Monday, 18 August 2025

How I Handle 1.5k Users/Minute at Zero Cost on Enjoy Movie

on  

When I first built enjoymovie.live, I wasn’t trying to create some hyperscale system like Netflix. I just wanted a simple streaming site. But then traffic grew — and suddenly I was serving over 1,500 users per minute.

The crazy part? I pay $0 for infrastructure. No servers, no bandwidth bills, no scaling headaches. That, to me, is the real achievement. Not just handling traffic — but handling it for free.

The Philosophy: Remove Work Until Nothing Is Left

Most developers panic when traffic grows. They spin up servers, configure databases, add Kubernetes clusters, and watch costs spiral. I took the opposite approach: I stripped away everything unnecessary until there was nothing left to scale. The CDN does all the heavy lifting.

This is the mindset: don’t handle traffic yourself, make the edge handle it.

Step 1: My Origin Barely Exists

The “server” behind enjoymovie.live is just static files hosted on Cloudflare Pages. My build outputs:

  • HTML files for every page
  • Static JSON with movie metadata
  • Hashed JS and CSS bundles
  • Poster and thumbnail images

No PHP, no Node, no databases. In practice, 98% of requests never hit my origin because Cloudflare caches everything at the edge.

Step 2: Load Balancing Without Servers

I don’t run load balancers. Instead, traffic is automatically spread across Cloudflare’s 300+ global data centers. Users in Paris get content from Paris. Users in Delhi get it from Delhi. Los Angeles users are served in LA. That’s real load balancing — without me managing anything.

Example: On a Saturday evening spike, Cloudflare analytics showed:

  • North America: 4,500 requests (served from 8 edge POPs)
  • Europe: 3,000 requests (served from 6 edge POPs)
  • Asia: 4,000 requests (served from 10 edge POPs)
  • Origin: fewer than 200 requests total

This is what my architecture looks like in practice:

User ---> Nearest Cloudflare Edge ---> Cache Hit
                                   |
                                   v
                              (if cache miss)
                                   |
                                   v
                              Cloudflare R2 (storage)
                                   |
                                   v
                              Edge caches result

I don’t pay for servers, yet my site feels global.

Step 3: Workers Keep Traffic Smart

Cloudflare Workers act as traffic managers. They normalize requests and keep the cache clean. Example:

  • /movie/123?utm_source=facebook
  • /movie/123?utm_source=twitter

Both are the same movie, so my Worker strips out utm_source. Result: one cache object instead of many. Multiply that across thousands of users, and my origin load drops to near zero.

Step 4: Video Streaming for $0

Streaming usually kills projects because bandwidth costs explode. But I solved this by leaning entirely on the CDN:

  1. HLS segments cached at the edge: A movie is just .m3u8 playlists + .ts chunks. Once the first viewer loads them, they’re cached at the nearest edge. The next 500 viewers get them for free.
  2. Free storage with R2: Cloudflare R2 gives me free egress to Cloudflare’s CDN, meaning I pay nothing to serve videos worldwide.
  3. Embeds for some titles: For certain content, I serve only the player shell and let external infra handle the bandwidth.

Real Example: One movie got 500+ simultaneous viewers. The first requests pulled ~200MB from R2. After that, every segment was served directly from cache. My bill? $0.

Step 5: No Databases, Just JSON

Things like search, trending, and catalogs feel dynamic but are actually static:

  • A JSON index (movies.json) is generated at build time.
  • Search happens in the browser (client-side).
  • Trending counters use KV but are cached at the edge for 5 minutes.

No SQL, no MongoDB, nothing to scale or maintain.

Step 6: Frontend Discipline Matters

The frontend helps scaling by using less bandwidth:

  • Images are lazy-loaded (loading="lazy").
  • I preload the first movie page for instant feel.
  • The app is a SPA, so JS/CSS are fetched once.
  • Responsive srcset ensures phones don’t fetch 4K posters.

Less waste per user = better scale.

Step 7: The Real Achievement — Scaling for Free

Handling 1.5k users/minute is impressive. But the real achievement is doing it at zero cost. Here’s what my stack looks like:

  • Hosting: Cloudflare Pages (free)
  • CDN: Cloudflare global edge (free, unmetered caching)
  • Workers: Millions of requests covered by free tier
  • Storage: Cloudflare R2 with free egress
  • Database needs: KV/D1 free tier
  • Analytics: Cloudflare Analytics (free)

Now compare this to AWS/GCP pricing:

  • AWS CloudFront + S3: ~$100+ for 1TB video streaming
  • Google Cloud CDN + Storage: ~$80–120 for the same
  • My stack: $0

That’s the victory: I scaled like Netflix without paying Netflix’s bills.

Final Thoughts

Scalability isn’t always about adding servers. Sometimes it’s about removing work until there’s nothing left to scale. That’s how I run enjoymovie.live at 1,500 users per minute, with no infrastructure bills at all.

Share: