Cloud & serverless

Serverless (FaaS)

Run functions on demand without managing servers

Upload functions triggered by HTTP, queues, or schedules; cloud provider handles scaling and patching. Great for spiky or infrequent workloads; watch cold starts and vendor coupling.

Growth scalemedium complexity

Architecture diagram

High-level component relationships

Key components

Functions

Stateless handlers with timeout limits

Managed triggers

API Gateway, S3, SQS, EventBridge, cron

Managed data

DynamoDB, RDS Proxy, object storage

IAM & secrets

Per-function least-privilege roles

Data flow

Request triggers function instance (cold or warm start)
Function reads/writes managed services, returns response
Async work offloaded to queue-triggered workers
Billing per invocation and duration

Pros

Near-zero ops for scaling and patching
Pay only for actual usage — cost-efficient at low/spiky traffic
Fast iteration for APIs and glue logic

Cons

Cold start latency for user-facing paths
15-minute (typical) execution limits
Debugging and local parity are harder
Vendor lock-in on proprietary integrations

When to use

Webhooks, image processing, scheduled jobs
APIs with unpredictable traffic
Prototypes and event reactions alongside managed services

When to avoid

Long-running batch jobs or WebSocket-heavy realtime
Strict on-prem or multi-cloud portability requirements
Sustained high RPS where reserved compute is cheaper

Real-world examples

Image thumbnail pipelines
Webhook processors
Scheduled report generators

Related technologies

AWS LambdaAzure FunctionsCloudflare WorkersStep Functions