Teradata Highlights 'Data Driven' Strategy at PARTNERS 2014

Teradata's 2014 PARTNERS conference kicked off Monday morning with several announcements of big-data software enhancements, partnerships, and ambitious examples of data-driven technologies transforming business and marketing.

Teradata introduced customers to Connection Analytics, an analytics tool that finds unique patterns among interrelated connections and influences among people, machines, products, and processes. Tracking interactions between entities can mean understanding how people are influenced in a social networks analysis, tracking flow of money in a fraud analytics environment or influence between machines based on interactions in an Internet of Things scenario.  

Traditionally, analytics looks at entities as discrete units of analysis, says Chris Twogood, VP of product and services marketing for Teradata, but Connection Analysis is unique in that it is looking at the linkages of all information from all entities.

"What we're seeing as a key trend is there's a lot of dialogue about capturing big data and storing big data. But the problem with this is a lot of big data is raw, without any insider value." Customers are looking for new and different techniques to uncover unique patterns from their data. With Connection Analytics, business analysts and data scientists can analyze a massive amount of data and look for patterns.

"It literally will come back and identify key patterns," says Twogood. "It will identify where there are connections that might imply some fraudulent ring between people calling within a call network. It might look at different patterns where you see implications of a small change in pricing that will have better price elasticity within a market. It may say that by targeting this cluster of people within an overall network that I can drive more viral marketing."

Connection Analytics is built on Teradata Aster Discovery Platform, MapReduce, and Graph engines. It leverages machine learning, sentiment analysis, and more than 100 pre-built algorithms.

"This is about enabling the data to give us answers to questions we don't even know we should be asking. That's what these analytics lead us to do, and that's a huge driver for being able to get value from big data."

In-memory optimization
Teradata also announced advanced engineering and system efficiencies for in-memory optimization inside the Teradata database. The enhancements, available in early 2015, will reduce load on memory bandwidth with new algorithms, and by running queries without leaving memory, thus reducing data movement.

The enhancements include new weightings for measuring the temperature of data, meaning how frequently data is accessed. Weightings distinguish between tactical and strategic workloads, according to the press release, and "hot" data (frequently used data) is placed into memory faster, thus aligning data in-memory with business priorities.  

"When you move to in-memory from disk it is 30,000 times faster, but logical input/output (I/O) is always a bottleneck, so memory becomes the new bottleneck," says Twogood. "We're doing advanced engineering to go beyond that. We want to continue to drive performance because demand for analytics is just getting bigger and bigger."

“Teradata is relentlessly dedicated to engineering a smarter, simpler way to leverage memory and CPU to drive performance,” said Scott Gnau, president of Teradata Labs in the press release. “Blindly throwing additional memory at a problem has diminishing returns, particularly when it comes to big data. Teradata’s sophisticated approach automatically and efficiently places the right data in-memory to get the performance they need for the best cost.”