Can we set up druid in the same cluster as our hadoop

Hi,

We have a fairly big hadoop cluster (90 nodes) running cdh5. Is it possible to set up Druid in the same cluster to run in parallel? If that’s possible will that reduce query performance?

Thanks

-Gokila

Hey Gokila,

Yes, you can colocate Hadoop and Druid. If you want to guarantee consistent quality of service from Druid, you could use cgroups or virtualization to ensure that it has a guaranteed amount of resources. Otherwise you run the risk of Druid performance suffering when Hadoop is doing an expensive job.