Total Pageviews

Wednesday, September 18, 2013

With Help, Teradata Speeds Up

Teradata, the venerable data analytics company, is about to offer capabilities in data analysis that seem to be fast, powerful, versatile and accessible. It is a somewhat cobbled-together approach, involving other companies, but it is a start.

It could be a big deal, because this is not a combination seen previously from the analysis offerings of the biggest companies. There is a lot of expectation and hype around the analytic and predictive potentials of Big Data, but in general it remains hard to work with. When an idea does seems promising, either from smaller start-ups or new initiatives formed by big companies, products have not come out. When they do, it is not clear how quickly they will be accepted by customers.

Teradata is among the largest businesses in data analytics, with a big customer base. This new product, part of its latest release, has a fast kind of so-called in-memory processing, and uses a library of over 1,000 types of different analytic tools. It is used in conjunction with an open source analytic language, R, that is popular with many researchers, academics and analysts.

“There are no size limitations here,” said Chris Twogood, the company’s vice president of products and services marketing. “We have e-commerce customers generating 50 billion items a day who are using this. A manufacturer of silicon wafers is looking at 100,000 different versions of design.” For lots of data, he said, the product also can use disk memory, which is somewhat slower.

In this move to serve what Mr. Twogood called “mere mortals” in the business world, however, Teradata did need some help. The R analytics it is using come from another company, Revolution Analytics. Customers of Teradata that want to use the service will have to subscribe to Revolution separately. In addition, 600 of the 1,000 tools in Teradata’s analysis library come from a third company, Fuzzy Logix.

In-memory computing is already done by SAP and Oracle, among others, but Mr. Twogood said the Teradata system would prove more efficient by allocating data by how frequently it is used for problem solving.

The analysis tool, a part of Teradata’s 14.1 product release, also shows a change in what customers want from Teradata’s products, he said.

“The world is going to more predictive analysis,” he said. “It has been about report generation, now predictive is more pervasive. It is a change from known quantities to unknowns - what question is the data showing we should ask.”