serverless cost optimization for cheapskates

Your serverless bill just made you question your life choices. Welcome to the club.

Here’s the dirty secret about serverless: it’s cheap as hell when you do it right, and expensive as fuck when you don’t.

The “I’m bleeding money” starter pack:

Lambda functions that run for 14 minutes and 59 seconds. Congratulations, you’ve invented the world’s most expensive cron job. 1

API Gateway with no caching. Every request hits your Lambda. Every. Single. One. Your wallet is crying. 2

DynamoDB with provisioned capacity set to “whatever the console suggested.” Spoiler alert: the console wants your money.

The “I’m actually smart about this” approach:

Use Lambda’s ARM-based Graviton processors. Same performance, 20% cheaper. It’s literally free money. 3

Set up DynamoDB auto-scaling or go on-demand. Stop paying for capacity you’re not using at 3am. 4

Cache everything at API Gateway. That $0.0000002 per request adds up when you’re doing millions of them. 2

Pro cheapskate moves:

EventBridge is cheaper than Lambda for simple routing. Stop using Lambda as a glorified if/else statement. 5

S3 event notifications are free. Lambda invocations from S3? Standard pricing applies, but no extra notification fees. Use this. 6

CloudWatch Logs cost money to store. Set retention periods or watch your “free” serverless architecture slowly bankrupt you. 7

The nuclear option:

Compute Savings Plans for Lambda. Commit to consistent usage and save up to 17% on compute costs. It’s like reserved instances, but for serverless. 8

Serverless isn’t about spending less money. It’s about spending money more efficiently.

There’s a difference, and your bank account knows it.

Generated by Amazon Q (Claude Sonnet 4.5) - because even AI assistants have opinions about your AWS bill.