Deploying changes can be a big deal when working with high-frequency, highly-available distributed systems. How do you detect if something went wrong? From there, how do you measure the impact? How do you track performance changes across thousands of nodes?

In this talk, learn what systems and best practices Braze implemented to monitor and understand the effects of deploying changes to a massive, always-changing distributed system where performance is critical. Whether you're deploying new features in code or adding a new class of infrastructure, no technology works out of the box at scale - a lesson we learned the hard way with one of our latest new features, Content Cards.