Cloud-native applications
Select and engineer the right Azure runtime for each application.
We compare managed web, serverless, container, and Kubernetes runtimes before building the APIs, messaging, supply chain, identity, telemetry, scale, upgrade, and recovery foundations around them.
Runtime decision
Workload drivers
- Protocol and execution model
- Scaling and traffic shape
- Isolation and networking
- Operational control required
Use the least operationally complex fit
- Runtimes
- Four primary choicesApp Service, Functions, Container Apps, and AKS.
- Integration
- API and messagingSynchronous and asynchronous boundaries are explicit.
- Lifecycle
- Build through recoverySupply chain, telemetry, upgrade, scale, and restore are included.
Runtime selection
Runtime choice follows application needs, not container preference.
Final selection also considers region availability, quotas, team skills, network design, and support requirements.
| Area | Strong fit | Trade-off to test | Customer ownership |
|---|---|---|---|
| App Service | Managed web apps and APIs | Runtime constraints and scaling model | Application, deployment, and configuration |
| Azure Functions | Event-driven and scheduled compute | Execution limits, cold start, and state | Function code, triggers, retries, and idempotency |
| Container Apps | Managed containers, jobs, revisions, and event scaling | Platform limits and network features | Images, revisions, scale rules, and dependencies |
| AKS | Kubernetes control and ecosystem requirements | Cluster lifecycle and higher management burden | Workloads plus agreed cluster duties |
Application foundation
The runtime is only one part of a production application platform.
Supporting services are selected around application boundaries and failure modes.
API and messaging
Design API Management policies, authentication, versioning, throttling, Service Bus or Event Grid messaging, retries, dead-letter handling, and idempotency.
- API Management
- Service Bus
- Event Grid
Supply chain
Secure repositories, builds, Azure Container Registry, image scanning inputs, environment promotion, approvals, provenance, and rollback.
- Azure Container Registry
- Azure DevOps
- GitHub Actions
Identity and configuration
Use managed identities, Key Vault references, private endpoints, RBAC, Policy, and environment-specific configuration.
- Managed identities
- Key Vault
- Azure Policy
Service behavior
Engineer telemetry, autoscale, health probes, deployment strategies, runtime or cluster upgrades, backup, and regional recovery.
Modernization workflow
Applications move from runtime selection to measurable production behavior.
A thin vertical slice validates platform assumptions before broad migration.
- 01
Characterize
Map traffic, state, dependencies, protocols, runtime constraints, deployment frequency, scale, and recovery needs.
- 02
Prove a slice
Implement one representative flow with identity, network, data, telemetry, release, and rollback controls.
- 03
Build and migrate
Increment application components, automate tests and infrastructure, migrate data, and validate integrations.
- 04
Harden and transfer
Exercise scaling, upgrade, failure, backup, recovery, alerting, and support procedures before service acceptance.
Engineering assets
Cloud-native delivery includes application and platform lifecycle assets.
The exact code boundary is stated by repository and component during scoping.
- 01
Runtime selection record
Requirements, compared services, limits, cost factors, team impact, risks, and selected target.
- 02
Reference implementation
Representative service, infrastructure code, identity, secrets, API or messaging integration, telemetry, tests, and pipeline.
- 03
Release design
Build, registry, security checks, configuration, promotion, deployment strategy, rollback, and upgrade ownership.
- 04
Service runbooks
Scaling, dependency failure, message backlog, deployment failure, secret rotation, restore, and regional recovery.
Application fit
Cloud-native work needs application ownership and production constraints from the start.
A runtime assessment can precede implementation when the current application is poorly understood.
Best suited to
- New Azure applications
- Selective replatforming
- Container and Kubernetes rationalization
Needed to begin
- Source code and build access
- Application owners and test scenarios
- Landing-zone, identity, and network prerequisites
Customer responsibilities
- Own product behavior and acceptance
- Provide data and integration access
- Approve runtime support and upgrade responsibilities
Not included by default
- Full application rewrite unless named
- AKS selection without a Kubernetes requirement
- Production support beyond the contracted service