It's not always clear what would be the final state of the cluster after applying the suggested changes (obtained from /autoscaling/suggestions), especially on a large and busy cluster where several autoscaling rules have to be considered.
This issue proposes to use the simulation framework for simulating the effects of the suggestions.
First, the simulator would be initialized from the current state of a real cluster. Then it would run several rounds of simulated execution of suggestions until there were either no more suggestions (the cluster would be perfectly balanced) or the iteration count limit was reached.
This simulation could be executed using either the deployed autoscaling config or one provided by the user, which would make it easier to test the effects of various configurations on the cluster layout.
Support for this functionality would be integrated into the existing SolrCLI autoscaling tool.