TrustRadius
Apache Drill vs PrestoDB
https://www.trustradius.com/nosql-databasesApache DrillUnspecified8.54101
Anson Abraham profile photo
February 13, 2018

Apache Drill vs PrestoDB

Score 8 out of 101
Vetted Review
Verified User
Review Source

Overall Satisfaction with Apache Drill

Apache Drill was being used initially to evaluate running queries on data stored in multiple data stores (hDFS, postgres, cassandra). We were testing it out, over the use of PrestoDB. But saw that Drill also supported HBASE and other engines. it was from that aspect, used Drill to query data across the multiple datasources, as more ec2 instances were created, since some of the data was also stored in s3.
  • queries multiple data sources with ease.
  • supports sql, so non technical users who know sql, can run query sets
  • 3rd party tools, like tableau, zoom data and looker were able to connect with no issues
  • deployment. Not as easy
  • configuration isn't as straight forward, especially with the documentation
  • Garbage collection could be improved upon
Garbage collection is a pain. tuning GC for apache drill isn't so easy. So it's a trial and error basis, for each different use case. but once it's properly been configured, then querying w/ drill becomes easier. creating multiple nodes is easy and adding them to existing cluster fairly easy. However configuration isn't so straight forward. Also use of SSDs is not easy to use w/ Drill. SSDs have dropped in price, that configuring drill to use it, isn't straight forward.
  • Configuration has taken some serious time out.
  • Garbage collection tuning. is a constant hassle. time and effort applied to it, vs dedicating resources elsewhere.
  • w/ sql support, reduces the need of devs to generate the resultset for analysts, when they can run queries themselves (if they know sql).
  • prestodb, impala and apache phoenix
compared to presto, has more support than prestodb.
Impala has limitations to what drill can support
apache phoenix only supports for hbase. no support for cassandra.
Apache drill was chosen, because of the multiple data stores that it supports htat the other 3 do not support. Presto does not support hbase as of yet. Impala does not support query to cassandra
if Presto comes up with more support (ie hbase, s3), then its strongly possible that we'll move from apache drill to prestoDB. However, Apache drill needs more configuration ease, especially when it comes to garbage collection tuning. If apache drill could support also sparkSQL and Flume, then it does change drill into being something more valuable than prestoDB
if you're doing joins from hBASE, hdfs, cassandra and redis, then this works.
Using it as a be all end all does not suit it. This is not your straight forward magic software that works for all scenarios. One needs to determine the use case to see if Apache Drill fits the needs. 3/4 of the time, usually it does.