Deployment and operations

Deployment determines where services run and who operates their runtime and infrastructure. Platform operations maintain that deployment as traffic, application versions, and infrastructure change.

Deployment options

The deployment model describes where Akka runs and who operates it. Your service uses the same runtime and component model across deployment options; operational responsibilities change.

Who operates what

Option Infrastructure Operator

Serverless

Shared infrastructure in Akka’s cloud

Akka

Dedicated

Single-tenant infrastructure in Akka’s cloud

Akka

BYOC

Your cloud account

Akka operates the platform

BYOK8s

A Kubernetes cluster you provide

You operate the cluster; Akka operates its platform components

Self-managed

Infrastructure you choose

You

Managed options use Akka Automated Operations for deployment and platform operations. Self-managed operation gives you control of persistence, networking, certificates, and runtime configuration, with the work needed to operate them.

Local development runs services without a cloud deployment and provides a console for inspection. It is a development environment, not another production responsibility model.

Services, regions, and planes

A service is the unit of deployment and scaling. Projects group services and their shared resources. A region contains the clusters and storage that run those services; multi-region projects replicate state between regional clusters.

The application plane runs services within regions. The federation plane coordinates organizations, access, and deployment across them. Application data remains in its configured regions, while defined management interactions cross to the federation plane.

A cluster stays inside one region. A project spanning clouds or data centers uses several regions, not one global cluster. See Multi-region operations.

Choosing an option

Choose according to infrastructure ownership, operational control, data residency, and recovery needs. BYOC and BYOK8s do not remove the federation-plane interactions or make every task Akka’s responsibility. Use the specific shared-responsibility model when planning operations.

Automated operations

Akka Automated Operations (AAO) operates the runtime in Serverless, Dedicated, your own cloud account, and your own Kubernetes. With self-managed operations, you perform these operational tasks yourself.

Runtime behavior and platform operations

The runtime owns component state, clustering, shard placement, and recovery. AAO manages the deployment around it: instance capacity, health, rolling updates, runtime patching, certificates, and keys.

Scaling changes the capacity available to a service; the runtime then rebalances state across its instances. AAO scales within the service’s configured bounds.

A service update can run two application versions concurrently. Persisted events and state must remain compatible throughout that overlap. Platform patching is a separate operation: updating runtime or infrastructure components does not require a new version of your application.

AAO also supplies platform monitoring and project isolation. The customer remains responsible for application code and dependencies, and for the infrastructure assigned to them by the deployment option’s shared-responsibility model.

Next steps