From Availability and Reliability to Chaos Engineering - Why Breaking Things on Purpose is a Good Idea

This talk examines availability, reliability, and large-scale software architectures, then introduces chaos engineering as a way to find failures before they become outages. It discusses resilience patterns and designing experiments to measure their effects.

Syllabus

Intro
Computing evolution
The micro-services architecture
Building Confidence Through Testing
Jesse Robbins GameDay: Creating Resiliency Through Destruction
Netflix 2013
Failure Injection
System Availability
Availability in Parallel
Auto-Scaling
Infrastructure as Code
Immutable Infrastructure
Read/Write Sharding
Database Federation
Message passing for async. patterns
Exponential Backoff
Circuit Breaker
Dynamic Routing with Route53
Stateless Services
CAP Theorem
Eventual Consistency
Non-blocking UI
Exception Handling
Service Degradation & Fallbacks
What is Steady State?
Business Metrics at work
Designing Experiment
Canary deployment
Quantifying the result of the experiment
Big Challenges to Chaos Engineering