The instances property in the conf.yaml let you customize the way the agent collects the metrics, with that being said, you can manually load-balancing the API requests performed by the agent splitting the agent load through different instances and gathering datapoints as per your requirements.

Let's assume you are interested in getting updated data of your CloudHub and On-Prem apps behavior, through the Development: Optimizations dashboards, you'd like to monitor the CPU, memory and networking in a constant rate, but you won't like to push the agent to get at the same rate the metrics of your resource allocation. You can easily set up the following configuration to fulfill the depicted requirement.

  
instances:
   - min_collection_interval: 172800
      threads: 32
      api_filter:
         - access_management
   - min_collection_interval: 10
      threads: 32
      api_filter:
         - cloudhub
  

With the above example, you are running two isolated executions of our metric reader program, if you have a scenario where customization of API calls is needed you can refer to the configuration section to locate the configuration YAML file and set your configuration properly. Once done, you only need to restart the agent to load the new configuration and that's it you will now have a customized agent.