System Design Cheat Sheet with 28 Chapters for Those Who Don't Want to Read Two Volumes by Alex Xu

I remember a couple of years ago when I first sat down to prepare for a system design interview. You open the popular two-volume book by Alex Xu "System Design Interview" and realize it has nearly a thousand pages. You usually have a week, at most two, before the interview, and your head is already spinning from the abundance of diagrams. And recently, I stumbled upon the liquidslr/system-design-notes repository in my bookmarks.

The project author did all the routine work for us. They took both parts of the book, removed the unnecessary fluff, and organized 28 topics with ready-made architectural solutions. The project has already gathered nearly 15,000 stars on GitHub, and there are clear reasons for that.

What's Inside the Notes

There are no abstract discussions about clean architecture here. The repository strictly follows the structure of the original books and is divided into two major groups of topics: basic building blocks and designing specific services.

The first part covers the basic mechanics without which it's impossible to design any reliable backend:

• Back-of-the-envelope estimation. The ability to calculate incoming RPS, disk space for terabytes of logs, and required network bandwidth in a couple of minutes.
• Rate limiting algorithms for protecting public APIs from overloads and spam.
• Consistent hashing, which saves the day when dynamically sharding databases and balancing cache.
• Generating unique IDs in a distributed environment without locks, like Twitter's Snowflake or Flickr's ticket servers.

The second group of topics is dedicated to analyzing full-fledged services. The author breaks down tasks that are most commonly seen in big tech interviews:

• URL shortener service architecture with hash collision calculations.
• Scalable chat at the level of Slack or Discord with persistent WebSocket connections.
• Video hosting like YouTube with an asynchronous transcoding pipeline for different resolutions.
• Payment gateways and digital wallets, where operation idempotency and two-phase commit are critically important.

Real Engineering Articles Instead of Dry Theory

The main advantage of the repository isn't even the condensed retelling of chapters. At the end of the README, the author collected links to original technical articles and papers from Discord, Uber, Netflix, and Amazon teams.

Book diagrams often look sterile. On paper, all components work perfectly, queues never fail, and database replicas synchronize instantly. Real production is different.

When you read the chapter on messaging systems, you can immediately open the Discord engineers' article about how they stored billions of messages and why they initially chose Cassandra, later rewriting the storage to Rust and ScyllaDB. There's also an article about Flannel architecture by the Slack team, explaining how the application-level cache works at the network edge.

The section on search and autocomplete includes a link to Prefixy and materials on Prefix Hash Tree. For those diving into distributed data storage, original papers on Google BigTable and Amazon Dynamo DB are collected.

This approach connects book theory with real experience from large companies.

How to Read and Use These Notes

Materials are available in two formats. You can study the folders directly in the GitHub repository, or open the web version via Pagefy, with a link at the very beginning of the README. The second option is more convenient for reading on a smartphone or tablet on the go.

In each folder, you'll find:

• Problem statement with functional and non-functional requirements.
• High-level component diagram (load balancers, caches, databases, queues).
• Detailed breakdown of architectural bottlenecks and ways to scale them.
• Final conclusions and trade-offs of the chosen solution.

No code is written here; the focus is specifically on data flow logic, communication protocols, and choosing storage types for different workload patterns.

That said, the repository has a nuance: the project is still marked as work in progress. Some chapters are formatted in detail, with formulas and neat diagrams, while some chapters currently only have brief outlines. The author updates the content from time to time, but before a serious interview, it's best to additionally check the mentioned articles from engineering blogs.

Who the Project Will Save Time For

If you're preparing for a job change to a Middle+ or Senior level position, this repository will save you dozens of hours.

The notes will also be useful for those who:

• Are designing a new service at work and choosing between SQL, NoSQL, and key-value storage.
• Want to understand the details of how geo services work (like finding nearby objects or friends on a map) through GeoHash and QuadTree.
• Are writing integrations with payment systems and want to avoid double charges during network failures.
• Are tired of typical tasks and want to improve their understanding of distributed systems in general.

If time is tight, I recommend starting with the first three topics: scaling from zero to millions of users, quick back-of-the-envelope calculations, and a step-by-step framework for answering system design interview questions. This will create a solid foundation onto which you can then layer any specialized scenarios.