In the modern world, tech companies build their products on extremely reliable servers that never break. They’re stacked in the racks with highly reliable switches with firmware that is rock solid and guaranteed to have no bugs. These switches talk to each other over super low latency networks that have close to zero packet loss rates. And this whole thing is located in the building with infinite and redundant power supply. Just kidding, it’ll all break.

Companies can buy the most expensive top notch hardware, platinum support, pick the best vendors in the industry, but sooner or later everyone realize that everything fails. We’re going to talk about the inevitability of a failure and the ways how engineers can design their systems to be able to tolerate them.