Some solid ops content this week, on the changing role of operations, NAT networking, tagging cloud resources, logging, testing infrastructure as code, and more.
env0 sponsors Devops Weekly
When developing for the cloud, separate branches aren’t enough anymore - you want full on separate environments for whatever you’re working on.
A great post on the changing role of operations. Some good tips for those wondering what modern ops looks like, with tips on vendor management, outsourcing infrastructure and the importance of understanding sociotechnical systems.
A good introduction to NAT networks, for anyone wanting to understand this area of networking better. Good diagrams and examples and lots of details.
Metrics are used for lots of different purposes, including reporting to the top of an organisation. This post explores engineering KPIs for board room conversations.
Ever wanted to ensure that messages between services are kept in order, with a retry mechanism for any lost messages? This post describes a specific pattern, but is also part of a set of articles on distributed computing patterns that’s worth exploring.
Incident reviews are increasingly common but often hard to do well. This video and detailed transcript has various tips for improving the process.
An ambitious idea for a new journal for Systems research. Definitely relevant to the interests of some readers of Devops Weekly I think.
Pulumi, the Infrastructure as Code tool, now supports using Open Policy Agent to validate the resulting resources. This post explores why and how.
Even if you’re not writing applications in Java, it’s often useful to have some knowledge of how logging works as you’ll probably end up running at least some Java applications. These posts provide a solid foundation.
Tags are critical to managing AWS resources at scale. Awstaghelper provides a command line tool to ease adding and managing tags to and from CSV files across the wide range of AWS resources.
The GitOps Toolkit is a set of composable APIs and specialized tools that can be used to build a Continuous Delivery platform on top of Kubernetes. They should provide the underpinnings for the v2 of Flux, but could also be used to build other interesting high-level tools that take the same control loop approach.
Kip is a Virtual Kubelet provider that allows a Kubernetes cluster to transparently launch pods onto their own cloud instances. Handy if you require additional workload isolation.