Which modules of OpenStack are used to perform actions to achieve autoscaling and fault-tolerance?
I know that Heat (with Ceilometer) achieves this through given templates.

My doubt is if even Nova does automatically something like:

 "O gosh, that machine is exploding!  
 The workload is to high!  
 Let's launch another VM to share the load!"

or not.

I ask you this 2 main reasons:

• I'm trying to understand if OpenStack has already a built-in MAPE cycle (Monitoring Analysis Planning Execution), or if it relies only on Heat templates;
• I'm trying to understand if OpenStack isolates all the "adaptive" and "automatic" part into Heat, leaving Nova only as e mere executor of actions (launching, stopping and so on).

I hope that now the question is clearer.
Thank you