The VM agent (roxagent) supports a "daemon mode" which is expected to be used by customers to enable periodic reporting. Currently it sends reports at fixed intervals of (by default) 5 minutes. This presents several challenges:

• The default interval is too short for ACS to be able to handle reports from the desired amount of VMs (see the VM scale test for details).
• Enabling the agent in many VMs at the same time (e.g. via Ansible) results in peaks of reports hitting central at the same time and either crashing it or being dropped by the rate limiter (when merged).

These concerns should be minimized by adjusting the agent's behavior, e.g. increasing the default interval and adding a random delay to initial sending.