Workload rebalancing in heterogeneous resource environments

• US 10,673,981 B2
• Filed: 06/09/2017
• Issued: 06/02/2020
• Est. Priority Date: 06/09/2017
• Status: Active Grant

• Current Assignee: Nutanix Incorporated
• Original Assignee: Nutanix Incorporated
• Inventors: Sahay, Manjul, Chandra, Ramesh U.
• Primary Examiner(s): Dinh, Khanh Q
• Application Number:
  US15/619,180
• Publication Number:
  US 20200028935A1
• Time in Patent Office: 1,089 Days
• Field of Search: 709203, 709220, 709224, 709227
• CPC Class Codes: G06F 2009/45562 Creating, deleting, cloning virtual machine instances
• CPC Class Codes: G06F 2009/4557 Distribution of virtual machine instances; Migration and load balancing
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects
• CPC Class Codes: G06F 9/4856 resumption being on a different machine, e.g. task migration, virtual machine migration G06F9/5088 takes precedence
• CPC Class Codes: G06F 9/5083 Techniques for rebalancing the load in a distributed system
• CPC Class Codes: G06F 9/5088 involving task migration
• CPC Class Codes: H04L 67/1001 for accessing one among a plurality of replicated servers
• CPC Class Codes: H04L 67/1004 Server selection for load balancing
• CPC Class Codes: H04L 67/60 Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources admission control or resource allocation H04L47/70

Abstract

Techniques for rebalancing computing workloads between a set of on-premises resources of an on-premises system and a set of cloud computing resources of a cloud computing system. A method embodiment commences upon detecting a rebalancing trigger event. Responsive to the trigger event, a set of resource data corresponding to observations pertaining to the on-premises resources and resource data corresponding to observations pertaining the cloud resources is accessed. Based at least in part on the resource data, a plurality of candidate workload rebalancing plans are generated. The candidate workload rebalancing plans are evaluated against the likelihood of achieving quantitative objectives resulting from the rebalancing. A portion of the scheduling commands to carry out the workload rebalancing plans are sent to the cloud computing system, and another portion of the scheduling commands are sent to the on-premises computing system. The receiving computing systems carry out the scheduling commands.

Claims

1. A method comprising:

monitoring an on-premises computing system and a cloud computing system to detect a rebalancing trigger event, wherein the rebalancing trigger event is detected by applying a rebalancing trigger rule emitted by a machine learning module, a detector module, or a constraint satisfaction algorithm, or wherein the rebalancing trigger rule comprises a manually-specified trigger rule or a threshold-sensitive rule; and

dynamically rebalancing a workload in response to the rebalancing trigger event by;

generating a workload rebalancing plan based at least in part on resource usage data from the on-premises computing system and from the cloud computing system, the workload rebalancing plan specifying movement of the workload between the on-premises computing system and the cloud computing system; and

initiating the workload rebalancing plan by executing a command to move the workload between the on-premises computing system and the cloud computing system.

2. The method of claim 1, wherein the workload rebalancing plan comprises a workload attribute translation instruction.

3. The method of claim 2, wherein the workload attribute translation instruction comprises a binary image of a virtual machine or container, an installation specification pertaining to the virtual machine or container, a configuration specification pertaining to the virtual machine or container, an instruction to use a particular hypervisor type or container daemon, an instruction to create the virtual machine or container, or an instruction to register a migrated workload with the virtual machine or the container.

4. The method of claim 1, wherein the workload rebalancing plan comprises a scheduling command to carry out the workload rebalancing plan.

5. The method of claim 1, wherein the resource usage data corresponds to an on-premises resource of the on-premises computing system and to a cloud resource of the cloud computing system, resource usage data corresponding to the on-premises resource being received using a first application programming interface (API) and resource usage data corresponding to the cloud resource being received using a second API.

6. The method of claim 1, wherein generating the workload rebalancing plan is also based on resource environment data, and resource rule data, and the resource usage data, the resource environment data and the resource rule data is represented using a structured data interchange format.

7. The method of claim 1, wherein the rebalancing trigger rule is associated with a resource usage, a resource configuration, a resource geography, a resource functionality, a workload, a policy, a rebalancing schedule, or a resource usage cost.

8. The method of claim 1, wherein the rebalancing trigger rule is emitted by the machine learning module.

9. The method of claim 1, wherein the workload rebalancing plan is generated using a planning algorithm selected from a plurality of algorithms based at least in part on the rebalancing trigger event.

10. The method of claim 9, wherein the planning algorithm is a multi-objective optimization algorithm or a logistic regression algorithm.

11. A non-transitory computer readable medium having stored thereon a sequence of instructions which, when executed by a processor performs a set of acts comprising:

monitoring an on-premises computing system and a cloud computing system to detect a rebalancing trigger event, wherein the rebalancing trigger event is detected by applying a rebalancing trigger rule emitted by a machine learning module, a detector module, or a constraint satisfaction algorithm, or wherein the rebalancing trigger rule comprises a manually-specified trigger rule or a threshold-sensitive rule; and

dynamically rebalancing a workload in response to the rebalancing trigger event by;

generating a workload rebalancing plan based at least in part on resource usage data from the on-premises computing system and from the cloud computing system, the workload rebalancing plan specifying movement of the workload between the on-premises computing system and the cloud computing system; and

initiating the workload rebalancing plan by executing a command to move the workload between the on-premises computing system and the cloud computing system.

12. The computer readable medium of claim 11, wherein the workload rebalancing plan comprises a workload attribute translation instruction.

13. The computer readable medium of claim 12, wherein the workload attribute translation instruction comprises a binary image of a virtual machine or container, an installation specification pertaining to the virtual machine or container, a configuration specification pertaining to the virtual machine or container, an instruction to use a particular hypervisor type or container daemon, an instruction to create the virtual machine or container, or an instruction to register a migrated workload with the virtual machine or the container.

14. The computer readable medium of claim 11, wherein the workload rebalancing plan comprises a scheduling command to carry out the workload rebalancing plan.

15. The computer readable medium of claim 11, wherein the resource usage data corresponds to an on-premises resource of the on-premises computing system and to a cloud resource of the cloud computing system, resource usage data corresponding to the on-premises resource being received using a first application programming interface (API) and resource usage data corresponding to the cloud resource being received using a second API.

16. The computer readable medium of claim 11, wherein generating the workload rebalancing plan is also based on resource environment data, and resource rule data, and the resource usage data, the resource environment data and the resource rule data is represented using a structured data interchange format.

17. The computer readable medium of claim 11, wherein the rebalancing trigger rule is emitted by the machine learning module.

18. A system comprising:

a storage medium having stored thereon a sequence of instructions; and

a processor that executes the sequence of instructions to perform a set of acts comprising;

monitoring an on-premises computing system and a cloud computing system to detect a rebalancing trigger event, wherein the rebalancing trigger event is detected by applying a rebalancing trigger rule emitted by a machine learning module, a detector module, or a constraint satisfaction algorithm, or wherein the rebalancing trigger rule comprises a manually-specified trigger rule or a threshold-sensitive rule; and

dynamically rebalancing a workload in response to the rebalancing trigger event by;

generating a workload rebalancing plan based at least in part on resource usage data from the on-premises computing system and from the cloud computing system, the workload rebalancing plan specifying movement of the workload between the on-premises computing system and the cloud computing system; and

initiating the workload rebalancing plan by executing a command to move the workload between the on-premises computing system and the cloud computing system.

19. The system of claim 18, wherein the workload rebalancing plan comprises a workload attribute translation instruction.

20. The system of claim 19, wherein the workload attribute translation instruction comprises a binary image of a virtual machine or container, an installation specification pertaining to the virtual machine or container, a configuration specification pertaining to the virtual machine or container, an instruction to use a particular hypervisor type or container daemon, an instruction to create the virtual machine or container, or an instruction to register a migrated workload with the virtual machine or the container.

21. The system of claim 18, wherein generating the workload rebalancing plan is also based on resource environment data, and resource rule data, and the resource usage data, the resource environment data and the resource rule data is represented using a structured data interchange format.

22. The system of claim 18, wherein the rebalancing trigger rule is emitted by the machine learning module.