Start with constraints and workload

Data classes, residency, latency, volume and workload variability define the real choices. Include connectivity to systems that already run locally.

One organisation may need several valid environments.

Compare operating models

Cloud provides managed services but requires cost control and new skills. On-premise offers direct control while placing upgrades and scaling on the team.

  • Security and compliance
  • Total cost
  • Team capability
  • Recovery plan

Design portability

Document data, transformation code and metric definitions so they can move. Hybrid architecture needs especially clear boundaries.

The platform should support analytics and AI growth without creating a technology trap.