Solution
Getting ready for 11.11
Two months before the 11.11 (Singles’ Day) sale, we fully prepared the infrastructure: load testing, profiling and optimizing heavy SQL queries, moving sessions and carts to Redis, and fine-tuning Nginx and PHP-FPM for high concurrency. We also optimized WooCommerce specifically for peak loads — HPOS, object caching, and deferred task processing via queues.
We deployed full-scale monitoring on Grafana + Prometheus: CPU, RAM, disk I/O, the Nginx request queue, MySQL response time, and the number of active PHP processes. We set up alerts in Telegram, so any deviation from the norm reached us instantly.
The 11.11 sale: results
When the mailing went out, thousands of shoppers hit the site at once. Thanks to the preparation, the server withstood the load: carts worked, orders went through, and payments were processed. Total downtime over the entire sale was about 5 minutes — not critical for the business and nothing like what it used to be.
Black Friday
Building on the 11.11 experience, we further optimized the bottlenecks that surfaced under real load. We prepared runbooks for rapid response: what to do when the queue grows, when the database slows down, or when PHP processes spike. Engineers were on call around the clock.
Black Friday went even more smoothly: the server didn’t dip for a single second, and all systems ran normally. We monitored in real time and adjusted settings on the fly.
The outcome
The two biggest sales of the year — with no lost orders and no critical failures. The business ran at full capacity exactly when it mattered most. The infrastructure investment paid for itself in a single day of the sale.
