Configuring Compute Performance: 10 practice questions

Question 1 of 10

Which setting changes worker count between a configured minimum and maximum as demand varies?

1. AAutoscaling
2. BRuntime preloading
3. CAutomatic termination
4. DDedicated access mode

Question 2 of 10

Which node family best fits large joins and aggregations that otherwise spill to disk?

1. AGPU-accelerated nodes
2. BMemory-optimized nodes
3. CCompute-optimized nodes
4. DStorage-optimized nodes

Question 3 of 10

What does automatic termination do after the configured inactivity period?

1. AReduces the cluster to its minimum worker count
2. BStops the cluster while preserving its configuration
3. CReturns every worker to an instance pool but keeps the driver running
4. DDeletes the cluster configuration and all metadata

Question 4 of 10

Which compute component directly sets the maximum task parallelism available to Spark?

1. AThe total executor core count
2. BThe pool's idle-instance timeout
3. CThe workspace access mode
4. DThe automatic termination period

Question 5 of 10

Why can fewer large workers outperform more small workers in a shuffle-heavy analytical workload with equal total cores and memory?

1. AThey reduce network traffic during shuffle operations
2. BThey eliminate every write to local storage
3. CThey disable parallel task execution
4. DThey make the cluster a single-node resource

Question 6 of 10

Which architecture can add workers to scale horizontally?

1. AA single-node cluster
2. BA stopped cluster configuration
3. CAn idle instance without a cluster
4. DA multi-node cluster

Question 7 of 10

What is the defining startup benefit of an instance pool?

1. AClusters can take already provisioned virtual machines
2. BIdle virtual machines have no cloud-provider cost
3. CEvery workload becomes serverless after pool attachment
4. DClusters receive unlimited workers without a capacity setting

Question 8 of 10

A nightly transformation performs many large joins and shows disk spill. The total cluster budget is fixed, and the team wants to reduce both spill and shuffle traffic without reducing total memory. Which change best fits?

1. AKeep the nodes and shorten automatic termination
2. BUse more, smaller compute-optimized workers
3. CMove to a single-node storage-optimized cluster
4. DUse fewer, larger memory-optimized worker nodes

Question 9 of 10

Classic clusters are created many times each workday, startup delay is unacceptable, and demand varies during each run. The team accepts a controlled idle-infrastructure cost. Which configuration addresses both startup and active demand?

1. AA single-node cluster with runtime preloading
2. BA fixed cluster with a short inactivity timeout
3. CAutomatic termination without a pool or worker range
4. DAn instance pool combined with autoscaling

Question 10 of 10

A stable production workload uses the same resources throughout each run. Operations wants predictable performance and simple capacity planning, while compute must stop after inactivity. Which configuration best matches?

1. AAutoscaling with no inactivity termination
2. BMore small workers with no termination setting
3. CA pool with maximum idle capacity and no cluster
4. DA fixed worker count with automatic termination