Reference Architecture: Overview
Purpose: For platform engineers, explains the layered architecture model from physical hardware through Kubernetes workloads.
Layer Model
Physical Layer
The foundation of any deployment:
- Compute - Server hardware specifications
- Network - Switching and routing
- Storage - Disk configurations
- Power & Cooling - Environmental requirements
- Rack Layout - Physical placement
- Cabling Standards - Connectivity
Infrastructure Layer
Services required before Kubernetes:
- Hypervisor - vSphere or KVM/OpenStack
- Virtual Networking - Overlay networks
- Virtual Storage - Datastores and policies
- Infrastructure Services - DNS, NTP, DHCP
Platform Layer
Kubernetes and supporting services:
- Control plane (HA with 3 nodes)
- Worker nodes (scaled to workload)
- Platform services (cert-manager, Kyverno, etc.)
- Observability stack (Prometheus, Loki, Tempo)
Application Layer
Customer workloads:
- Stateless applications
- Stateful services
- Managed data services (Kafka, PostgreSQL)