NoSQL Databases

In recent year all new types of database have emerged for different reasons such a scalability, flexibility and usability. The underlying reasons are that much has changed in the underlying hardware and the way that databases are used. Some notable changes are:

  • hard-disks are getting bigger, memory banks and SSDs are getting cheaper
  • distributed processing has become easier, scaling out more important
  • the architecture of the hardware is changing: CPUs have more cores and more caches

This has lead to a new class of DBMSs which are sometime called NoSQL databases. This is a heterogeneous class of databases that includes key-value stores, document (or JSON) stores and graph stores.

The research in WISE on NoSQL databases conerns the following topics:

  • Indexing in NoSQL databases: The changed use of disks, the changed caching archtitectures and the often distributed processing of storage and procesing requires new indexing techniques that can optimally function in these contexts. The research focuses on developing new and adapted indexing structures for this.
  • Query Languages for NoSQL databases: The JSON data model is similar to the XML data model, but is at the same different enough to warrant its own query language. The research focuses on investigating, designing and implementing query languages for JSON. This involves languages similar to SQL (or XQuery) as well as languages similar to datalog, a simplified version of Prolog specialized for formulating queries.
  • Schema definition languages for NoSQL databases: Although JSON often is seen as a lightweignt semi-strucutred (or even unstructure) data model, and this is in fact seen as one of its advantages, there is a rising trend to introduce powerful schema defintion langauges that allow constraining the contents of a JSON store. The research focuses on comparing and designing such languages.

Related Publications