What are the three core principles of DevOps/SRE?

• A strong focus on automation.
• The idea that "failure is normal".
• Redefining "availability" as what the business can tolerate, not 100%.

These three build on each other:

• Automation comes first because repeatable, fast and reviewable changes are what make frequent releases safe.
• "Failure is normal" accepts that in a large distributed system something is always broken somewhere. Rather than trying to prevent every failure, you design for recovery: redundancy, rollback, self-healing.
• Availability as tolerance. Once failure is accepted as normal, 100% uptime is no longer the target. The real question becomes "how much unavailability can our users and business live with?", and that number is written down as an SLO (see the next card).

Go deeper:

• Google SRE book — Embracing Risk — why 100% is the wrong reliability target.