System Design Fundamentals I Wish I Had Learned Earlier

The article discusses fundamental system design concepts that are often overlooked in favor of interview-focused resources. The author emphasizes the CAP theorem, which states that a distributed system can guarantee at most two of three properties: Consistency, Availability, and Partition tolerance. Since network partitions are unavoidable, the practical trade-off is between consistency and availability.

Consistency ensures every read receives the most recent write, while availability ensures every request receives a response, even if it's slightly stale. Most web applications opt for availability over strict consistency, accepting eventual consistency, where data becomes consistent eventually despite potential slight staleness during propagation.

The article then covers horizontal vs. vertical scaling. Vertical scaling involves adding more resources to a single machine, which is simple and requires no architectural changes but hits hardware limits or becomes costly. Horizontal scaling adds more machines and distributes load, requiring stateless architecture, load balancers, session management, and careful state management.

The author advises starting with vertical scaling and only moving to horizontal scaling when necessary, emphasizing the importance of code and query optimization before scaling horizontally.

Caching is highlighted as the most significant performance tool. It stores the results of expensive computations to serve them quickly on subsequent requests, offering high leverage improvements. Caching can occur at the client-side (browser cache), CDN cache, application cache (Redis/Memcached), or database query cache. The author emphasizes the importance of cache invalidation and presents three strategies: TTL, write-through, and cache-aside.

Finally, the article underscores the importance of database indexing. Without an index, a database must scan every row to find matching records, which becomes increasingly costly as the number of rows grows. Indexes allow the database to find matching rows in O(log n) time, significantly improving performance. The author advises indexing primary keys, foreign keys, columns in WHERE clauses, and ORDER BY clauses, balancing the trade-off between index benefits for reads and potential slowness for writes.