Drives fail, databases crash, fibers get cut and unindexed queries hit production. Do you know how your application reacts to those events? Are they covered by tests? What about the failures you haven't even thought of? To avoid cascading failures applications must adopt general patterns to defend against misbehaving dependencies, including themselves. This talk covers the resiliency techniques Shopify has successfully put into production at scale, and how we write tests to ensure we don't reintroduce single points of failure. You'll walk away from this talk equipped with the tools to make your applications resilient and to better sleep at night.