Teradata upgrades data warehouse


Teradata Corp. has overhauled the top and bottom of its data processing platforms as well as added capabilities to help Hadoop users in Teradata environments.

On Monday the company showed off the EDW 6700, which it says has up to a 40 per cent increase in performance over the previous 6690/6650 models thanks in part because one version runs new eight-core 2.6 GHz Intel Xeon processors.

There’s also up to a 10 times increase in traffic performance because the 6700 uses switched fabric InfiniBand network technology for connectivity to storage, the first time in this class of chassis. Previous models used Ethernet.

The 6700 can handle “hundreds of thousands of users concurrently, ” said Tasso Argyros, the company’s senior vice-president for global product deployment and strategy.

Like the previous models, the 6700 can be equipped with both solid-state and regular hard drives, with technology that can load more frequently used “hot” data onto the SSDs and leave less used data on HDDs.

Depending on the model, the 6700 has up to eight times more memory capacity per cabinet (256 GB of memory per node).

On the software side, the EDW 6700 includes version 5 of Teradata’s Bynet interconnect,

The other new hardware is the Data Mart Appliance 670, a low end appliance for test development and small data marts which Teradata says is a single processing unit version of the 6700. It also uses the new Xeon CPU and can be configured with mixed storage.

No pricing was announced.

Curt Monash of Monash Research, an industry analyst who specializes in analytic technologies, said in an interview the 6700 is a straightforward annual refresh of the line. As for adopting InfiniBand, he noted that Teradata used to be negative on the technology. However, an increasing number of vendors are backing it.

For Teradata users who also have Hadoop storage clusters, the company has created Teradata SQL-H, which allows staff to use Teradata’s SQL engine to do SQL queries on Hadoop data.

“What SQL-H does is allow you from within Teradata to see all of the tables that exist in Hadoop,” said Argyros.

Last year Teradata released SQL-H for its Aster analytics database.

Now users and query Aster, Teradata and Hadoop in the same SQL statement, Argyros said, “so you don’t need to be an Hadoop expert. All you need to know is how to use a BI tool or SQL.”

To make things easier, Teradata Studio, a database administrator, now has a smart loader for transferring data between Hadoop and Teradata.

Good connectivity between analytic relational databases and Hadoop is important, said Monash “There are many things that relational databases do better than Hadoop does, so you want the technologies to work together.”


Please enter your comment!
Please enter your name here