Alternative Ways To Think About Serverless
summary
Benchmarks on AWS compare Lambda to instances using a custom “unit of compute” (large primes/factorials) and storage tests in S3 versus RDS. Surprising results: smallest instances give the best $/compute, RDS micro sustains 4,000 TPS, S3 reads hit ~30 ms, and Lambda shows doubling memory doubles compute so cost per unit stays constant. Cost crossover is clear — point of parity at one instance — and operational/dev overhead, outage risk and feature velocity favor serverless for smaller, event-driven workloads.
AI-written summary · about
chapters (4)
- 0:00Start
- 4:40IS SERVERLESS RIGHT FOR MY PROJECT?
- 10:08BUT WHAT ARE THE REAL COSTS?
- 11:50Code encapsulation
related talks
01h 00mServerless - reality or BS - notes from the trenches
30mPerformance Tales of Serverless
19m(N+1) Lessons I learned about serverless
58mServerless: Five Key Things you need to Know
54mServerless beyond Functions
35mEfficiency of Serverless Architectures with AWS Solutions
42mNo Server Is Better Than No Server: Building Blocks for Serverless Architectures
44mServerless Architecture on Production: Good, Bad and Ugly Parts