Answer: They can use clusters that are from a cluster pool

## Explanation **Correct Answer: D - They can use clusters that are from a cluster pool** **Why this is correct:** 1. **Cluster Pools**: When you use clusters from a cluster pool, Databricks maintains a pool of pre-warmed, idle instances that are ready to be assigned to clusters. This significantly reduces cluster start-up time because: - The instances are already provisioned and warmed up - No need to wait for cloud provider instance provisioning - The instances are already configured and ready to use 2. **Job Clusters vs. All-Purpose Clusters (Option B)**: While job clusters are optimized for jobs, they still need to be provisioned from scratch each time unless they come from a pool. Using job clusters alone doesn't solve the slow start-up problem. 3. **Single-Node Clusters (Option C)**: Configuring clusters to be single-node might reduce some complexity, but it doesn't address the fundamental issue of instance provisioning time. 4. **Autoscaling (Option E)**: Autoscaling helps with handling varying workloads but doesn't improve initial cluster start-up time. 5. **Databricks SQL Endpoints (Option A)**: These are for SQL analytics workloads, not for general Spark jobs, and don't address cluster start-up time for job tasks. **Best Practice**: For production jobs that run regularly (like nightly jobs), using cluster pools is a recommended best practice to minimize cluster start-up latency and ensure consistent performance.

Author: Keng Suppaseth