chp 2 - section 3:

• reliability - it keeps running even if there is an issue, and the entire system does not grind to a halt
• a fault - a part of the system not working
• a failure - the how whole system is not working
• SPOF - single point of failure
• chaos engineering - purposely causing faults to test fault-tolerance
• tolerate faults instead for preventing them
• hardware faults:
  • increase hardware fault tolerance through redundancy - so that if one hard drive/ram/whatever fails there re other's already in place to do the work
  • if your system is a single node - it will have to come down for upgrades and patches
  • if your system is multi-node you can have rolling upgrades
• software faults: a fault in a software propagates to all nodes its deployed to
• sociotechnical: interplay between social and technical aspects in complex systems.
• humans:
  • make mistakes
  • blaming them is counter-productive
  • design the systems to configure things to encourage the right things, the same way you designing for customers
  • Preventable mistakes should be seen as a problem of the system that allowed not the human you might have done it while rushed and overwhelmed
• unreliable software has far reaching consequences beyond app being down and lose of productivity, humans have been wrongly sent to jail due to bugs