Relational databases have been ubiquitous in the enterprise world for some time now. In terms of storing data in tabular form, relational databases are fast, time-tested, and powerful.
However, over the past years, NoSQL databases have risen in popularity. The term NoSQL encompass a number of different technologies - such as document databases, key / value stores, graph stores, and wide-column stores.
Each of these technologies solves a different problem, but in general they are designed to cope with the scale and agility challenges that relational databases were not built to address. I will go over the specifics on how NoSQL databases differ from SQL.
SQL databases require a schema to be defined before data can be stored. The system must know not only what attribute must be stored - i.e., a phone number or a name - but what type of data that attribute will be - i.e., text, number, etc. If a team decides to change the data that is stored, a new schema must be designed and the current database must be migrated. This can be a slow process for large data sets. In the world of Agile Development, an SQL database can represent a significant deterrent to fast iteration.
NoSQL databases, on the other hand, do not have need predefined schemas. Teams can iterate over their product, design, and business model without having to worry about changing the database.
Relational databases rose to popularity before the era of Cloud. It is no surprise that relational databases were designed to be deployed on a single server. Due to the large volume of information that today's businesses must deal with, horizontal scaling has become a constraint. The answer is vertical scaling, in which capacity is added to a database in the form of multiple servers, not a single massive server.
NoSQL databases were designed to be hosted on multiple servers. They have built-in support for spreading data across servers, ensuring consistency of the data, load balance request for resources, and combining the data once more when a query warrants it.
Most NoSQL databases also support automatic replication. This allows for high availability and responsive disaster recovery.
NoSQL databases are increasing in popularity and have real benefits to teams that require non-tabular data structures, the ability to iterate quickly without having the database slow that down, and the ability to scale a database vertically.