Scaling, what you need to know
summary
Advocates splitting application into process and storage with concrete choices: PostgreSQL for transactional rows, Redis for TTL sessions, S3 (or similar) for blobs, RabbitMQ (or Redis where appropriate) as event broker. Warns against using RAM or the local filesystem for shared state. Recommends reverse proxy in front, automated build/deploy pipelines, and small services — example: one system split into five services handling hundreds–thousands of users, allowing each service to be redeployed six times a day with zero visible downtime.
AI-written summary · about
chapters (4)
related talks
13mLow-Maintenance Backend Architectures for Scalable Applications
01h 02mTools for simplified large scale systems
32mScaling Done Right
18mScaling APIs to Billions: Extreme-Scale Reliability
50mBuilding an Application Platform: Lessons from
44mIntroduction to application scaling
46mDeploying and maintaining software for today's infrastructure
47mDeploying Like a Boss: From 2 Servers to 20,000