NoSQL database-as-a-service vendor Clusterpoint SIA is using the year’s hottest word – “disruption” – to describe the latest version of its namesake product. In version 4, Clusterpoint has added computational capability to the database engine, enabling data to be shared among multiple nodes, with code-execution capability on each node.

Clusterpoint is positioning the new release has a full replacement for relational DBMS, saying it supports enterprise-class performance with near-infinite scalability, thanks to an underlying parallel architecture. The Riga, Latvia-based company, which was founded by a former Google search engineer, entered the US market in June with a novel pay-as-you-go pricing model that charges only for resources used.

The new release is intended to address a common problem with document-oriented NoSQL databases: The volume of information they must process exceeds the capacity of the database to deliver it in a timely fashion. This limits the ability of customers to achieve the promise of near-real-time analytics. Clusterpoint says it can now deliver speed, and at lower cost by scaling out rather than up.

Clusterpoint already had a distributed foundation with support for persistent data structures and numeric, full-text and geospatial indices. The new version supports ACID (Atomicity, Consistency, Isolation, Durability)-compliant multi-document transactions and a computational model based on JavaScript, which the company said are two features that are “often ignored by other NoSQL databases in the market today.”

Computation is managed by a new JavaScript-compatible data manipulation and query language called JS/SQL that the company termed “universal non-proprietary.” JS/SQL can be used to transform data during the retrieval stage as well as to access indices as raw objects.

The company is providing more details of its unified database/computing architecture on its blog.

Riga, Latvia photo by Bryan Ledgard via Flickr CC