MEPX
Chapter 3 of 10All chapters
  1. 1Start with numbers
  2. 2Tiers and statelessness
  3. 3Scaling
  4. 4Caching
  5. 5Data at scale
  6. 6Queues and async work
  7. 7Consistency
  8. 8Designing for failure
  9. 9Knowing what is happening
  10. 10Making the trade-off

Chapter 3 of 10

Scaling

Up, out and the cost of each.

Two directions

Scaling up buys a bigger machine and needs no code changes, until you reach the largest machine available. Scaling out adds machines and needs coordination, but has no ceiling.

  • Scale up first: it is cheaper in engineering time than distributed complexity.
  • Scale out when you need redundancy as well as capacity.

Load balancing

A load balancer spreads requests and removes unhealthy instances. Health checks are the part people forget, and they are what makes failure invisible to users.