As the data landscape changes, so must the databases used to gather, store and analyze the rich information within them. The emergence of web 2.0, social networking and user-contributed content have pushed traditional relational database management system (RDBMS) architectures to the limit. Consumer-facing Internet companies such as Google, Amazon, Yahoo! and LinkedIn are able to scale by using NoSQL (Not Only SQL)* data stores. Sites like these have realized that the controls of an RDBMS limit performance and lack the ability to conduct massive numbers of concurrent reads and writes. In cases where data must be stored and analyzed — and doesn’t require the controls and rigidity of RDBMS — CIOs can learn from what’s worked for hugely successful web sites, and therefore create architectures that promote flexibility and speed.
There is no reason to consider unseating RDBMS as the guardian of transactional data and its integrity, but organizations struggling to cope with the recent explosion of unstructured data must consider separate non-relational database solutions where warranted in order to enhance the data ecosystem with performance and scale.
The goal of this report is to explain the role of NoSQL databases within the context of the database market in general, to provide information regarding primary vendors and projects, to illustrate how NoSQL databases provide the most benefit to enterprise and web information systems architects, and to provide recommendations for evaluation, implementation and development.
Table 1: Overview of Relational and Non-Relational Data Stores
* We prefer to use this version of the term, rather than No SQL, as we believe that defining these solutions as simply “they don’t do SQL,” is outdated and, frankly, no longer accurate.