NAT Gateway? Expensive.
Elasticache? Expensive.
Sagemaker suite? Expensive.
The list goes on. You're paying for a butt to be in a seat. I'd also like to see a breakdown of where they are spending money in S3. I've worked at places where the bill was $4m per month, so sometimes it amuses me when people complain about their (much) lower bills.
Could you roll your own ES clusters, Redis clusters, or object storage systems? Certainly. Is that the best use of your engineering capacity vs revenue generating work? That’s the math you’ve got to work out, and the role of your engineering management. If you’re a steady state business, then it probably makes sense to start performing aggressive cost optimization. If you’re a rocket ship, spend a bit more and optimize to grow the top line as fast as you can; you can optimize costs later.
With all that said, I’d agree AWS egress bandwidth is extortionate.
They had around 60 employees, which put the salary and expenses on employees close to 12M a year.
Assuming the "tens" were minimum at 20M, their revenue would be around ~$35M.
I spend in cloud to ensure we're not the bottleneck in the org- we pay our scientists, engineers, and labs far more money than we spend on cloud.
Also, previously (October 2022) from DHH: https://world.hey.com/dhh/why-we-re-leaving-the-cloud-654b47...
