Case Study: Moz achieves faster failover and slow-query visibility with Percona (ProxySQL & PMM)

Moz Uses Percona to Optimize the Architecture and Find Slow Queries in Their Database Environment

Moz is an SEO software company serving over 20,000 customers with a highly sharded database environment—more than 100k schemas across about 64 primaries, 64 replicas and ~250 servers, handling a few hundred to a few thousand QPS. They faced two critical challenges: slow master/slave failovers (DNS-based failovers took 5–15 minutes and often forced app restarts) and a lack of effective slow-query visualization and analysis across their architecture.

Percona performed an architecture audit and recommended and guided implementation of ProxySQL for fast, reliable failover/load balancing, Percona Server for MySQL for a stable platform, and Percona Monitoring and Management (PMM) to find and analyze slow queries. The changes reduced failover times to the required level, gave Moz continuous slow-query visibility for root-cause diagnosis, and improved overall availability and performance so customers can reliably access up-to-date data.

Moz

Phil Hildebrand

Senior Database Admin