Immutable infrastructure with Terraform
This document summarizes a talk on immutable infrastructure with Terraform. It begins by covering the devops journey from manually managed on-premise hardware to configuration management tools. It then discusses the rise of cloud computing and challenges it introduced. Next, it defines immutable infrastructure and contrasts it with mutable infrastructure. The rest of the document focuses on Terraform, explaining what it is, why it is useful, and demonstrating how to use it through examples like deploying a web app on EC2, handling failures and incremental changes, using remote state and locking, and best practices for teams. It also shows examples of rolling deployments, output variables, supporting multiple environments, and using modules.
Immutable infrastructure with Terraform
- 1. Immutable Infrastructure with Terraform - Prashant Kalkar & Rahul Baphana (Expert Talk - Pune - 9 June 2017)
- 3. Limited hardware, manually managed. On Premise Hardware
- 4. » Limited hardware » Configured generally once. » Long running. » Servers might not look identical. » Change propagation requires changes in all the server instances. » Manageable with small no of servers. On-Prem Hardware
- 5. » Client Server Architecture » Easy to propagate configuration changes. » Instances mostly look identical. » Long running server. » Can face issues of Configuration drift. Configuration Management Tools
- 7. » Elastic » Disposable resources. » On-demand » Cost effective Cloud properties
- 8. » Infrastructure Provisioning. » Configuration of new instances. » Delays at server runtime for server configuration. » Possible failure of server configuration at runtime. Challenges in the Cloud
- 10. Machine Image / Container Ubuntu Chef / Puppet client Manages EC2 Instances Configuration Management Tool App artifact Java Tomcat Logstash client Mutable Infrastructure
- 11. Machine Image / Container Ubuntu Server Templating Tools App artifact Java Tomcat Logstash client EC2 Instances Chef / Puppet client Manages EC2 Instances Configuration Management Tool App artifact Java Tomcat Logstash client Immutable Infrastructure
- 12. 3 Terraform
- 13. » Declarative style for infrastructure as code » Supports multiple cloud providers » Easier to adopt for new teams » Open source » Under active development Why Terraform
- 15. » Sample web app on EC2 instance » AMI for spring-boot-app Demo
- 16. » No client-server » Terraform tfstate (show command) » Resource dependency & Parallel Execution » Terraform failures » Incremental changes » Non-managed infrastructure is invisible to Terraform How Terraform is working?
- 17. » Remote state » State locking » Practices : ⋄ Use centralise server to apply Infrastructure changes ⋄ Always apply Terraform on a pushed code base ⋄ Branches or no branches? Using Terraform in a team
- 18. » Rolling deployments » Output variables Demo
- 22. » Supporting multiple environments ⋄ Per environment tfstate (recommended) ⋄ Keep state as minimum as possible ⋄ Read only state ⋄ Use profiles Demo
- 23. » Using modules to dry out Terraform code ⋄ Relative modules ⋄ Terraform Variables ⋄ Modules with versions Demo
Editor's Notes
- #4: Limited hardware They are generally configured once with required applications. Once created test to use them for long period of time. Long running servers - Need to keep them running. Instances might not look identical.
- #16: Simple instance without name Add name incrementally Add security group
- #17: Failures : Availability zone failure in demo 03
- #19: Environments
- #23: Demo for variables Environment tfstate
- #24: Modules in Environments