Oncall or Not Oncall?

Microservices and other loosely coupled architectures create a new set of challenges for engineers who are responsible for reliability and performance: not only is the performance of each service important on its own, but understanding the way services interact is also a critical part of improving end-user experience. And while adopting DevOps practices creates tighter feedback loops between teams and their users, it often makes it more difficult to catch performance issues before they are pushed to production.

Whether you're oncall or not, you need to understand dependencies between services and the role each service plays in handling requests. In this presentation, we'll show how distributed tracing provides context that can:

• Map out dependencies in ways that are specific to each type of request
• Quickly connect changes in service level indicators (SLIs) back to their root causes
• Make sure that alerts are delivered to teams that are best positioned to address them
• Catch performance regressions earlier in your deployment pipeline
• Inform performance optimization work so you are making the best use of your time

We'll also talk about the steps you need to take to leverage distributed tracing for your application and your organization, including when and how to instrument your code and how to manage the costs of your observability solution as a whole.