Choose between Serverless, Containers and VMs for a First Deployment

A practical approach

Step 1

Describe the workload: request-response web traffic, long-running jobs, WebSocket connections, scheduled tasks or heavy computation.

Step 2

Prefer serverless or edge platforms for spiky, request-based traffic with short operations and for teams that want minimal operations work.

Step 3

Prefer containers on a managed platform for steady services, long-running processes or apps with dependencies that serverless runtimes do not support.

Step 4

Choose virtual machines when you need full control of the operating system or software that cannot be containerised.

Step 5

Check the platform's limits, such as maximum execution time, memory, supported runtime APIs and database connection handling, against your application before committing.

Worked example

A team deploys a Next.js learning site with a PDF-generation feature to a serverless platform. Pages are fast and cheap, but generating large certificates hits the execution time limit. Keeping the site serverless and moving certificate generation to a small container worker fed by a queue gives each workload a suitable home.

A mistake to avoid

Do not choose virtual machines for a simple web app only because they feel familiar. You become responsible for patching, scaling, backups and security hardening, work that managed platforms do for you.

Try it yourself

Write down three workloads in a project you know, such as page rendering, file processing and scheduled emails. For each, choose serverless, containers or a virtual machine and note the platform limit that most influenced your decision.

Your implementation checklist

• Workloads are described before choosing a platform
• Spiky request traffic considers serverless
• Long-running and steady services consider containers
• Virtual machines are chosen for specific control needs
• Platform limits are checked against requirements