⚠️ This post is based on a real production setup. Some server and domain info are anonymized.

Overview

Deploying a WordPress site that serves 57,000 pageviews/week using AWS Free Tier sounds impossible — but it’s doable with the right stack and tuning.

This post walks through my setup using:

• t2.micro (1GB RAM)
• LiteSpeed Enterprise (Starter License – \$0)
• MariaDB + Redis
• Cloudflare (Free Plan)

Final Result:

• ~250 concurrent users (simulated)
• Avg TTFB: 189ms
• Redis hit rate: 99.91%
• RAM peak usage: 798MB
• Zero cost infra (excluding domain)

🔧 Stack

Key Optimizations

1. Multi-layer caching

• Redis for object cache
• LiteSpeed Cache plugin for page-level cache
• Cloudflare for CDN + static asset caching

2. RAM & CPU efficiency

• Swap enabled (2GB) to extend t2.micro stability
• Redis tuned to 256MB, allkeys-lru eviction
• Peak memory under 800MB under 250 concurrent users

3. Deployment tricks

• Uploaded large site zip via wget (faster than scp)
• Adapt vHost config from OLS
• Changed LiteSpeed WebAdmin port to avoid scans

Benchmarks

Loader.io: 250 concurrent users / 60s

• ✅ 100% success rate
• ⚡ Avg Response: 189ms
• 💾 Peak RAM: 798MB → stabilized ~430MB
• 🔥 CPU: ~100% short burst, no crash

Redis hit rate:

redis-cli info stats
→ keyspace_hits: 1,623,661
→ keyspace_misses: 1,480
→ Hit rate: 99.91%

Lighthouse

• Score: 95 / 100 (desktop)
• FCP: 0.3s
• LCP: 0.5s

Pingdom (San Francisco)

• Load time: ~1.78s
• CDN: Cloudflare

Lessons Learned

• 1GB RAM works if you optimize smart.
• Redis + full page cache reduces load by 70–80%
• Externalizing media (Imgur) cuts disk & CPU
• Benchmark everything (ab, loader.io, redis-cli, curl -I)
• Don’t rely on plugins — know your stack

🚫 Performance isn’t about bigger servers. It’s about smarter engineering.

Want the full guide?

👉 Full config & setup: GitHub Repo 👉 Live demo: https://demo.assettolab.com dead as result of AWS Account suspension 🎁

Like it?

Give it a ❤️ and let me know your setup below. Always happy to trade war stories.

(Originally deployed from scratch by Aleix, a DevOps-leaning generalist obsessed with lean infra & real-world performance.)