• Hungry Minds
  • Posts
  • 🍔🧠 Netflix's 3-Tier Cache System That Powers 700M Hours Daily

🍔🧠 Netflix's 3-Tier Cache System That Powers 700M Hours Daily

PLUS: Caching Explained in 5 Minutes 🗝️, AI Agents Starter Kit 🧰, Time Management for Engineers ⏳

Author

Alex Zajac
April 14th, 2025 • Time-to-digest: 11 minutes

In partnership with

Happy Monday! ☀️

Welcome to the 534 new hungry minds who have joined us since last Monday!

If you aren't subscribed yet, join smart, curious, and hungry folks by subscribing here.

📚 Software Engineering Articles

🗞️ Tech and AI Trends

👨🏻‍💻 Coding Tip

  • Use Docker's --mount instead of -v for clearer volume management

Time-to-digest: 5 minutes

Big thanks to our partners for keeping this newsletter free.

If you have a second, clicking the ad below helps us a ton—and who knows, you might find something you love.

💚

The #1 AI Meeting Assistant

Still taking manual meeting notes in 2025? Let AI handle the tedious work so you can focus on the important stuff.

Fellow is the AI meeting assistant that:

✔️ Auto-joins your Zoom, Google Meet, and Teams calls to take notes for you.
✔️ Tracks action items and decisions so nothing falls through the cracks.
✔️ Answers questions about meetings and searches through your transcripts, like ChatGPT.

Try Fellow today and get unlimited AI meeting notes for 30 days.

Netflix powers 15% of global internet traffic and streams 94 billion hours of content globally. Their sophisticated distributed system architecture enables instant, uninterrupted video delivery to millions of concurrent viewers across 190 countries through cloud services and custom content delivery networks.

The challenge: Building a globally distributed system that maintains sub-second latency and high reliability while streaming HD video to millions of concurrent users, handling 15% of global internet traffic.

Implementation highlights:

  • Two-plane architecture: Control plane on AWS for browsing/recommendations, data plane (Open Connect) for video delivery

  • Custom CDN network: Purpose-built Open Connect servers placed directly within ISP networks

  • Microservices at scale: 700+ services enabling independent scaling and deployment

  • Chaos engineering: Tools like Chaos Monkey randomly break things to ensure resilience

  • Sophisticated data pipeline: Real-time analytics processing millions of events/second

Results and learnings:

  • Global reach: Successfully streams to 301M+ subscribers across 190 countries

  • Reliability: Maintains 99.99% availability despite massive scale

  • Optimization: Reduced bandwidth needs by 20-40% through per-title encoding

Netflix's architecture shows how thoughtful system design can solve incredibly complex technical challenges a massive scale. Their willingness to build custom solutions rather than accept off-the-shelf limitations demonstrates how architecture can be a competitive advantage.

Remember: Even Netflix's system isn't perfect. Just ask anyone who tried to watch the Tyson vs Paul fight! Sometimes the best architectures still buffer.

How PostgreSQL Works: Internal Architecture Explained

PostgreSQL has emerged as one of the most powerful and versatile open-source relational databases

How to develop a Senior mindset

5 specific real-world examples of the difference between Junior and Senior mindset!

How to Test Software 🔍

A practical and opinionated take on one of engineering's most divisive topics.

3 Steps to Weather Tough Tech Markets

It’s no secret that we’re currently facing a difficult tech market. From recurring mass tech layoffs and rapid advancement in artificial intelligence to a more recent stock market correction, many software engineers are feeling uneasy about the future. While this feeling is reasonable, tough markets are inevitable—because of this, it’s crucial to learn how to weather them tactfully. Here are the four critical moves that catapulted my career and wealth during the COVID crash, while others panicked.

Time Management

10 Ways to Take Ownership of Your Time

Building Robust React Apps with Zustand and Immer

For years, I dodged React like the plague. In fact, I avoided JavaScript altogether, even in web-related tasks. Take static site generators, for example. For my old chemistry blog, I experimented with…

Every Caching Strategy Explained in 5 Minutes

SWE Quiz - Become the smartest dev on your team.

The Best Programmers I Know

I have met a lot of developers in my life. Late…

ARTICLE (docs-or-it-didnt-happen)
Why Documentation Fails Developers

GITHUB REPO (zoom-zoom-multiplayer)
Multiplayer at the speed of light

Want to reach 180,000+ engineers?

Let’s work together! Whether it’s your product, service, or event, we’d love to help you connect with this awesome community.

Brief: Meta launches Llama 4 Scout (17B params, 10M-token context) and Llama 4 Maverick (beats GPT-4o/Gemini 2.0), both open-weight and multimodal, while teasing the training of Llama 4 Behemoth, a 288B-parameter powerhouse.

Brief: Midjourney's V7 Alpha introduces model personalization by default, Draft Mode for rapid iteration, and Turbo/Relax modes, offering higher-quality textures, better coherence, and 10x faster rendering for AI-generated art.

Brief: Google’s new Ironwood TPU shifts focus from training to AI inference, tackling soaring costs with double performance per watt and 6x more memory than its predecessor.

Brief: Samsung’s Ballie rolling companion will debut Gemini AI integration in May, promising smarter home automation and voice-controlled tasks in its latest revival.

Brief: OpenAI upgrades ChatGPT’s memory to reference past conversations by default, raising privacy concerns but enabling more personalized responses for Plus and Pro users.

This week’s coding challenge:

Build Your Own HTTP Server

Real-world proficiency projects designed for experienced engineers. Develop software craftsmanship by recreating popular devtools from scratch.

This week’s tip:

Use Docker's --mount flag instead of -v for explicit and more maintainable volume declarations. The --mount syntax is more verbose but self-documenting and supports additional options like bind-propagation and readonly.

Wen?

  • Microservice deployments: Clearly document volume requirements in deployment scripts with explicit mount options.

  • Shared development environments: Prevent accidental writes with readonly flags and ensure consistent mount behavior across teams.

  • Production systems: Better control over volume properties and easier debugging of mount-related issues.

If you don't read the newspaper, you're uninformed. If you read the newspaper, you're misinformed.
Mark Twain

That’s it for today! ☀️

Enjoyed this issue? Send it to your friends here to sign up, or share it on Twitter!

If you want to submit a section to the newsletter or tell us what you think about today’s issue, reply to this email or DM me on Twitter! 🐦

Thanks for spending part of your Monday morning with Hungry Minds.
See you in a week — Alex.

Icons by Icons8.

*I may earn a commission if you get a subscription through the links marked with “aff.” (at no extra cost to you).