You are designing a data pipeline in Snowflake to process streaming data from Apache Kafka. You need to choose the appropriate virtual warehouse size. The data arrives at a rate of 100,000 messages…

Question

Options

• AOnly the total volume of data (100MB/second) needs to be considered. Choose the smallest
• BThe complexity of the transformations and aggregations performed on the data significantly
• CThe number of concurrent users accessing the processed data is the primary factor. Choose a
• DAlways choose the largest available warehouse (X-Large or larger) to ensure optimal performance,
• EThe frequency of data arrival (100,000 messages/second) requires a warehouse with sufficient

• A
  32% (10)
• B
  48% (15)
• C
  6% (2)
• D
  13% (4)

Explanation

Selecting the right warehouse size requires weighing both the complexity of workload and the ingestion throughput simultaneously - options B and E capture these two distinct, non-redundant dimensions. Complex transformations and aggregations (B) are CPU/memory-intensive operations that demand more compute regardless of data volume, while a high-frequency arrival rate of 100,000 messages/second (E) demands sustained throughput capacity to prevent backpressure or lag - neither factor alone is sufficient to make the sizing decision.

Why the distractors fail:

• A is wrong because raw volume (100 MB/s) without considering transformation complexity leads to undersizing - a small warehouse handling simple SELECTs vs. complex window functions are fundamentally different workloads.
• C is wrong because concurrent reader count is a separate concern (handled via multi-cluster warehouses or query concurrency settings), not the primary driver for a pipeline processing warehouse.
• D is wrong because always choosing the largest warehouse ignores cost efficiency; Snowflake bills by credit consumption, so over-provisioning without justification wastes money without guaranteed benefit.

Memory tip: Think "What + How Fast" - What you do to the data (B = complexity) and How fast it arrives (E = frequency) are the two pillars of pipeline warehouse sizing. Volume alone (A) and readers (C) are secondary concerns, and "always biggest" (D) is never a sound engineering principle.

No community discussion yet for this question.