A production Databricks cluster is configured with one driver and three executor nodes, all utilizing the same virtual machine type. While monitoring the Ganglia Metrics dashboard, which of the following observations would most likely indicate a performance bottleneck caused by code executing sequentially on the driver rather than being parallelized?