Predictive auto-scaler for a hierarchical computing infrastructure
A cloud configuration, including public and private clouds, is organized hierarchically, with a top level and any number of lower levels. A parent level cloud receives resource workload metrics from respective child level cloud(s), makes predictions, based in part on the metrics, as to future resource needs at the child level. The parent level sets up runtime-modifiable business rules and scaling plans based on the predictions. The parent level cloud sends the scaling plans to respective child level(s). The parent level automatically triggers a scaling plan at a child level, if conditions written into the business rules are satisfied. Resources are scaled upward or downward automatically as needed to maintain optimal resource usage.
1. A computer-implemented method for automatic resource scaling in a multi-level computing platform, the method comprising:
receiving a first workload metric for a first resource of a multi-level computing platform;
predicting, based on the first workload metric, a scaling action for the first resource;
inserting, based on the predicted scaling action for the first resource, a predictive metric generated with respect to a future point in time into an existing runtime modifiable rule set for triggering an implementation of a scaling plan for the first resource, wherein the predictive metric modifies one or more rules in the existing runtime modifiable rule set for triggering the implementation of the scaling plan;
generating the scaling plan, based on a combination of the predicted scaling action and the runtime-modifiable rule set;
sending the scaling plan to a level of the multi-level computing platform associated with the first resource; and
triggering, based on the runtime-modifiable rule set, implementation of the scaling plan.
2. The method of claim 1 , wherein the multi-level computing platform comprises a hybrid cloud platform comprising a plurality of cloud levels structured in a hierarchical manner, where each cloud level is selected from the group consisting of: a private cloud computing platform and a public cloud computing platform.
3. The method of claim 2 , further comprising:
receiving a second workload metric with respect to a second resource of the multi-level computing platform;
generating a second predicted scaling decision based on the second workload metric;
generating a run-time modifiable auto-scaling rule set local to a respective level of the multi-level computing platform;
modifying an application-level configuration file, and an application-level setting;
generating a local configuration change plan; and
deploying and managing a service running on the respective level of the multi-level computing platform, based on the local configuration change plan.
4. The method of claim 1 wherein the first workload metric is selected from the group consisting of: processor utilization, memory usage, storage usage, network bandwidth usage, arrival rate, interarrival time, response time, throughput, and service load pattern.
5. The method of claim 1 wherein the first resource is selected from the group consisting of: memory, storage, network bandwidth, and processor utilization.
6. The method of claim 1 wherein triggering implementation of the scaling plan further comprises:
determining the first workload metric exceeds a threshold value for a pre-defined period of time; and
in response to determining the first workload metric exceeds the threshold value for the pre-defined period of time, deploying the scaling plan.
7. The method of claim 1 wherein triggering implementation of the scaling plan further comprises:
determining a threshold value exceeds the first workload metric for a pre-defined period of time; and
in response to determining the threshold value exceeds the first workload metric for the pre-defined period of time, deploying the scaling plan.
8. A computer program product for automatic resource scaling in a multi-level computing platform, the computer program product comprising:
one or more non-transitory computer readable storage media, and program instructions collectively stored on the one or more non-transitory computer readable storage media, the program instructions comprising instructions programmed to perform:
receiving a first workload metric for a first resource of a multi-level computing platform;
predicting, based on the first workload metric, a scaling action for the first resource;
inserting, based on the predicted scaling action for the first resource, a predictive metric generated with respect to a future point in time into an existing runtime modifiable rule set for triggering an implementation of a scaling plan for the first resource, wherein the predictive metric modifies one or more rules in the existing runtime modifiable rule set for triggering the implementation of the scaling plan;
generating the scaling plan, based on a combination of the predicted scaling action and the runtime-modifiable rule set;
sending the scaling plan to a level of the multi-level computing platform associated with the first resource; and
triggering, based on the runtime-modifiable rule set, implementation of the scaling plan.
9. The computer program product of claim 8 , wherein the multi-level computing platform comprises a hybrid cloud platform comprising a plurality of cloud levels structured in a hierarchical manner, where each cloud level is selected from the group consisting of: a private cloud computing platform and a public cloud computing platform.
10. The computer program product of claim 9 , further comprising instructions programmed to perform:
receiving a second workload metric with respect to a second resource of the multi-level computing platform;
generating a second predicted scaling decision based on the second workload metric;
generating a run-time modifiable auto-scaling rule set local to a respective level of the multi-level computing platform;
modifying an application-level configuration file, and an application-level setting;
generating a local configuration change plan; and
deploying and managing a service running on the respective level of the multi-level computing platform, based on the local configuration change plan.
11. The computer program product of claim 8 wherein the first workload metric is selected from the group consisting of: processor utilization, memory usage, storage usage, network bandwidth usage, arrival rate, interarrival time, response time, throughput, and service load pattern.
12. The computer program product of claim 8 wherein the first resource is selected from the group consisting of: memory, storage, network bandwidth, and processor utilization.
13. The computer program product of claim 8 wherein instructions programmed to perform triggering implementation of the scaling plan further comprises instructions programmed to perform:
determining the first workload metric exceeds a threshold value for a pre-defined period of time; and
in response to determining the first workload metric exceeds the threshold value for the pre-defined period of time, deploying the scaling plan.
14. The computer program product of claim 8 wherein instructions programmed to perform triggering implementation of the scaling plan further comprises instructions programmed to perform:
determining a threshold value exceeds the first workload metric for a pre-defined period of time; and
in response to determining the threshold value exceeds the first workload metric for the pre-defined period of time, deploying the scaling plan.
15. A computer system for automatic resource scaling in a multi-level computing platform, the computer system comprising:
one or more computer processors; and
one or more computer readable storage media;
wherein:
the one or more computer processors are structured, located, connected and/or programmed to run program instructions stored on the one or more computer readable storage media; and
the program instructions include instructions programmed to perform:
receiving a first workload metric for a first resource of a multi-level computing platform;
predicting, based on the first workload metric, a scaling action for the first resource;
inserting, based on the predicted scaling action for the first resource, a predictive metric generated with respect to a future point in time into an existing runtime modifiable rule set for triggering an implementation of a scaling plan for the first resource, wherein the predictive metric modifies one or more rules in the existing runtime modifiable rule set for triggering the implementation of the scaling plan;
generating the scaling plan, based on a combination of the predicted scaling action and the runtime-modifiable rule set;
sending the scaling plan to a level of the multi-level computing platform associated with the first resource; and
triggering, based on the runtime-modifiable rule set, implementation of the scaling plan.
16. The computer system of claim 15 , wherein the multi-level computing platform comprises a hybrid cloud platform comprising a plurality of cloud levels structured in a hierarchical manner, where each cloud level is selected from the group consisting of: a private cloud computing platform and a public cloud computing platform.
17. The computer system of claim 16 , further comprising instructions programmed to perform:
receiving a second workload metric with respect to a second resource of the multi-level computing platform;
generating a second predicted scaling decision based on the second workload metric;
generating a run-time modifiable auto-scaling rule set local to a respective level of the multi-level computing platform;
modifying an application-level configuration file, and an application-level setting;
generating a local configuration change plan; and
deploying and managing a service running on the respective level of the multi-level computing platform, based on the local configuration change plan.
18. The computer system of claim 15 wherein the first workload metric is selected from the group consisting of: processor utilization, memory usage, storage usage, network bandwidth usage, arrival rate, interarrival time, response time, throughput, and service load pattern.
19. The computer system of claim 15 wherein the first resource is selected from the group consisting of: memory, storage, network bandwidth, and processor utilization.
20. The computer system of claim 15 wherein instructions programmed to perform triggering implementation of the scaling plan further comprises instructions programmed to perform:
determining the first workload metric exceeds a threshold value for a pre-defined period of time; and
in response to determining the first workload metric exceeds the threshold value for the pre-defined period of time, deploying the scaling plan.