Apache Drill 1.0 tears into data, with or without Hadoop

Drill 1.0 queries Hadoop data via SQL, but may have a life of its own outside of the framework

Apache Drill 1.0 tears into data, with or without Hadoop

How many ways can you mine Hadoop data with plain old SQL queries? Lots, but Apache Drill, one of the most versatile of the bunch, has hit its 1.0 milestone, and it's set to work with more than Hadoop alone.

Drill is an open source implementation of Google's Dremel/BigQuery engine. It was designed to query multiple kinds of data, including unstructured JSON, structured CSVs, the Apache Parquet format for columnar storage, schemas in Apache Hive's Metastore, and more conventional structured data sources.

Aside from needing nothing more than ANSI SQL to run queries and using conventional ODBC/JDBC connectors to allow access to the data, Drill doesn't require schemas to be defined for the data before querying. This means less involvement from IT to prepare data for analysis; anyone with a suitable tool set and the proper permissions can plug in and begin querying.

Jack Norris, chief marketing officer for MapR -- makers of a Google Capital-funded Hadoop distribution that offers Drill as a supported component -- described how the 1.0 release was earned by bringing the project to feature completeness for the sake of production environments. The core feature set for Drill has been fixed for some time.

But another key aspect of Drill is its future apart from Hadoop, which is only one of many different kinds of data sources that Drill can query. Many of MapR's partners who are business intelligence providers, said Norris, are providing their perspective and support for Drill. "At the very least, Drill expands their flexibility," he said, "and their ability to access some of these new data formats directly, so that it's bringing self-service [data access] to these systems as well."

As questions arise about how much enterprise uptake can be realistically expected of Hadoop, it's prudent for projects conventionally thought of as Hadoop-centric efforts to have second lives. That includes not only Drill, but other Apache data-crunching projects like Spark. If Hadoop's flame turns out to be bright but short-lived, Drill won't be lacking for other areas to tear into.