Fruit of the Loom: Teradata ladles out free bowls of Hadoop data soup-strainer

Data about data on latest heffalump offering

As of Thursday, Teradata is giving out free dollops of big-data analytics for Hadoop, via the latest release of its recently purchased Loom tool.

The data warehouse giant is expected to release Loom 2.3, updating the data-provenance and metadata suite it bought in July from Revelytix.

Loom is used to prep data coming into BI tools from Hadoop from different sources and in different types.

Loom 2.3 understands a greater number of data objects, Teradata said, and it features an updated UI and has been extended to work on Cloudera’s CDH 5.1 and MapR.

Previously, it had worked on Hortonworks’ Data Platform (HDP).

The most intriguing development is Teradata’s decision to take the price tag off of Loom 2.3 with a free – if limited – edition and a paid-for enterprise version.

The free edition comes as Teradata also unveiled a hosted Hadoop service. Due at the end of the quarter, it’ll be paid for monthly, although pricing is not yet fixed.

Hadoop as a service follows Teradata Cloud (data warehouse as a service) in November last year.

Teradata is going deep on Hadoop to gain an edge over other rivals, such as Microsoft, Oracle and IBM, who have been cuddling up to Hadoop from the SQL side of life.

Chris Twogood, veep of product and services marketing, called Teradata’s decision to give away a version of Loom for free a “big change.”

Teradata is usually re-assuringly expensive.

According to Twogood, Teradata is trying to help drive Hadoop’s enterprise adoption, and we "are committed to a unified data infrastructure [as] we want to be able to leverage that information to share it across multiple best of breed solutions".

Hadoop has suffered from relatively limited use cases – it’s been largely used to crunch rather mundane server web logs.

Twogood estimates customers are beginning to heft more data types on to their organisations’ Hadoop clusters — data including POS sales records, call center records, video, and information from sensors.

“With Hadoop, people are just dumping data it in so you have no structure,” Twogood said. “You have all these data types that have no relationship. Data warehousing is a heavy lifting process where you clean the data, so when you hit it, it’s already ready for sharing."

Loom is a data-profiling tool that tracks the lineage of data and its movements in and out of Hadoop systems. It manipulates the data before its ready for analysis, too, summarising it and creating new views where necessary before it hits the BI layer.

The free and paid editions give you a meta-data repository, APIs, profiling and discovery, and the ability to track data’s lineage. The difference is on how far you can go: you can only go on 20 different operations, whereas the enterprise edition is unlimited. Also missing is Kerberos, automated linage tracking on Hive and MapR.®