Serverless
Serverless is the cloud-computing model in which the cloud provider runs and scales the underlying infrastructure transparently — the developer deploys code (functions, containers) and the provider handles servers, capacity, and most operational concerns. Pricing is usage-based (per request, per GB-second) rather than per provisioned instance.
Serverless covers a spectrum: FaaS (Lambda, Cloud Functions) is the strict definition; managed databases (DynamoDB, Aurora Serverless), managed queues, and edge runtimes are commonly grouped under the same umbrella. The economic model favours bursty or low-traffic workloads where the per-request cost beats the provisioned-instance cost; high-throughput steady-state workloads usually cost more on serverless than on right-sized provisioned instances. Other trade-offs: cold-start latency, vendor lock-in (most serverless code is provider-specific), and observability/debugging that's harder than for traditional servers. The strongest fit is event-driven async work — webhook handlers, scheduled jobs, fan-out processors.
Related terms
- Function as a service (FaaS)
Function as a Service is the serverless compute model where the deployment unit is a single function with a defined entry point — the provider invokes it in response to events (HTTP request, queue message, scheduled timer) and bills per millisecond of execution.
- Edge function
An edge function executes at the CDN's edge locations close to the user — typically with sub-millisecond cold starts (V8 isolates) and ~10-50ms latency advantage vs origin-region functions.
- Event-driven architecture
Event-driven architecture is the design pattern where services communicate by emitting and consuming events rather than by direct synchronous calls.