The starting point is business impact, application dependencies, data, users, recovery and ownership—not a preselected product.
Private, public and hybrid are design choices—not rankings. Suitability depends on application dependencies, data, demand, recovery and operating constraints.
Pattern A
Dedicated resources for selected stable, sensitive or business-critical workloads, designed around application, data, resilience and support requirements.
Pattern B
Private cloud connected to approved on-premise systems and optional public-cloud services, with identity, monitoring and data boundaries treated as one design.
Pattern C
A primary environment paired with backup and disaster-recovery options selected against business impact, recovery objectives and test procedures.
Business, technical, data and commercial reasons with stated assumptions.
Application traffic, administration, logs, replicas, backups and third-party integrations.
Provider, customer and joint responsibilities during normal operation and incidents.
Acceptance criteria, rollback, recovery tests, evidence and review cadence.
No model is inherently secure, economical or suitable. Architecture, configuration, governance and operating discipline determine the result.
| Decision area | Private cloud | Public cloud | Hybrid estate |
|---|---|---|---|
| Best fit | Stable, sensitive or specialized workloads needing dedicated control. | Rapid experiments, variable demand and access to broad managed-service ecosystems. | Estates that need deliberate workload placement across private, public and existing systems. |
| Control model | Greater design control, with capacity and lifecycle planned in advance. | Provider platform plus customer configuration under a shared-responsibility model. | A joined governance model is required across multiple platforms and teams. |
| Scaling | Planned against physical or contracted capacity and lead times. | Can expand quickly, subject to service, region, quota and architecture constraints. | Selected services can absorb changing demand while core placement remains intentional. |
| Commercial model | More scope up front; may improve predictability for suitable stable workloads. | Fast entry and usage-based charges; governance and data transfer affect total cost. | Requires an end-to-end model covering both environments, integration and operations. |
Pricing and commitment details remain TBD. These are the inputs a transparent assessment should cover.
Start with workload placement, connectivity, data flows, recovery objectives, migration risk and the full commercial scope.