Disclaimer: I intended to ask this on the AWS Developer forum, but since I just created a forum account, I "might not be able to post messages for a few hours" and I'll surely be asleep by then and forget I was going to ask this tomorrow....
I've read over the CloudWatch pricing page, have searched the AWS forums and even made it to page 3 on google.... But I still don't understand CloudWatch pricing.
My confusion comes from a few things...
- My astonishment that I can run a server (be it a t3a.nano) for $3.50 a month, but it cost a $2.10 to monitor that server....
- AWS's statement that "Metric pricing is prorated by the hour and our calculation assumes customers are sending custom metrics at least once an hour" (that explanation is pretty "cloudy" to me.... sorry)
- AWS's example for Custom Metrics uses a completely relatable assumption that my application "runs on 51,000 Amazon EC2 instances....."
So here's the meat of my question. If I have a custom metric, let's assume I'm measuring max memory usage of a Python script that runs once per hour, for a duration of 60 seconds. So I send 730 measurements to Cloudwatch a month.....
Questions: (use general assumptions for any data I might have left out)
- I'm assuming that's a single metric?
- Is that really going to cost me $0.30 for 730 data points?
- What if I send 60 metrics per hour, for a total of 43,800 data points, still $0.30?