Sometimes, it might be useful to change runtime behavior based on changes (e.g. disk failures) to the cache. For example, lets say I have 4 disks, and I know that I need 2 disks to function reasonably well. If I now lose two disks, I'd want ATS to take action (plugin would implement the action), such as shutting down, or going into no-accept mode etc.
Also, on startup, the same logic would apply. If I can't satisfy my cache requirements at startup, it should not let ATS pass the cache-ready lifecycle hook.