Using a generative model to facilitate simulation of potential policies for an infrastructure as a service system

1. A method for evaluating at least one potential policy for an infrastructure as a service (IaaS) system, comprising:

creating at least one generative model based on historical data corresponding to the IaaS system, wherein creating the at least one generative model comprises:

determining causal relationships between variables related to the IaaS system and how the causal relationships change over time; and

determining joint probabilities between the variables for which a causal relationship exists;

determining a predicted workload for the IaaS system based on the at least one generative model corresponding to the IaaS system;

simulating the at least one potential policy for the IaaS system based on the predicted workload, thereby producing one or more simulation metrics that indicate effects of the at least one potential policy; and

optimizing performance of the IaaS system based on the one or more simulation metrics.

2. The method of claim 1 , wherein the at least one generative model comprises a probabilistic graphical model.

3. The method of claim 1 , wherein:

the predicted workload comprises a plurality of tasks that are predicted to be performed by the IaaS system; and

simulating the at least one potential policy comprises implementing the at least one potential policy in a virtual IaaS system and causing the virtual IaaS system to perform the plurality of tasks with the at least one potential policy in effect.

4. The method of claim 1 , wherein optimizing the performance of the IaaS system comprises:

determining whether the one or more simulation metrics satisfy at least one implementation condition;

providing a first recommendation to implement the at least one potential policy if the one or more simulation metrics satisfy the at least one implementation condition; and

providing a second recommendation to not implement the at least one potential policy if the one or more simulation metrics do not satisfy the at least one implementation condition.

5. The method of claim 1 , wherein optimizing the performance of the IaaS system comprises selecting a first potential policy for the IaaS system over a second potential policy for the IaaS system because first simulation metrics corresponding to the first potential policy are more favorable than second simulation metrics corresponding to the second potential policy.

6. A system for evaluating at least one potential policy for an infrastructure as a service (IaaS) system, comprising:

one or more processors; and

memory comprising instructions that are executable by the one or more processors to perform operations comprising:

determining a predicted workload for the IaaS system based on at least one generative model corresponding to the IaaS system;

simulating the at least one potential policy for the IaaS system based on the predicted workload, thereby producing one or more simulation metrics that indicate effects of the at least one potential policy; and

optimizing performance of the IaaS system based on the one or more simulation metrics, wherein optimizing the performance of the IaaS system comprises:

determining whether the one or more simulation metrics satisfy at least one implementation condition;

providing a first recommendation to implement the at least one potential policy if the one or more simulation metrics satisfy the at least one implementation condition; and

providing a second recommendation to not implement the at least one potential policy if the one or more simulation metrics do not satisfy the at least one implementation condition.

7. The system of claim 6 , wherein the operations further comprise creating the at least one generative model based on historical data corresponding to the IaaS system.

8. The system of claim 7 , wherein creating the at least one generative model comprises:

determining causal relationships between variables related to the IaaS system and how the causal relationships change over time; and

determining joint probabilities between the variables for which a causal relationship exists.

9. The system of claim 6 , wherein the at least one generative model comprises a probabilistic graphical model.

10. The system of claim 6 , wherein:

the predicted workload comprises a plurality of tasks that are predicted to be performed by the IaaS system; and

simulating the at least one potential policy comprises implementing the at least one potential policy in a virtual IaaS system and causing the virtual IaaS system to perform the plurality of tasks with the at least one potential policy in effect.

11. The system of claim 6 , wherein optimizing the performance of the IaaS system comprises selecting a first potential policy for the IaaS system over a second potential policy for the IaaS system because first simulation metrics corresponding to the first potential policy are more favorable than second simulation metrics corresponding to the second potential policy.

12. A non-transitory computer-readable medium having computer-executable instructions stored thereon that, when executed, cause one or more processors to perform operations comprising:

determining a predicted workload for an infrastructure as a service (IaaS) system based on at least one generative model corresponding to the IaaS system;

simulating at least one potential policy for the IaaS system based on the predicted workload, thereby producing one or more simulation metrics that indicate effects of the at least one potential policy; and

optimizing performance of the IaaS system based on the one or more simulation metrics, wherein optimizing the performance of the IaaS system comprises selecting a first potential policy for the IaaS system over a second potential policy for the IaaS system because first simulation metrics corresponding to the first potential policy are more favorable than second simulation metrics corresponding to the second potential policy.

13. The non-transitory computer-readable medium of claim 12 , wherein the operations further comprise creating the at least one generative model based on historical data corresponding to the IaaS system.

14. The non-transitory computer-readable medium of claim 13 , wherein creating the at least one generative model comprises:

determining causal relationships between variables related to the IaaS system and how the causal relationships change over time; and

determining joint probabilities between the variables for which a causal relationship exists.

15. The non-transitory computer-readable medium of claim 12 , wherein:

the predicted workload comprises a plurality of tasks that are predicted to be performed by the IaaS system; and

simulating the at least one potential policy comprises implementing the at least one potential policy in a virtual IaaS system and causing the virtual IaaS system to perform the plurality of tasks with the at least one potential policy in effect.

16. The non-transitory computer-readable medium of claim 12 , wherein optimizing the performance of the IaaS system comprises:

determining whether the one or more simulation metrics satisfy at least one implementation condition;

providing a first recommendation to implement the at least one potential policy if the one or more simulation metrics satisfy the at least one implementation condition; and

providing a second recommendation to not implement the at least one potential policy if the one or more simulation metrics do not satisfy the at least one implementation condition.