The Boulder BI Brain Trust


June 2008 Archives

Kognitio brings flexibility to complex analytics

| | TrackBacks (0)

The BBBT pondered the past and future of the British firm, Kognitio, which has deep roots into the database community. In 1992 ex-Teradata people formed WhiteCross Systems, which merged with Kognitio in 2005. It currently employs over 70 persons at Bracknell, UK, building a group in Chicago to serve the US market, and has over 30 customers across several industries, like finance, retail, and telecom. We were briefed by Sean Jackson, VP Marketing, John Thompson, EVP and General Manager of US, and Roger Gaskell, Chief Technology Officer.

The focus is high-performance analytics at a low cost. The scalable MMP architecture executes on any collection of x86 blade servers under Linux interconnected with TCP/IP. There is no indexing and no materialized aggregations. The system is said to perform well with high data volumes and high workload concurrency.

A distinctive of Kognitio is that they have a full SQL functional row-oriented database engine that achieves good performance with complex query processing. This runs contrary to industry wisdom that only column-oriented engines can achieve such performance. The magic comes from: spreading data evenly across many nodes, extensive use of in-memory processing, generation of x86 machine code for query processing, mature cost-based optimizer, and smart pipelining of temp data among the nodes. The pipelining reminded me of the old Y-bus unleashed.

Another distinctive of Kognitio is that they have three ways of delivering their product/service. First, they licence their database as a normal software product. Second, they will sell a complete appliance as a hardware/software bundle. And third, they offer data warehousing as a service, hosted in their own data center and third-party data centers.

Now that is flexibility! Check them out! These brits have more to offer than Newcastle Brown.



This page is an archive of entries from June 2008 listed from newest to oldest.

May 2008 is the previous archive.

July 2008 is the next archive.

Find recent content on the main index or look in the archives to find all content.