This document discusses chaos engineering and using chaos experiments to test systems reliability. It introduces chaos engineering as a way to acquire knowledge about how systems behave under stressful conditions in order to make better decisions. Chaos experiments follow a protocol of defining a baseline, impacting the system in a controlled way ("what if..." experiments), and observing if the impact deviates from the baseline. The Chaos Toolkit is presented as an open source tool that automates running chaos experiments across different platforms and services. It has drivers for common platforms and tools to inject faults and monitor system behavior. Chaos engineering helps build more reliable, secure and performant systems.