Drill has different virtues than Impala or Hive or Spark SQL or Tajo or any
of the gazillion other SQL on Hadoop offerings. In particular, the ability
to work with schema-less data, dynamically typed SQL and efficient vector
representation are all Drill strengths.
Note also that Impala is a closely held open source project while Drill is
community based and driven. That means that you can join in on Drill and
contribute so that you have a voice in where it goes while Impala is
whatever Cloudera wants it to be, end of story.
MapR is particularly customer-driven and we hear demand for all the
different options. Hive is (by far) most widely used. Impala has some
usage, but surprisingly little. Drill has strong pull as well. Being
customer driven, support for one option does not mean lack of support for
another. It simply means that customers are asking about that option.
Regarding whether Drill is a live project, take a look at the JIRA
220 bugs reported and 140 fixed in the last month sounds like a pretty
lively project to me.
On Sun, May 18, 2014 at 10:24 AM, Ashish Gupta <[EMAIL PROTECTED]>wrote: