s0sta
EN NL عربي
Filters
How to search
  • Plain words find names, texts and examples.
  • "two words" finds the exact phrase.
  • --name or -x finds an option.
  • #word looks only inside commands.
  • @grep finds lessons and labs that use grep.
  • Filters: cat:network · in:labs · in:notes · level:rhcsa · ch:8 · root:yes. Put ! before a word to leave it out. OR between words means one of them.
  • ↑ ↓ to move · Enter to open · Esc to close · / to search

Cloud: AWS & AzureVisual series

A serverless web app on AWS, drawn

The first serverless architecture you draw looks empty: six managed services and not one server to patch. The request path still has three tiers, and each tier fails differently.

Quick reference

  • Tier 1 · Edge: Route 53 (DNS) · API Gateway (API)
  • Tier 2 · Compute: Lambda (FN) · SQS queue (SQS)
  • Tier 3 · Data: DynamoDB (DB) · S3 objects (OBJ)

Request flow: Route 53 → API Gateway → Lambda → DynamoDB; Lambda → S3 objects; API Gateway → SQS queue → Lambda

Explained

Edge Route 53 answers DNS and API Gateway is the front door. API Gateway can check the caller, applies throttling and sends the request to an integration; it can invoke Lambda directly or put a message on a queue. Neither service is a machine you own, which is the point.

Compute Lambda runs the request. Fifteen minutes is the hard ceiling per invocation, so it suits short work rather than a nightly batch. When work can wait, API Gateway writes to SQS and Lambda polls the queue in batches through an event source mapping, adding concurrency as the backlog grows. The queue is the part that survives a traffic spike.

Data DynamoDB stores the session and the order, S3 stores the uploads, and the execution role is what lets Lambda write to either. Permissions live in that role, not in the function code. A VPC is optional here: add it only when the function must reach a private database, and then give it a NAT gateway for outbound internet calls.

The honest caveat: serverless moves the operational work rather than deleting it. Cold starts, concurrency limits and per-request pricing become your capacity plan, and a retry storm on a queue can bill you twice for the same failure.

The lesson: in a serverless diagram the arrows are permissions and queues, not cables.

In one line

Draw the arrows, not the servers. In serverless, the arrows are permissions and queues.

Follow the function

aws logs tail /aws/lambda/fn --follow

More notes: Cloud: AWS & Azure

Open in the notes app: search, filters, copy This note on GitHub