Joshua Resnick - Fotolia
The chase to "SQLize" Hadoop continues. Independent Hadoop distribution provider MapR's latest version of Hadoop supports Apache Drill 0.5.0 SQL-on-Hadoop analytics. It provides yet another means to bring SQL users into the sphere of Hadoop, joining Tez, Spark and others.
A plus for Drill is that it can work with schema-less data, schema-on-the-fly data and nested data, while using familiar SQL commands. The widely used JSON data format is a likely target for much of the Drill work to come.
"A major problem for Hadoop has been that it doesn't have a good query mechanism," said Jack Norris, chief marketing officer at MapR. He positioned Apache Drill as a particularly favorable choice here, because it can query complex data in native formats.
Besides supporting Drill, the new MapR 4.0.1 software supports a slew of SQL-on-Hadoop technologies that include Hive, SparkSQL, Impala and Vertica.
The release also includes core platform updates. Norris said the latest distribution supports mixed-mode configurations of Hadoop 2.0 YARN applications running along with MapReduce 1.0 applications. Such capabilities are important as organizations look to efficiently run existing batch-oriented MapReduce jobs while at the same time bringing on new "2.0" applications.
Administrators can explicitly deploy different jobs within MapR clusters, according to Norris. With the MapR update, "they are not forced to run everything on YARN on 'Day 1,'" he said. In the near term, such dual-mode operation may become the standard procedure.
Lights out at InfiniDB
Database maker InfiniDB last week announced it was filing for Chapter 7 and closing up shop. Word came by way of an open letter from CEO Bob Wilkinson, who said the company's InfiniDB software will continue to be available in open source form.
Originally formed as Calpont, the company was renamed InfiniDB earlier this year, taking the name of its InfiniDB database, which was first released in 2010. Categorized as an analytic columnar database management system with parallel query support, the database struggled to find wide use. Among noted use cases were digital advertising and communications.
Wilkinson became CEO in April, after serving as vice president of engineering and chief operating officer. He succeeded Jack McDonnell, who served as CEO for just eight months.
InfiniDB faced difficulties as columnar databases quickly moved from the periphery to the mainstream. IBM, for example, has ramped up columnar enhancements to its flagship DB2 database. Wilkinson addressed these in his open letter, writing, "as a startup on the leading edge of technology, business can be challenging especially as differentiation in the market is murkier and competition from bigger entities increases."
Learn about new big data tools
Listen to a podcast on SQL and Hadoop
Catch up on summer reading on Hadoop 2.0