Druid cluster configuration assistance required

Hi All,

I’m new to Druid. We have requirement to form 9 node cluster and configure Druid.

Need assistance in how to approach in configuring on a cluster. Kindly advise on how many servers can be allotted to Master, query & Data nodes. What are all the process that can be split across various nodes

Well, the size of your cluster will depend on a number of factors, including the amount of data, your rollup, how many concurrent queries and what type of ingestion tasks. Druid is unlike a traditional database, and allows for a high level of flexibility.

In general, I recommend fairly beefy nodes with fast attached storage for Data. To use AWS speak, i3.2XL or larger. You scale these out as you need more storage or CPU power.

For query nodes, smaller machines that have a good CPU to RAM ratio are better choices, and you will want 1 for every 10 - 15 data nodes you have.

For master nodes, I recommend you have 3 of them to create an HA cluster, and you can scale those up as needed. You never need more than 3, but you can get away with just 1 if you aren’t worried about HA.

Cheers,

Rachel