Automation and Scalability

Question 1

Designing for scalability

Answer 1

Resource usage should increase linearly.
Design for 10x growth in data, users, traffic.

needs to be available (resilient to failure, recoverability from failure, graceful degradation)

Question 2

How to achieve scalability and availability

Answer 2

• Automate as much as possible to save time and prevent human error
• Asynchrony as much as possible/background jobs. Avoid client-side transactions as much as possible. Helps with latency and helps avoid peak load time (cost). Do offline batch processes where appropriate (generate recommendations, import data, etc.)
• Assume everything fails
• Partition everything. Split every problem into manageable chunks.
• Ability to learn/improve over time with minimal human effort
• Rollback: all changes must be undoable!
• Real-time failure detection