Skip to content
Performance

Apache to OpenLiteSpeed: How We Cut Load Times by 94%.

· Jan 2026 · 2 min read

Why we moved away from Apache

Apache served us well for years. But as our client portfolio grew — and traffic volumes with it — we kept hitting the same bottleneck: Apache’s process-based architecture doesn’t scale efficiently under high concurrency.

Every connection spawns a thread (or process, depending on MPM). At 500+ concurrent connections, memory usage balloons. Add PHP-FPM into the mix and you’re running two process managers that don’t coordinate well.

Why OpenLiteSpeed over Nginx

The obvious alternative is Nginx. We evaluated both and chose OpenLiteSpeed for three reasons:

  • Native LSAPI — LiteSpeed’s PHP interface is 50% faster than PHP-FPM in our benchmarks because it eliminates the socket overhead
  • .htaccess compatibility — our WordPress sites use .htaccess extensively; Nginx requires rewriting all rules into its config format
  • Built-in cache — LSCache integrates at the server level with cache tags for granular invalidation, eliminating the need for Varnish

The migration process

We migrated 6 production WordPress sites over 3 weekends. Each migration followed the same playbook:

  • Provision a parallel server with OpenLiteSpeed configured identically
  • Sync the site files and database using rsync and mysqldump
  • Run our test suite against the new server (visual regression, functional tests, load tests. This is part of our standard cloud infrastructure playbook)
  • DNS cutover during lowest-traffic window with 5-minute TTL
  • Monitor for 48 hours, keep old server as instant rollback. (See also: our CI/CD pipeline template for zero-downtime deploys.)

The results

Across all 6 sites, the results were consistent:

  • TTFB dropped from ~800ms to ~50ms on cached pages (94% reduction)
  • Memory usage dropped 40% — same traffic, smaller server
  • PHP execution time reduced 30% thanks to LSAPI
  • Zero downtime during migration — DNS propagation handled the transition transparently

The combined infrastructure savings let us downsize 3 of the 6 servers, saving roughly $400/month. The performance gains improved Core Web Vitals scores across every site, with noticeable SEO improvements within 4 weeks.

Need help with your project?

We'll review your architecture and recommend the right path forward.

Book a Strategy Call →