in two phases: evaluation over behavioural characteristics and experimentation for functional characteristics. Various available databases are stacked up against the behavioural characteristics like partitioning, sharding, querying, tolerance, reliability, indexing and many more. Databases ranked high are then measured by implementing basic functional operations like read, write and delete. Experimental results are represented in two ways one is graphically for time consumption for each operation and other is tabular ranking representation of all these against each other. This paper concludes that NoSQL databases apparently perform better as compared to SQL databases in for read operation while for others it varies depending upon the operation. A few are instantiating fast but are slow during write operation and vice versa. Not all Database Management System can be used purely for one type of database architecture, and selection of database depends solely upon the nature and needs of application.