Realtime Delivery Event Retention Rules for Recovering Missed Map Updates

Realtime delivery event retention rules are crucial for ensuring correct map updates after connectivity issues like phone sleep, tunnel drops, or browser restarts. The key elements include a short, durable event window, a reconnect cursor, and a retention contract that combines a snapshot with replay capabilities.

In a Python-based system, the retention logic is implemented in a gateway that keeps an abstract store to test policies without choosing a specific broker or database. The gateway handles three main responses: replaying events when the cursor is within the retained window, returning a snapshot if the cursor is too old, and doing nothing if the client is up-to-date. This response structure helps maintain a consistent map experience across various scenarios.

To size the retention log, consider three consumers: reconnects, delayed readers, and operations. Estimate the event rate per delivery and the expected offline interval to determine the necessary log size. Implement a replay evaluator to test the system under various conditions, such as simulated delays, duplicate events, and out-of-order frames. Monitor key metrics like the oldest retained sequence, replay-hit rate, snapshot-fallback rate, bytes per delivery, and time from ingestion to rendering.

By following these guidelines and designing a robust retention policy, you can ensure that your realtime delivery service provides accurate and reliable map updates even in challenging network conditions.