Infrastructure Decisions at a Startup: Endorsements and Regrets After 4 Years

The text describes the author's infrastructure decisions and whether they endorse or regret them after running infrastructure at a startup for 4 years. The decisions cover various aspects, such as cloud providers, managed services, process automation, software tools, and infrastructure choices. Some of the key points include endorsing AWS over Google Cloud, endorsing the use of RDS and Redis ElastiCache, regretting the use of EKS managed addons, and endorsing the use of GitOps and prioritizing team efficiency over external demands. Additionally, the author expresses regret for not adopting an identity platform early on and not using open telemetry from the start.

• Endorsements:
  • Picking AWS over Google Cloud
  • EKS
  • RDS
  • Redis ElastiCache
  • ECR
  • Using pager duty’s incident templates
  • Moving off JIRA onto linear
  • Picking Flux for k8s GitOps
  • Using ExternalSecrets, ExternalDNS, and cert-manager to manage infrastructure components
  • Go for services
• Regrets:
  • EKS managed addons
  • Not using Function as a Service(FaaS) more
  • Manging post mortems in datadog or pager duty
  • Not adopting an identity platform early on
  • Not using open telemetry early
  • Using SealedSecrets to manage k8s secrets
  • Bottlerocket for EKS
  • Not using more code-ish IaC solutions (Pulumi, CDK, etc)