Hi
This is my first post at IH. I have read a lot of grievances about using AWS services (both on job & off) They have some generous credit-program me. I have been using it for a couple of years to experiment with their services.
So, I want to ask the community what is the biggest pain-point while using AWS?
Also what is the one feature (if introduced) would enable you to adopt it more freely.
Waiting for the monthly Cost Explorer refresh is the wrong
feedback loop for this turn on AWS Cost Anomaly Detection
instead, it's free and runs near-real-time ML monitors per
service or account, so a resource spun up in the wrong region
shows up as a spike within a day instead of 30.
Pair that with an SCP or tag policy that restricts which regions
your account can actually provision in that turns "wrong region"
from a billing surprise into something that just fails at creation
time.
Cost Explorer is fine for trend analysis after the fact, but it
was never built to be your alarm system.
issacguerrero67-web/cloud-cost-toolkit on GitHub.
For me, thier unpredictable pricing. Even with reserved instances for 3 years, you will be billed absurd amounts.
I am not saying they have hidden charges, but you can't predict how much you are going to spend on AWS.
In an instance my RDS bill was in $8k for the month, just due to high IOPS.
@tusharb - IOPS can cause cost-overruns! did you try using the pricing calculator to estimate the running costs?
We had a similar situation when someone had set up resources in different region and we ran up a bill of $3k bill in a single week!
After that incident I took all the precautions. But since now I moved to DO, I don't think I will have to worry about the same.
your wrong-region story is the one that gets me — that's not runaway usage, that's a config mistake that silently compounds for days. The $3K was probably burning for 72+ hours before anyone caught it. Has anyone built a reliable way to catch "this resource is in the wrong place / shouldn't exist" type anomalies before they compound? I keep hearing "set a billing alert" but by the time a monthly alert fires the damage is already done. Curious if people have found anything that detects this kind of structural mistake faster.
that's interesting! So with DO do you use their managed DB instances ?
Yes, that's the only way to go. Managing the DB is a tedious job on a medium scale.
I've not used AWS but their pricing is so ridicuously complicated. I was asked to try and find an estimate cost for a new project at work and it was nearly impossible to try and come up with something as it was just pure guess work - thankfully we didn't use AWS in the end and went somewhere else that had more easy to understand pricing.
you know they have price calculator?
The right decision, I guess. Even we have moved from AWS to Digital Ocean this month after using them from past 4 years.
Yea, we went with Digital Ocean in the end as well, it was either DO or Linode that we chose between in the end.
Thats interesting. I guess if your using bare-bone servers. Its cheaper to use DO/Linode. They have the cheapest servers in the market.
I am curious what issues did you face with the pricing calculator?
https://calculator.aws/
The fact that they need a pricing calculator proves their pricing is complicated 😊 I'm not sure they had that specific pricing calculator at the time, but one of the problems was working out what size of VM we needed as they had so many different versions, plus had to calculate things like expected mb in and out which we had no idea for this project so it was pure guess work. Potentially it would have been a lot so would ha e worked out quite expensive
This comment was deleted 5 years ago
@Arrigo - when you say "configurations", I believe you mean IAM roles & permissions.
I have not used Netlify, but does it have better configuration management?
This comment was deleted 5 years ago