Skip to main content
Version: v2.10.0

Cluster Size

Cluster size defines the compute power allocated to a Yeedu cluster: the compute shape, the instance range, and the parallel execution slots available on each instance. Those three settings interact, and misjudging them is expensive in either direction, because an undersized cluster queues work that could have started immediately while an oversized one bills for instances that never receive anything to run.

What you're configuring here is the envelope rather than the exact allocation.

Parameters Involved​

  • Compute : The machine shape (cores, memory, disks, GPUs) chosen from the Select Compute modal
  • Min Instances : Minimum compute required to keep the cluster active
  • Max Instances : Maximum capacity the cluster can scale to
  • Number of Parallel Execution : Max concurrent executions per instance

Note: Workers per Node isn't a field in Yeedu v2.10.1. Per-instance concurrency is controlled by Number of Parallel Execution on the Create Cluster form.

We optimise the sizing automatically in response to the jobs you submit, within the envelope those parameters describe.

Cluster Size