There is a lot of unstructured and semistructured data produced by application such as social media programs, the IoT (internet of things), and logs, where you have a very dynamic schema. These data types may have different schemas from each set of records. Storing this data in a relational database could be a very tedious task. Everything has to be filed into fixed schema, which can either cause lots of null values or data loss. The nonrelational or NoSQL database provides you with the flexibility to store such data without worrying about fixed schema. Each record can have a variable number of columns and can be stored in the same table.

NoSQL databases can store a large amount of data and provide low-access latency. They are easy to scale by adding more nodes when required and can support horizontal scaling out of the box. They can be an excellent choice to store user session data and can make your application stateless to achieve horizontal scaling without compromising user experience. You can develop a distributed application on top of the NoSQL database, which provides good latency and scaling, but query joining has to be handled at the application layer. NoSQL databases don't support complex queries such as joining tables and entities.

There are various choices available for the NoSQL database—for example, Cassandra, HBase, and MongoDB, which you can install in a cluster of virtual machines; however, the public cloud-like AWS provides a managed NoSQL database called Amazon Dynamo DB, which provides high throughput and sub-millisecond latency with unlimited scaling.

You can use OLTP for a relational database, but it has limited storage capacity. It doesn't respond well to queries for large amounts of data, and those that perform aggregations as required for data warehouses. Data warehousing needs are more analytical than transactional. The OLAP database satisfies the gap of the OLTP database to query a large dataset. Let's learn more about the OLAP database.