Infrastructure Plan

Meeting notes 2024-11-12

Discussed plans for new infrastructure:

• we wish to move all infrastructure except for the pods/daemonsets currently running in the platform cluster to the new azure subscriptions in the short term
  • this includes (but is not limited to):
    • postgres flex server
    • key vaults
    • application insights/logg workspaces
  • this requires that the new infrastructure (databases, key-vaults, etc.) is reachable from the old cluster
• we wish to consolidate all compute resources running outside the platform cluster to be of a single type (either container apps or Deployments in a cluster) in the medium term
  • the decision on whether to go for AKS and use kubernetes directly, or use container apps hinges on whether or not the platform team can manage AKS maintainance for us
• we plan to make all new deployables using the strategy decided on in the previous point - the first one out is metadata-api
• we REQUIRE that all of the components (BOTH infrastructure AND application versions) are deployable from github pipelines
  • we have a strong preference for a deployment scheme where we can know in a pipeline when the deployment is complete (and if it failed) such that for instance E2E tests can be run and the deployment can be rolled back if required.