Knock, a company specializing in providing simple APIs and dashboards for developers to integrate transactional notifications into their products, recently introduced a Throttle Function to their notification workflow engine. This feature aims to regulate the frequency of workflow triggers per recipient by implementing a throttle step that halts workflow runs once a specified limit is exceeded, allowing for a cooldown period. The design involves using Redis Streams, a data structure that records allowed workflow runs in a temporally ordered, append-only format. This approach enables efficient comparison of workflow run counts against throttle limits and allows for the expiration of outdated records to optimize memory usage. Redis was chosen for its lightweight, fast performance, and ability to handle concurrency with simple locks, despite potential data loss risks due to its lack of durability and constraints typically found in relational databases. Knock's development philosophy focuses on solving customer problems with straightforward solutions, leveraging existing tools, and maintaining curiosity about the technology landscape to identify optimal solutions.