Teradata bulks up its universe, joins data warehouse to data lake

Dashboard engine for Hadoop

ThinkBig, Teradata's Big Data consultancy business, has introduced a Dashboard Engine for Hadoop (DEH). It says: "dashboards and online analytic processing (OLAP) roll-up summaries have been around for decades in the business intelligence market. But... no-one has delivered OLAP summary analysis from multi-structured Big Data at scale", until now.

It handles many forms of multi-structured data, such as click stream, mobile phone logs, sensor and JSON (JavaScript Object Notation).

Teradata says DEH distils terabytes of data per hour into billions of aggregates, which can then be queried by business users. Ron Bodkin, ThinkBig's president, said: “[It] offers organisations the ability to get summarised Hadoop data in under a second, without having to go through transformation and modelling processes.”

ThinkBig provides installation and expert implementation services to help customers use DEH.

Data Warehouse Appliance 2800

Teradata says this product – full name: Teradata Data Warehouse Appliance 2800 – is an integrated data warehouse with more than twice the compute power and up to four times more data capacity in a single cabinet than the previous version.

Teradata Data Warehouse Appliance 2800

The 2800 is optimised for in-memory processing of analytics and is a ready-to-run rack crammed full of components. It can be used as a data warehouse, data mart, disaster recovery system or analytical sandbox for test and dev.

It features dual Intel Xeon Haswell 14-core CPUs, up to 512GB of DDR4 memory and SuSE Linux. Teradata says the system uses Intel’s vector instructions and on-board cache to improve performance. The interconnect is Teradata's BYNET V5 on InfiniBand.

It has Teradata Intelligent Memory for "automatic, real-time, usage-based monitoring of data access", used "to place the most heavily used data in memory". Total available memory scales automatically and linearly as the system scales out, with up to 12 nodes in a cabinet and 200 nodes overall mentioned.

It's claimed that database administrators never have to worry about data placement or data reorganisation, because memory and storage are designed to work together as a unit.

The storage is based on Dot Hill's Ultra48 AssuredSAN array and available 2.5-inch drives are 300GB, 600GB, 900GB and 1.2TB. The system is scalable to more than 54 petabytes, with 1.2TB drives. The tray design has disks both at the front and the back, which helps to improve capacity.

The rack cabinet includes the Teradata Database with its shared-nothing cluster design, managed servers and optional back-up technology. Customers can install an optional second production system, test and development capability, or back-up and restore storage hardware within the same cabinet.

Check out a DWA 2800 datasheet here (pdf).

Teradata Labs boss heads off to Hortonworks

Scott Gnau, president of Teradata Labs since May 2011, is leaving to become CTO at Hortonworks, the IPO'd Hadoop distributor. We understand he provided direction for R&D and sales support activities for Teradata's integrated data warehousing, Big Data analytics and associated products. He also drove the investments and acquisitions in Teradata's technology related to the solutions from Teradata Labs.

In a canned quote, Gnau said: "Hortonworks is enabling the entire Hadoop ecosystem and accelerating enterprise adoption of the platform through its products' centralised architecture." He's passionate about data etc, etc.

Teradata will look to replace him with another BI/BD (Business Intelligence/Big Data) heavy hitter.

Availability

QueryGrid's Hortonworks functionality will ship in May. The beta version of the Dashboard Engine for Hadoop is currently available. GA is intended in the third quarter. The Teradata Data Warehouse 2800 Appliance is available worldwide and is compatible with Teradata Database 14.10 and later. ®