Kognitio to offer virtual cube analytics
- 15 February, 2011 09:13
Analytical database provider Kognitio has augmented its flagship WX2 database so that it will be able to offer data cubes, eliminating the need to run a separate data warehouse, the company says.
The technology, called Pablo, consists of a set of metadata that allows the database to build out OLAP (online analytical processing) cubes on the fly, so they can be analyzed when called.
Pablo is best suited for customers who already work with data cubes and want to reduce "the overhead from creating those cubes and repopulating them on an ongoing basis," said Sean Jackson, Kognitio vice president of marketing.
Typically organizations render data cubes, which are multidimensional rows of data taken from live databases, in separate data warehouses, where they can be analyzed without slowing performance of the source databases.
However, database servers have gotten so much faster in the past few years that they could now easily take on additional analytical duties, argued John Thompson, Kognitio CEO for U.S. operations. He points to the emergence of in-memory databases, in which entire databases run from within the working memory of one or multiple servers, speeding response time.
Pablo works on a similar principle. It can create data cubes upon request, using a set of metadata within an additional database table. "So when someone wants to come in with Microsoft Excel, or a Business Objects or MicroStrategy [business intelligence tool], they don't have to hit a middle tier," Thompson said. "With Pablo, you can still work with the cubes, but you are getting a huge performance advantage."
With the technology, Kognitio offers an MDX (multidimensional expression) connector, allowing any outside applications supporting MDX to interact with MX2's virtual cubes as they would with any other data cubes.
The virtual cubes can do everything that regular cubes can, and would have no limit of dimensions, Thompson said. And because WX2 supports high concurrency, it can be run across multiple servers, which means that the additional duties of building virtual cubes shouldn't slow performance, assuming that the user has enough hardware in place, Thompson said.
In addition to cutting the costs of setting up a separate data warehouse, Pablo offers the additional advantage of allowing an organization to analyze live data, as opposed to waiting until the latest data is offloaded to the data warehouse, Thompson said.
The company plans to release a full production version of the software in June, and will run beta programs with select customers in the next few months.