Scale

Complexity

Less is more.

Reduce complexity at all costs.

Over-engineering is one common sources of complexity.

  • over-complicated solutions
  • over-complicated designs (too many classes and objects)
  • trying to predict the future

Keep scalability in mind when designing, developing, and deploying.

Scale Cube

The Scale Cube | AKF Partners

https://upload.wikimedia.org/wikipedia/commons/8/82/AKF_Scale_Cube.gif

  • x-axis: duplication, such as application code and database
    • easiest to achieve, just make copies
  • y-axis: split by verbs and nouns
    • focus on the “final” verbs that operate on the data directly
  • z-axis: split by properties of things, like hash-value of data
    • data of users from region A are stored in a database near region A

Vertical or Horizontal Scaling

  • Vertical: scale up
    • add more resources to an system
  • Horizontal: scale out
    • add more machines

Prefer scale-out

Use more and cheap devices.

  • Cheaper to maintain
  • Easier to scale
  • Less time to fix

Use more multiple data centers.

Use cloud smartly. Easy to set up and shut down as soon as we are done.

  • QA
  • batch processing
  • pike load