Efficiency of Serverless Architectures with AWS Solutions
summary
Claims concrete numbers and outcomes: cold starts measure about one second, Lambda deployments cut from 40 minutes to 3 minutes, execution limit is 15 minutes so long jobs use a sequential recursive invocation via SQS, provisioned concurrency and dummy warmups fail to stop intermittent latency spikes, migrating the Checker service to ECS proxy on Fargate (NLB + EFS + S3 + LetsEncrypt + Route53) fixes ICMP support and stabilizes response times under 50 ms while raising per-run cost compared with Lambda.
AI-written summary · about
chapters (15)
- 0:00Serverless architectures intro
- 1:03Serverless performance myths
- 2:34Avoiding vendor lock-in
- 3:42AWS serverless services
- 8:23Serverless efficiency benefits
- 10:58Serverless limitations
- 12:01Saga pattern with AWS
- 14:52Lambda as microservices
- 17:00Inter-service communication
- 18:28Lambda execution limits
- 19:33Real-world architecture challenges
- 23:59Migrating to ECS Fargate
- 24:41Custom status page
- 26:51Custom domains and SSL
- 33:57Final architecture results
related talks
47mAdvanced Serverless Architectural Patterns on AWS
30mPerformance Tales of Serverless
44mServerless Architecture on Production: Good, Bad and Ugly Parts
01h 00mServerless - reality or BS - notes from the trenches
29m2 Years, 0 Servers: Lessons learned from running a startup 100% on Serverless
54mServerless Architectures: no servers allowed!
15mAlternative Ways To Think About Serverless
27mServerless Performance on a Budget