This is specifically for AWS EC2 monitoring but I'm also interested in the generic context of monitoring input/output.

How do I know what the network or disk i/o limit is for any server/service and whether it's approaching that limit (and if that's even a problem)? How do I know what's abnormal (where I set my alert thresholds) and should be investigated? For AWS I have disk read/write (bytes and operations) network in/out (bytes and packets) but no idea when I should be concerned about these metrics.

  • 1
    You can monitor actuals with CloudWatch. Limits are in documentation. – Tim Aug 4 at 4:42

Limits for network and disk i/o are listed in the documentation. Be aware that they vary depending on instance size and class, and that if you are using network i/o (like EBS), that can also run into issues.

What is abnormal and whether you should be concerned is entirely application dependent. For most web services, its very unlikely that you'd hit the limits on disk i/o, but if you are running a database, you might hit disk i/o limits, network limits, or both.

You'll definitely want to trend these metrics over a period of time and then set alerts accordingly; just remember to also factor in whether hitting a given limit actually presents some kind of business problem, or if it just means you are spinning up another instance; the former you are more likely to want to alert on than the latter.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.