Menu

When do you run Spark jobs in Cluster Mode on Dataproc?

Reading 0:00
15s threshold

I get the basic argument that running concurrent jobs on dataproc can cause Driver OOM and poor performance. But how do you decide between increasing the master node memory/compute or adding 1 more master node vs choosing cluster mode?…

Anonymous readers can preview up to 1024 characters here. Log in to unlock the full article once ingest succeeds.