Snowflake distinguishes between scaling up a warehouse to improve the resources available to individual statements and scaling out through multiple clusters to accommodate concurrent workloads.
Option C is the intended answer because a multi-cluster warehouse can increase parallel query-processing capacity without requiring a continuously running 2X-Large warehouse. Smaller clusters reduce the cost of each running cluster, while auto-scaling provides additional clusters when concurrency increases.
Option A introduces another expensive warehouse without addressing inefficient workload execution. Option B increases compute capacity but does not necessarily improve the throughput of numerous small INSERT statements. Option D introduces additional tables without establishing that table-level contention is the primary bottleneck.
However, Snowflake's official cost-optimization guidance explicitly identifies frequent single-row INSERT operations as inefficient. These operations can generate substantial cloud-services overhead because Snowflake is designed primarily for analytical workloads rather than traditional OLTP processing.
Technical qualification: Option C is the strongest available answer, but it does not address the principal ingestion-design problem. Snowflake recommends batching INSERT operations or using bulk-loading mechanisms such as COPY INTO whenever appropriate. Smaller multi-cluster warehouses are not guaranteed to outperform a larger single warehouse for every DML workload.
For a production-quality examination item, the scenario should distinguish concurrency bottlenecks from inefficient ingestion architecture.
Official Snowflake references: Optimizing Cloud Services for Cost; Planning a Data Load; Multi-cluster Warehouses.
Submit