09-21-2017 10:18 AM
Hi, is it supported to have some Worker/Data nodes on premise and others in the AWS or Azure cloud within the same cluster managed by the same NN/YARN? What are the issues with this?
09-25-2017 09:42 AM
Cloudera Director is designed for provisioning cloud clusters and cannot create a mixed cloud/on-premise cluster.
My initial thoughts on a mixed cluster is that network latency between the cloud nodes and on-premise nodes would lead to poor performance. You may be better served by creating two separate clusters and using some other mechanism for sharing data. For example, you can export/import data to S3.