Design KV store with sliding-window QPS metrics

Problem Statement

Design an in-memory key-value store that supports put(key, value) and get(key) operations. In addition to standard storage functionality, the system must expose a get_qps(window_size_seconds) method that returns the number of requests processed within the last N seconds.

Requirements

• The store must be thread-safe.
• get_qps should provide a reasonably accurate sliding-window count.
• Optimize for high-throughput writes (i.e., put and get operations should not be blocked by complex metric aggregation logic).
• Assume a scale of up to 100,000 operations per second.

Implementation Signature

class MetricsKVStore:
    def __init__(self):
        # Initialize data structures

    def put(self, key, value):
        # Store key-value pair and record metric

    def get(self, key):
        # Return value and record metric

    def get_qps(self, window_size_seconds):
        # Return total requests in last window_size_seconds

Example

• put('a', 1)
• put('b', 2)
• get_qps(60) returns 2 (assuming these were the only calls in the last minute).

Considerations

• How will you handle the memory consumption of time-stamped request logs?
• How do you handle concurrency when updating metrics vs. reading them?