Why on production recommends to co-locate Overlord and Coordinator?

On the Druid Production Cluster configuration document( http://druid.io/docs/latest/configuration/production-cluster.html) it says “Overlord and Coordinator nodes should be co-located on the same hardware.”
Could someone explain why so?



I believe the reasoning is that from a RAM and CPU perspective, they can generally be co-located on the same hardware.

I definitely would implement your production solution with multiple Overlord/Coordinator nodes in your cluster for service redundancy.

Both Coordinator and Overlord services elect one leader, so only one service (for each) on a healthy node of your cluster will doing the work.

Hope this helps.


Both services are very lightweight and don’t have a lot of resources. Co-locating is away to manage the cost of the infrastructure. This is pretty common in the big data space.

Rommel Garcia
Director, Field Engineering

Hi Christine,

I believe the doc page you linked has gone stale. I think it’s not mandatory to colocate overlord and coordinator in the same machine. But, you can do if your machine has enough resource.

Check http://druid.io/docs/latest/operations/recommendations.html and http://druid.io/docs/latest/configuration/index.html for latest recommended configurations.


Thanks all for the explanation. It’s what we thought, good to be confirmed.