High-Traffic WooCommerce Marketplace Optimization

Overview #

A multi-vendor WooCommerce marketplace with tens of thousands of products was experiencing checkout timeouts and slow catalog browsing during traffic spikes — particularly around seasonal sales, when concurrent sessions could jump 5-10x overnight. The goal was to make the platform reliably handle peak traffic without a full replatform.

The Problem #

Diagnosing the platform surfaced three compounding issues:

• No persistent object cache. Every WP_Query call, including repeated ones on the same request, hit MySQL directly.
• Synchronous order processing. Payment confirmation, inventory sync, and vendor notification emails all ran inline during checkout, so slow third-party API calls directly extended checkout latency.
• Unbounded session growth. The wp_woocommerce_sessions table had never been pruned and had grown to millions of rows, slowing every session lookup.

The Solution #

The fix was staged rather than a rewrite:

1. Redis object cache in front of MySQL, cutting repeated query load on catalog and cart pages significantly.
2. Action Scheduler-based queueing for anything that didn't need to block the checkout response — inventory sync, vendor notifications, and order confirmation emails moved to background jobs.
3. Nginx fastcgi_cache for catalog and product pages, with cache bypass rules for cart/checkout/account routes and automatic purge hooks on stock or price changes.
4. Session table cleanup via a scheduled job pruning expired sessions, plus an index review on the sessions and postmeta tables.
5. Dockerized environment parity across local, staging, and production so the fixes validated in staging matched production behavior exactly.

map $request_uri $skip_cache {
    default 0;
    ~*/(cart|checkout|my-account) 1;
}
 
fastcgi_cache_bypass $skip_cache;
fastcgi_no_cache $skip_cache;
fastcgi_cache_valid 200 10m;

Technology Stack #

WooCommerce and WordPress remained the core platform — the work was infrastructure and architecture around it: Redis for object caching, Nginx for page caching and reverse proxying, MySQL with targeted index and schema cleanup, and Docker for environment parity across local, staging, and production.

Lessons Learned #

The instinct when a WooCommerce store struggles under load is often "replatform to something custom." In this case, the platform itself wasn't the bottleneck — the caching layer and synchronous request handling were. Fixing those in place was faster, lower risk, and preserved the vendor and catalog management workflows the business already depended on.