#58.2 - 10.12.2019
#58.2 5 reasons why developers do not buy in the observability hype
Do observation tasks have to be boring, tedious and tiring?
During my speech, I will talk about typical solutions used to monitor applications built on the basis of microservices. I will use an example of a Java application running on the Google Kubernetes Engine. With its help, I will demonstrate the challenges associated with monitoring microservices: “high churn”, “high cardinalitylabeling”, problems with rarely reported metrics, and potential difficulties in choosing a sampling strategy.
I will tell about monitoring and tracing both from the perspective of a programmer who would like to keep the code clean, and an architect who tries to meet the requirements of observability, but at the same time does not want the team implementing them to forever fight with the tools he chose. In the end, I also comment on the trend “observability as a code”.
Marta is a Software Engineer at SignalFx, a Splunk company, where she implements monitoring features for various cloud services. Prior to that, she worked in Motorola Solutions, contributing to mission-critical public safety software. For the last 8 years, Java has been Marta’s main language. Having struggled with monitoring cloud-hosted microservices herself, now she writes tools to lighten this burden for other developers. She considers keeping up with changing tools and shifting paradigms one of the biggest challenges in her job.