Use cases · Leave Kubernetes overhead

Keep the appsDrop the cluster work

Web, workers, and data in production. You do not run Kubernetes to ship.

ticket

cluster ops

charts · ingress · upgrades

punched

worker node

web · worker · database

Before

The platform became another product.

A product change still touches manifests, CI, networking, and a rollout.

Off-ramp

The apps move. The cluster work does not.

Web, worker, and database become services. Charts and ingress stay behind.

cluster

  • charts

    owned by the team

  • ingress

    manual TLS

  • rollouts

    CI glue

dFlow

  • web

    service on a worker node

  • worker

    same deploy history

  • database

    private reference

Scale

CPU, memory, replicas. Per process.

Capacity is a process setting, not cluster tuning.

storefront-web

cpu

memory

replicas

2

Network

Domains stay in the service flow.

Hostname and SSL follow the app. Ingress yaml does not.

was

ingress.yaml

Routing and certs as cluster objects.

now

app.up.dflow.sh

Preview hostname, then custom SSL on the service.

off-ramp

Keep production.Drop cluster ops.

worker node release

Use the same deploy path on dFlow compute, your cloud, or an existing VPS.