How can I count the number of requests in the last second, minute and hour?

When designing high-concurrency systems, understanding how to calculate request counts in the last second, minute, and hour is crucial, as it directly impacts system performance monitoring and scaling strategies. Below, I will outline several common methods to achieve this.

1. Sliding Window Algorithm

The Sliding Window Algorithm is a widely used approach that dynamically calculates the total number of requests within a time window by leveraging timestamps. Specifically, it employs a double-ended queue (deque) to store each request's timestamp.

Example (for request counts in the last second):

1. When a new request arrives, add the current timestamp to the end of the queue.
2. Simultaneously, remove timestamps older than one second from the front of the queue.
3. The size of the queue directly represents the number of requests in the last second.

This method can be easily extended to calculate request counts for the last minute or hour by adjusting the window size.

2. Counter Method

Another effective approach involves using multiple counters to track request counts per second, minute, and hour. This method excels with high data volumes but requires proper synchronization mechanisms to handle concurrent requests.

Example:

• Maintain three counters: secondCounter, minuteCounter, hourCounter.
• For each received request, increment all three counters.
• Every second, reset secondCounter.
• Every minute, reset minuteCounter.
• Every hour, reset hourCounter.

3. Time Bucketing

Time Bucketing is a detailed technique for recording data within specific time intervals. It involves creating buckets for each second, minute, and hour, where each bucket stores the request count for that period.

Example:

• Create an array where each element corresponds to the request count for one second.
• For each received request, increment the count in the relevant second bucket.
• Every second, minute, and hour, aggregate the associated buckets to compute the total request count.

4. Redis and Memory Data Structures

In practical implementations, memory data structures like Redis can efficiently handle this functionality by utilizing its expiration policies and atomic operations.

Example:

• Use Redis's INCR command to increment specific keys.
• Set key expiration times to 1 second, 1 minute, or 1 hour.
• Retrieve the values using the GET command, which provide the request counts for the last second, minute, and hour.

Summary

When selecting an implementation, consider the system's specific requirements, expected load, and available resources. For instance, if request volumes are extremely high, solutions like Redis may be preferable to reduce application server load. If high real-time accuracy is critical, the Sliding Window Algorithm is often the better choice. Each method has distinct advantages and use cases, and the key is to choose appropriately based on the actual context.