How to compare the load and resources used by different applications?

Question

linux

Hello everyone,
There is a PaaS system that runs on ESXi, KVM and bare-metal.
An application is running inside, which is an api for applications inside LXC / docker containers.
You need to choose the optimal application for api and its configuration.
What to monitor:

• Resource consumption by api-server: RAM/CPU/ disk io/ number of read+write operations / response speed / dependence of consumption of these resources on the number of clients.
  
• API client resource consumption: RAM/CPU/ disk io/ number of read+write operations / request sending speed.
  

Answer 1

If the HTTP API works, you can throw a test plan in jmeter , the data can be uploaded to influxdb and beautifully visualized in Grafane . In Grafan, you can attach zabbix monitoring data . You can set up trashholds in the metrics, if they are exceeded, they will change color on the dashboard. There is a grafana fork with the ability to make alerts and notifications - raintank.

Answer 2

If in Docker, then install cAdvisor for CPU / RAM / DISK ( https://github.com/google/cadvisor ), if "by fast"
"number of read + write operations" / response speed / number of clients - read, write , speeds of what, what clients?