Skip to content

Latest commit

 

History

History
3 lines (2 loc) · 345 Bytes

preemptible worker node.md

File metadata and controls

3 lines (2 loc) · 345 Bytes
  1. first of all is preemptible nodes are much less expensive. I think, they're about 80% less than the cost of a typical version of that particular instance.
  2. The other reason is because Spark and Hadoop are both designed to be resilient to failures in nodes. So if Google takes back your preemptible node, that's not necessarily a problem.