Configuration templating vs configuration as code, 12th round

Im sorry to say that, but Infrastructure is everywhere; no matter if you are using k8s, GCP, AWS, or any other offering… you need to describe the infrastructure requirements your application needs to run… with code? Are we really using code, or are we back to configuration management? Should we?
We will be reviewing how we have been interacting with Cloud Providers in the last 10 years, using scripting, actual code, and configuration files. Code snippets will be supporting us in this time travel, while we will be blaming the Twelve-Factor App methodology (and ourselves?) given we ended up messing with templates… Some newer alternatives will be presented (Terrascript and Pulumi), including also the patterns we are applying in k8s (Helm, Kustomize and others), sharing finally some thoughts about the convenience of changing (or not?) the model.


  • Audience should be able to position different IaC models and complexities behind cloud deployments
  • We should acknowledge limitations and issues generated by describing infrastructure configuration with declarative languages, getting a clear idea of the pros and cons of each approach
  • The audience should leave the session with interest in playing with the new models and tools we will be presenting (mostly Pulumi), with the guidance also to the examples we will be providing, as well as having a better view of what the future of IaC may consist of.

“Senior SRE Engineer at Fastly, also part-time lecturer in La Salle University (DevOps program I had the honour to introduce), previously working as Staff Devops at Schibsted, NTT, Oracle (and others!), where I have been trying to code, build and maintain backend services, infrastructure and (automation) tools. In parallel, author and maintainer of tcpgoon and gcs-pit-restore, but also contributing to OSS as much as I can”

This post is also available in: German