No, they always allowed 100GB, now they'll allow unlimited if you contact support. Of course, this seems more likely to be a result of European regulation than them being good citizens.
I had a pet project that I tried to turn into a SaaS and I did everything possible to keep it lean. A single Fargate container, two small EC2 instances to run my own Postgres cluster, one bastion. That's it.
Amazon is diabolical in the sense that the free tier expires over months, and by the time your bill starts creeping up with every single cycle, you've already built your infra around AWS. Thankfully I kept it as vanilla as I could, so it took my just a few days to bolt to Digital Ocean.
Best decision ever. A managed Postgres instance for $15 a month? That will get you one crappy micro EC2 box with Amazon. My bill went from $100 to $30! Imagine doing that at actual scale.
It was common reasoning that with very large scale going on-prem could be cost-effective, but you don't even have to go that far. It's a racket. It's not like there are no alternatives. Industry standard shmandart.
Using AWS as a VPS provider will likely always be expensive; but, as you learned, that level of abstraction makes it easy to move to an actual VPS provider.
To be fair on Amazon, the managed Progress instances on AWS are competitive with Digital Ocean when you move up the performance scale. The $12/month (or even $8 if you commit to a year) db.t4g.micro AWS instances are rock bottom in performance
9 comments
[ 3.2 ms ] story [ 35.7 ms ] threadhttps://techcrunch.com/2024/03/05/amazon-follows-google-in-a...
Amazon is diabolical in the sense that the free tier expires over months, and by the time your bill starts creeping up with every single cycle, you've already built your infra around AWS. Thankfully I kept it as vanilla as I could, so it took my just a few days to bolt to Digital Ocean.
Best decision ever. A managed Postgres instance for $15 a month? That will get you one crappy micro EC2 box with Amazon. My bill went from $100 to $30! Imagine doing that at actual scale.
It was common reasoning that with very large scale going on-prem could be cost-effective, but you don't even have to go that far. It's a racket. It's not like there are no alternatives. Industry standard shmandart.
For example: https://levelup.gitconnected.com/performance-aws-rds-postgre...
Thread from earlier in the week: https://news.ycombinator.com/item?id=39612607