Some of the most popular databases people are aware of are SQL-based and relational, MySQL, PostgreSQL, SQLite, MSSQL, Oracle, etc. There is now a rising trend towards non-relational databases. It is understood now that relational databases are not the be-all and end-all, for many purposes a relational model is not required, inefficient or just counter-intuitive.
Such no-relational databases are classified under the NoSQL banner, since SQL is most commonly associated syntax for querying relational databases.While technically NoSQL would include relation databases that don’t use the SQL syntax, the term NoSQL is most popularly used only to include non-relational databases.
There are many different non-relational databases available today, some popular ones being Cassandra, CouchDB, MongoDB, BigTable, Hadoop, Berkely DB, and the recently-released LevelDB by Google. Even Indexed DB the standard specification for database that can be used by web applications, is a non-relational database.
Currently each database has its own mechanism to query the data it contains. Now work has begun on creating a standard way to access the data no NoSQL databases.
The new specification is being called UnQL, for Unstructured Query Language, and is being developed by the creators of the popular SQLite database system, and the creators of CouchDB. They intend the query language to work with “JSON, semi-structured and document databases.”
By creating a standard way to access data from NoSQL databases, it is possible to easily create applications that are agnostic of the database software being used in the backend, allowing them to easily be plugged and swapped.
You can read more about the specification on its website here.