Infrastructure as Code
What, Where, How, Why and we do it
Michael Schmid (he/him)
@Schnitzel
CTO amazee.io
What?
What?
- Server, Infrastructure Configuration in Code
- Scripts implement (provision) your configurations
basically:
Where?
Where?
Infrastructure
VMs / OS
Services
Application
VMs, Loadbalancers, Networking, Firewall, etc
OS configuration, SSH Keys, Tools, Monitoring, Services
Configs of Nginx, PHP, Varnish, etc.
Performance Settings, Security Configs, Domains, etc
How?
How?
- Configuration Management Tooling like
- Ansible, Puppet, Chef, Pulumi, Otter, SaltStack, CF Engine, Terraform, DSC, Docker
- Git repository for storage
- Push vs. Pull
- Declarative vs. Imperative
- Updatable vs. Non-Updatable
Updatable
- Most cfg mgmt tools
- Allows to update an already provisioned System
- More complex to test, can sometimes lead to weird cases
Non-Updatable
- Docker (Containers in general)
- Always provision from scratch
- No way to update an existing container
- Create a new container and replace it
- After Build done: much faster provisioned
Why?
Why?
- Recreatability
- Super easy scaling
- Broken? Just create again
- Self documenting
- Everybody can see what is installed
Why?
- Change Management
- Merge Requests for Review
- Git blame for historical data
- Automated Testing
- Create complete separate stack for a test
- Makes sure no changes on production have been missed
Containers
What?
- Containers bring IoC to Application Level
- Forces Developers to put everything into Git
- Dislike at beginning
- Understand benefits over time
IaC at amazee.io
Infrastructure
VMs / OS
Platform (Lagoon)
Application
Services
- Puppet and Ansible
- Weekly
- No automated testing yet
- Development Environments
- Manual changes and then copied to Ansible
Infrastructure
VMs / OS
Platform (Lagoon)
Application
Services
- OpenShift & Docker
- Weekly
- Open Source!
- Separate OpenShift for Testing Lagoon and Base Images
Infrastructure
VMs / OS
Platform (Lagoon)
Application
Services
- Docker
- Developers
- Lagoon Base Images
- Same Images everywhere
- Any adaptions possible
- No changes in running containers
Infrastructure
VMs / OS
Platform (Lagoon)
Application
Services
Pro
Con
Con
It takes time!
- Configuration
- 1 min for the fix
- 2 hour to add to cfg mgmt
- Testing
- Lagoon: 1-2h for full test
- Allow to run partial tests
- Rollouts
- easily 20min Ansible Run
Con
- Needs additional knowledge
- don't mix too many tools
- Discipline needed :)
Pro
- Super easy scaling
- Less errors
- Faster Rollouts
- Higher Security
- Self Documentation
and most important: