Monitoring serverless applications on AWS: the monitoring monster becomes your friend.
summary
Built a conference-session demo using AWS Lambda, SNS, DynamoDB and Slack notifications and shows concrete results: structured logging with a propagated trace ID lets CloudWatch correlate logs and find an exception (stack trace at line 150), CloudWatch Logs Insights runs queries to pull milestones, X‑Ray visualizes flow and pins a ~3s bottleneck inside a lambda, alarms and automated smoke tests plus Artillery load tests reveal retries (3 attempts) and lead‑time spikes from ~10s to ~220s due to queuing.
AI-written summary · about
chapters (22)
- 0:00Intro
- 2:00Agenda
- 3:00Serverless is a new normal
- 4:00Why is Serverless important
- 4:30Infrastructure as Code
- 5:05MMO Landscape
- 5:50Conference Application
- 6:45Adding Session
- 8:20Milestones
- 9:25Challenges
- 9:45Observability
- 10:50Three Challenges
- 11:30Where did it go wrong
- 13:00Happy flow
- 15:40Logging
- 17:45Logging Challenges
- 21:35Structured Login
- 27:15Structured Logs
- 29:10Xray
- 34:35Testing Remote System
- 35:00Testing in the Cloud
- 38:18Load Test Results
related talks
41mLive Coding on Observability for Serverless
34mObservability in Serverless Application
39mObservability in DevOps
24mObservability across serverless asynchronous managed services
28mMeasuring Serverless Apps in Amazon Bedrock
17mHow do we use Serverless Architectures on AWS Lambda
53mLook Ma! No servers!
44mServerless Architecture on Production: Good, Bad and Ugly Parts