Serverless Containers

ELI5 - The Vibe Check

Serverless functions make you squeeze your code into a tiny handler function with a strict shape. Serverless containers say fine, just give us a Dockerfile, run whatever you want inside it, in whatever language, for as long as a real request takes, and we'll still scale it to zero and bill you by the second. It's the container flexibility you already know, minus the part where you manage the servers underneath it.

Real Talk

Serverless containers run arbitrary containerized workloads with the operational model of serverless functions: automatic scaling (including to zero), per-second billing, and no server or cluster management required from the developer. Services like AWS App Runner, Fargate, Google Cloud Run, and Azure Container Apps sit between raw Kubernetes and narrow FaaS platforms, supporting long-running processes, arbitrary runtimes, and larger request bodies that traditional FaaS handlers struggle with.

Show Me The Code

# Deploy any container as a serverless service (Cloud Run)
gcloud run deploy my-api \
  --image gcr.io/my-project/my-api:latest \
  --min-instances 0 \
  --max-instances 20 \
  --memory 512Mi

When You'll Hear This

"We moved the video processor off Lambda to Cloud Run, the 15-minute timeout was killing us." / "Fargate gives us serverless containers without touching Kubernetes."