Data Connector for Druid.io

[exalate-issue-sync[bot]]

Feasibility: I was looking into details:

• no HDFS-like API: that would be easy way, but it has a dedicated rest-based client API to load data and make queries
• columnar store: which is good for us from implementation point of view, easy 1:1 mapping
• i am not sure if it has Java API/client - the whole system is based on JVM, but during my review I did not find any explicit client (this is not showstopper, since they have also REST API)

From technical point of view:

• In this context we need to implement a new persist layer for fetching data from druid
  • I did that long time ago for Tachyon - it is nice 1week exercise.
• I am not sure about saving data into Druid right now (it is more complicated but feasible)

Result: it is feasible, but will need non-trivial technical effort

[exalate-issue-sync[bot]]

Michal Kurka commented: This project looks interesting: https://github.com/himanshug/druid-hadoop-utils

This might be the right approach for data ingesting because people complain about the SELECT API being slow: https://groups.google.com/forum/#!topic/druid-development/uWGWCPlg0c8

[h2o-ops]

JIRA Issue Migration Info

Jira Issue: PUBDEV-3720 Assignee: New H2O Bugs Reporter: Venkatesh Yadav State: Open Fix Version: N/A Attachments: N/A Development PRs: N/A