Cloud Deployment Models
Public cloud, private cloud, hybrid, and multicloud — and how to recognize which model a scenario describes.
Three (plus one) ways to deploy
| Model | What it means | Typical reason |
|---|---|---|
| Cloud (public cloud) | Everything runs on a cloud provider like AWS. | Full cloud benefits: elasticity, pay-as-you-go, global reach. |
| On-premises (private cloud) | Infrastructure runs in your own data center, sometimes with cloud-like virtualization tools. | Full control; legacy systems; strict data residency rules. |
| Hybrid | Some workloads on AWS, some on-premises, connected together. | Gradual migration; keeping sensitive systems local while extending to the cloud. |
| Multicloud | Using two or more cloud providers (e.g., AWS + Azure). | Avoiding vendor lock-in; using best-of-breed services. |
A scenario that connects a company's existing data center to AWS — via AWS Direct Connect or a Site-to-Site VPN — is describing a hybrid deployment. That word association alone answers several exam questions.
AWS services that enable hybrid
A dedicated, private physical network connection from your data center to AWS — more consistent than the public internet.
An encrypted tunnel over the internet linking your on-prem network to your AWS VPC — quick and low-cost hybrid connectivity.
AWS-managed racks of AWS hardware installed *in your own building*, for workloads that must stay on-site but want AWS APIs.
Lets on-prem applications seamlessly use AWS cloud storage — a hybrid storage bridge.
Public cloud is renting apartments in a managed building. On-prem is owning your house and fixing everything yourself. Hybrid is owning a house but renting a storage unit and a downtown office — connected by the roads (Direct Connect / VPN) between them.
A company runs its ERP system in its own data center and its customer-facing website on AWS, with a VPN connecting the two environments. Which deployment model is this?