Setting up and using Apache Druid
After learning hardware recommendations and software requirements, you add the Apache Druid (incubating) service to an HDP 3.x cluster.
Recommendations:
- Assign the Overload, Coordinator, and Router to one or more master nodes of size AWS m3.xlarge or equivalent: 4 vCPUs, 15 GB RAM, 80 GB SSD storage
- Co-locate the Historical and MiddleManager on different nodes from the Overload, Coordinator, Router, and Broker, and on nodes of size AWS r3.2Xlarge or equivalent: 8 vCPUs, 61 GB RAM, 160 GB SSD storage.
- Do not co-locate LLAP daemons and Historical components.
Software Requirements:
- A MySQL or Postgres database for storing metadata in a cluster for a
production
You can use the default Derby database installed and configured by Ambari if you are using a single-node cluster for development.
- Ambari 2.7.0 or later
- Database connector set up in Ambari
- HDP 3.0 or later
- ZooKeeper
- HDFS or Amazon S3
- YARN and MapReduce2