Design a Logger Rate Limiter

To design this logger, I would first clarify that we need to prevent printing a specific message within a 10-second sliding window. For Meta's scale, efficiency is critical, so I propose using a HashMap rather than just a HashSet, as we need to track the exact timestamp of the last print for each message key. The logic flow would be: when a new request arrives with a message and timestamp, I check the HashMap. If the message is absent, I immediately print it and insert it into the map with the current timestamp. If the message exists, I retrieve its last timestamp and calculate the difference between the current time and that value. If the difference is greater than or equal to 10 seconds, I consider it valid, print the message, and update the stored timestamp to the current one. Otherwise, I silently drop the request. This approach ensures O(1) average time complexity for every operation, which is vital for high-volume logging systems. Regarding the Hash Set requirement, a pure set would force us to store separate timestamp objects or manage parallel arrays, adding unnecessary overhead. The HashMap elegantly solves both uniqueness and timing in one structure. Finally, I'd note that for production systems at Meta, we might add a background thread to periodically prune old entries from the map to prevent unbounded memory growth over long-running services.