RiskTech Forum

Teradata: Liquid Analytics Gets Real for Enterprises

Posted: 2 May 2014  |  Author: Scott Gnau  |  Source: Teradata


Last week I wrote how our industry is evolving toward a “liquid” approach to analytics where queries can be dispatched across one cohesive, interconnected and complementary architecture involving bi-directional movement of data to analytics and/or analytics to data. I used the analogy of an automobile – easy to drive despite tons of complexity under the hood – to convey how this should operate seamlessly across a wide variety of resources and analytic techniques so enterprise customers can extract the most value from their data.

 

 

 

 

 

 

 

 

But just like a fancy concept car that never goes beyond the aspirational drawing board, sketching a vision for liquid analytics isn’t much good if you don’t take those next important steps to bring that concept into production.  We at Teradata have been laying the foundation for the liquid analytics reality for a while, most notably by developing the Teradata Unified Data Architecture –  a multi-product, best of breed analytic environment that helped earn us some market leader nods in recent months from Forrester, Gartner and other respected industry groups.

Now in an evolutionary push to make liquid analytics a reality for customers, we’re bringing to market some key solutions to further operationalize the vision for vast query options across diverse analytic engines, file systems, storage techniques, procedural languages and data types.

You can get the full story from the news releases on our introduction today of Teradata 15.0, the latest version of the Teradata Database, and our release of Active EDW 6750, the next generation of the Teradata Workload Specific Platform Family’s powerful Integrated Data Warehouse (IDW) platform model. Read those releases and you’ll see how TD 15 helps fill in the big data lake with new, developer-friendly enhancements that include better temporal and geospatial analytics and upgraded capabilities for workload management, availability and supportability. Active EDW 6750’s numerous benefits, meanwhile, include enhanced processing memory space and deft use of in-memory via Teradata Intelligent Memory to help support thousands of concurrent users of hundreds of apps performing complex and varied work.

But here I’d like to call out one particular element that I consider the software fulcrum for the nerve center a liquid architecture will need in place to coordinate queries across so many complex resources and analytic options. We call it Teradata QueryGrid, a set of intelligent connectors and product capabilities – now in beta tests underway as part of TD 15’s enhancements – that allows you to submit a single query to a single system and have it use data and specialized analytics capabilities from many other systems to get the answer.

Designed for minimum IT intervention and duplication of data, QueryGrid orchestrates seamless and self-service access to data and analytic processing across different systems from within a single Teradata Database or Aster Database query. In doing so, QueryGrid facilitates bi-directional capabilities that minimize data movement and duplication by allowing processing to happen where the data resides. This is a key liquid analytics requirement, and QueryGrid allows pushdown processing not just to Teradata Aster and other databases, but also to open-source Hadoop. Teradata SQL-H previously available in Teradata Database 14.10, for instance, could not write data to Hadoop and could only retrieve data from Hadoop as entire files or partitions.  With the help of Hive performance improvements from our partner Hortonworks, the enhanced connection to Hadoop leveraged by QueryGrid in Teradata 15 now meets all of these increased criteria.

QueryGrid can even employ the Integrated Data Warehouse (IDW) and Integrated Discovery Platform as an orchestration engine to exchange data beyond the UDA to other data repositories and processing nodes.  A Teradata-Oracle connector, for instance, provides access to an Oracle Database to retrieve data or perform part of a query and retrieve the results.  As I’ve said before, the liquid analytics vision demands that individual companies embrace other vendors’ products if the end result is a better analytic environment for customers.
For all these reasons, I consider today’s introduction of QueryGrid as a game changing multi-system, multi-directional approach to analytics that reaps more flexibility, value and integration. QueryGrid will become even more powerful as we continue to optimize it over time. But its presence today means a lynchpin has arrived for businesses that don’t want to wait any longer to start leveraging all the data in their enterprise architectures.