serverloze

Serverloze, known as serverless computing, is a cloud execution model where the provider runs and scales code, manages servers, and charges only for usage. The term is a misnomer: servers exist, but developers do not provision or manage them. Workloads are stateless functions.

Code is deployed as functions triggered by events such as HTTP requests, file uploads, database changes, or queue messages. The provider handles resource allocation and can scale these functions up or down automatically. Many serverless platforms also offer backend-as-a-service components, such as managed databases, authentication, and messaging, to support backends.

Key characteristics include automatic scaling, pay-per-use pricing, and reduced operational duties. The model favors event-driven, short-lived tasks and architectures that are stateless or store state externally. Vendors impose limits on execution time, memory, and concurrency, and performance may vary due to cold starts after inactivity.

Advantages include lower upfront costs, efficient resource use for irregular workloads, and faster deployment. Drawbacks include cold starts, vendor lock-in, observability challenges, latency variance, and the need to design applications as stateless services with external state.

Common use cases are API backends, data processing pipelines, real-time file processing, automation tasks, and integration between SaaS services. It is also used for small, event-driven microservices and prototypes.

Examples include AWS Lambda, Google Cloud Functions, Microsoft Azure Functions, IBM Cloud Functions, and Oracle Functions. Some platforms also provide serverless containers or FaaS-on-Containers alternatives for longer tasks or specific compliance needs.

Pricing is typically based on execution time, memory, and invocations, plus data transfer or external service costs. While serverless can reduce costs for sporadic workloads, it may be more expensive for steady, high-volume tasks and can complicate testing.