A Snowflake data warehouse has a large table named 'ORDERS' that is frequently queried by multiple virtual warehouses of different sizes for various analytical workloads. The data in the 'ORDERS' tabl

Question

Options

• ACreate multiple clones of the 'ORDERS' table, one for each virtual warehouse, to isolate workloads
• BUse Snowflake's caching mechanisms and query acceleration service to improve query
• CImplement data partitioning on the 'ORDERS' table based on a frequently queried column to
• DUtilize Snowflake's multi-cluster warehouses with auto-scaling enabled to handle concurrent
• ECreate multiple materialized views on the 'ORDERS' table tailored to the specific query patterns of

• A
  15% (5)
• B
  48% (16)
• C
  36% (12)

Explanation

B, D, and E represent Snowflake-native features purpose-built for concurrent, high-throughput workloads. Snowflake's result cache and local disk cache (B) serve repeated query results without recomputing against the base table, while the Query Acceleration Service offloads large scan portions to shared compute-both reducing warehouse load without duplicating data. Multi-cluster warehouses with auto-scaling (D) directly tackle contention by automatically spinning up additional clusters when query queues build, ensuring different workload sizes don't starve each other. Materialized views (E) pre-aggregate common query patterns and-critically-Snowflake automatically keeps them consistent with the base table during DML and batch loads, satisfying the data consistency requirement.

Why A is wrong: Table clones are point-in-time snapshots, not live-synced replicas-with constant batch loads, clones immediately become stale and create consistency problems rather than solving them. Why C is wrong: Clustering keys improve micro-partition pruning for individual queries but do nothing to reduce inter-warehouse contention or maintain consistency during concurrent writes-it addresses only one dimension of the problem.

Memory tip: Remember "CDM = Concurrent Data Management" - Caching, Distributed clusters (multi-cluster), Materialized views. These three are Snowflake's layered answer to the "many warehouses, live data" problem, each attacking a different bottleneck (redundant compute, queue contention, and repetitive aggregation respectively).

No community discussion yet for this question.