You are tasked with optimizing a continuous data pipeline that loads data from an external stage into a Snowflake table using streams.
The pipeline is experiencing significant latency during peak hours. The stream is defined on a very large table with frequent updates and deletes. Which of the following strategies would be MOST effective in reducing the latency of the data pipeline, considering stream performance and cost implications?
Correct Answer: B
Pruning on the base table directly reduces the amount of data the stream has to track, therefore improving performance. Increasing warehouse size (A) might help but is not directly related to the stream's efficiency. Creating multiple streams (C) will likely increase the overhead. Reducing RETENTION _ TIME (D) might lead to data loss if changes are not consumed promptly. A materialized view (E) consumes credits and does not directly affect stream performance. The correct answer focuses on minimizing the workload for the stream itself.