This talk is about modern complex architectures and how Chaos experiments help in keeping error budgets in control by proactively discovering and preventing failures. When we look at the modern full stack, the number of layers is only increasing. There is usage PaaS, IaaS and SaaS, all in the mix. There are more and more technologies/tools/services being offered to either improve the way we are working today or solve an unsolved problem earlier in an innovative way. At one end, problems are being solved and at other end, more and more dependencies are being added. As per Murphy’s law, “Anything that can go wrong will go wrong” and when we look at it technically, it becomes “Anything that can fail will fail”. When we apply this law to modern multi-layered architectures with multiple dependencies, there is a need to do many things to prevent as many potential failures as possible and detect those that remain. Chaos Engineering helps in prevention of failures and early identification on any missed out detections.