Mitigating replication lag and reducing read load with freno

This post introduces 'freno', a new centralized throttling service designed to replace existing, disparate throttling mechanisms. freno addresses operational issues and limitations by continuously and asynchronously polling MySQL replicas for replication lag. It provides a unified API for applications to query lag status and receive recommendations, enabling better control and visibility over large write operations. The post details the architecture of freno, including its continuous polling mechanism and its role in managing replication lag for background jobs and schema migrations.