Looking to deploy MetricBeat and FileBeat on potentially 1000's of AWS Linux EC2 instances. Is there a published CPU overhead number for each of these agents? Obviously, developers are concerned about the agentry consuming too much of the CPU. Secondly, how much can it change based on configuration parameters within metricbeat.yml? Trying to get a worst case scenario for example if most of system module metrics are enabled: cpu, memory, processes, etc.
We don't have any published values as this depends on lots of different parameters.
For filebeat it heavily depends on how many log events you have per second and also the way you configured it. For metricbeat the period on how often events are sent is important, but also the rest of the config. And obviously the size of your EC2 instance.
I strongly recommend you to do some tests in a near production envrionment to get the numbers for your use case. Base on this we potentially can also recommend some config improvements etc. if needed. But that only works if you have all your numbers together.
ok, thank you for your input. We are in the process of defining some use case tests.
This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.