Serverless Architecture

Serverless uses managed services and functions to handle compute on demand. You pay per use and delegate scaling, patching and capacity planning to the platform.

When it fits

• Spiky or unpredictable workloads
• Small teams that want to minimise operations

Risks

• Cold starts and provider limits
• Tight coupling to cloud-specific services

Real-world examples

• Media processing pipelines built on object storage events triggering functions for transcoding and thumbnails
• Public APIs running on API gateways with functions and managed databases, scaling to zero when idle

Practical tips

• Design for stateless functions; push state to managed stores
• Use queues and retries to smooth traffic and handle failures