After years of using @awscloud Aurora, we are moving back to dedicated hardware. MySQL K8s operators are great, storage is cheap, memory is cheap, cpu is cheap, I can run 5.7 as much as I like and no AI. I'll miss database cloning and instant read replicas. 1/n
It was a TECHNOLOGY company, now it is selling AI bs. I don't care about videos of people with 3 legs. I care about Lambda getting faster, having shared backend storage, SnapStart being properly implemented and ... whatever. 3/n
Long term we will keep using S3, Route53, DynamoDB, apigw and lambda for some risky endpoints but we will definitely turn off most of our Lambda efforts. The main reason is AWS losing its mind over the last year. #awsexodus and zero investment in anything non-AI makes me cry 2/n
My bill went up 50% during the last the year. Mostly because of Aurora. Any new tech that comes out of that kitchen is at least 4 times more expensive: Serverless V2! Ok, there were some great surprises like Optimized IO but still 50% ... 4/n
our monthly lambda bill is 1000$ What is serverless in setting memory and cpu and choosing between lambda and fargate and fargate not being able to listen to sqs events, etc. With no improvements in serverless for years. Poor execution of SnapStart made me super angry. 5/n
We use Java and @quarkusio and we were always able to operate in on-premises K8s and AWS lambda world with almost no code changes. Limited use of SQS, smart mid layers for S3 and authorization, Redis, MySQL with our own implementation of Queues and State machine. that's it. n/n
MySQL course by Meta released this week. Database that runs the Internet. From AWS application data planes, Wordpress, Jira to Meta. youtu.be/iwRneX7GIGI