SRE with Java Microservices

Chapter 5. Safe, Multicloud Continuous Delivery

The placement of this chapter in the second half of this book should signal the importance of telemetry in achieving safe and effective delivery practices. This may come as a surprise, because every organization ships software with an emphasis on testing as a means of guaranteeing safety, but not every organization measures it actively in a way that is directly related to end-user experience.

The concepts in this chapter will be introduced with a continuous delivery tool called Spinnaker, but as with earlier chapters, a different tool could achieve similar ends. I’d like to establish a minimum base for what you should expect from a worthy CD tool.

Spinnaker is an open source continuous delivery solution that started at Netflix in 2014 to help manage its microservices in AWS. It was preceded at Netflix by a tool called Asgard, which was really just an alternative AWS console organized with application developers in mind, and built for Netflix’s unusual scale of AWS consumption. At one point, I was interacting with an AWS console form that required me to select a security group. The UI element in the console was a plain HTML select (list box) with four visible elements. The available security groups were unsorted in the list box, and there were thousands of them (again, because of Netflix’s broad scale in this account)! Usability issues like this led to Asgard, which in turn led to Spinnaker. Asgard was really just an application inventory ...