1.9 KiB
1.9 KiB
proposal: how to (re)structure our infrastructure setup
what we have:
- different providers - currently OTC and kind
- different groups of environments - currently called 'environment_namespace'
- different environments in each group of environments
- environments are called 'cluster_environment' and are terragrunt stacks, i.e. a huge set of resources where the kubernetes cluster itself is only one of them
- the top domain name (like t09.de, buildth.ing) is bound to the environment group
issues:
- in provider OTC there are two attributes for the environment group - 'tenant' and 'environment_namespace' are the same
- missing hierarchy for other providers than otc - we directly start with 'prod' and 'non-prod' right now
opportunity
right now we start switching to go code. we can use this port to also restructure, generalize and improve the infrastructure and platfrom-provisioning.
proposed environment hierarchy
- new: the 'environment_namespace' layer is called 'platformzone': "A neutral, abstract term that encompasses both the isolation (account, tenant, subscription) and the operational context (prod vs. non-prod)." (ChatGPT)
- new: folder 'environments'
- new: level 1 hierarchy 'provider'
- old: level 2 hierarchy 'platformzone'
- old: level 3 hierarchy 'environment'
visualization
environments
├── provider:kind
│ └── platformzone:non-prod e.g.vm-1 or local
│ ├── edp
│ └── observability
├── provider:otc
│ ├── platformzone:non-prod i.e. tenant-1
│ │ ├── central-forgejo-aka-edp
│ │ ├── forgejo-dev
│ │ └── observability
│ └── platformzone:prod i.e. tenant-2
│ ├── edp
│ └── observability
└── provider:scaleway
├── platformzone:non-prod e.g. account-1
└── platformzone:prod e.g. account-2
├── edp
└── observability