Tagged: Distributed System

Metrics are vital for the distribution system, this article describes how to implement a metric function for a high TPS system.

The code can be found here: https://github.com/ADU-21/producer-consumer

Our server has 75M daily TPS which generates a lot of ops work, I’m often asked “What do you guys do when you’re on call?” this blog will demon straight how we do troubleshooting in a distributed system.

Throttling is one of the three effective methods for protecting a high concurrency system. The other two are respectively caching and downgrading. Throttling is used in many scenarios to limit the concurrency and the number of requests. Our service has tens of millions of TPS, with tens of thousands of hosts serving traffic. Throttling is vital for such a large distributed service.