The core problem is the long start-up time for a SQL endpoint that runs an intermittent query (every four hours). Serverless SQL endpoints (now called warehouses) are specifically designed to address this issue. They are managed by Databricks and provide "instant compute," eliminating the start-up latency associated with Classic or Pro endpoints, which must provision clusters in the customer's cloud account. This directly solves the start-up delay. The pay-per-use model of Serverless compute is also cost-effective for intermittent workloads, aligning with the goal of managing costs.

A. Reduce the SQL endpoint cluster size: This would likely slow down query execution and would not reduce the time it takes for the cluster to provision and start.

B. Increase the SQL endpoint cluster size: This might speed up query execution but would increase costs and does not address the initial start-up latency of the endpoint.

C. Turn off the Auto stop feature: This would eliminate start-up time for subsequent runs but would incur significant costs by keeping the cluster idle for nearly four hours between runs.

D. Increase the minimum scaling value: This setting is for handling query concurrency by adding more clusters, not for reducing the start-up time of the initial cluster.

1. Databricks Documentation

"What are SQL warehouses?": This document explicitly contrasts the different types of SQL warehouses. It states

"Serverless SQL warehouses... provide instant compute... Pro and Classic SQL warehouses take several minutes to start." This directly supports the choice of Serverless for eliminating start-up time.

Source: Databricks Documentation > Data Warehousing > What are SQL warehouses? > SQL warehouse types.

2. Databricks Documentation

"Configure SQL warehouses": This page details the configuration options. Regarding "Auto stop

" it recommends keeping it enabled (e.g.

at 10 minutes) to "avoid paying for idle time

" which explains why option C is incorrect from a cost-management perspective. The "Scaling" section clarifies that this feature manages the number of clusters to handle query load

not initial start-up

making option D incorrect.

Source: Databricks Documentation > Data Warehousing > Configure SQL warehouses > General & Advanced options.

3. Databricks Documentation

"Serverless compute for SQL warehouses": This page reinforces the benefit of Serverless endpoints. It highlights

"Benefits of serverless SQL warehouses include: ... Fast start-up time for SQL warehouses." This confirms that Serverless is the intended solution for the problem described.

Source: Databricks Documentation > Data Warehousing > Serverless compute for SQL warehouses.