Resource scheduling system and method based on multiobjective optimization

Abstract

The invention provides a resource scheduling system based on multiobjective optimization. The resource scheduling system comprises a monitoring assembly, a cross-cloud management assembly, an overallscheduling assembly and a user management assembly, the assemblies can be deployed on different physical machines or virtual machines respectively, and can be deployed on the same physical machine orthe same virtual machine simultaneously. All assemblies can support high-availability deploy according to the business scenario. The invention further provides a resource scheduling method based on the multiobjective optimization. The method comprises the steps of monitoring information collection, life cycle management and dynamic scheduling of all subsidiary cloud environments. Under a cross-cloud environment, unified management of resources is conducted, and the utilization rate of all resource pools is balanced; under the cross-cloud environment, the user authority management is unified, and the use of important resources is preferably guaranteed; under the cross-cloud environment, all resource information is automatically collected, the resources are dynamically scheduled with the change of the business, and the resources are reasonably reallocated; cross-cloud scheduling and dynamic optimization are supported; SLA is supported, and the resources are more reasonably allocated; through monitoring and statistics, the data is persistent, and a visual interface is provided.

Description

technical field [0001] The invention relates to a virtualized resource scheduling technology in the technical field of computer software, in particular to a resource scheduling system and method based on multi-objective optimization in a cross-cloud environment. Background technique [0002] Since Amazon launched the elastic computing cloud platform and achieved great success, the industry has set off a new business model (IaaS) that provides shared data center infrastructure based on virtualized elastic resource pools, internal integration (private cloud) shared resources or external leasing services. mode of public cloud) research whirlwind. [0003] Nowadays, in the process of building and reusing multiple heterogeneous private clouds (or park clouds, city clouds, etc.), the unified allocation and scheduling of resources has become a complicated issue. In actual use, resource allocation problems often arise, which are reflected in: [0004] 1) Uneven use of resource poo...

AI Technical Summary

Problems solved by technology

[0004] 1) Uneven use of resource pools, resulting in some resource pools being tight and some resource pools being too idle;

[0005] 2) Resource allocation is unreasonable, resulting in the SLA of some important users not being satisfied, and non-important users occupying high-quality resources instead;

[0006] 3) Serious waste of resources, resulting in memory and VCPU fragmentation, resulting in failure to be utilized;

[0007] 4) Resources cannot be shared, and there is a lack of a global and unified resource scheduling mechanism, resulting in the inability of resources to be shared and invoked among various clouds;

[0008] 5) Resources cannot be dynamically adjusted. The static allocation of resources may achieve a local optimum. However, as time goes by and business changes, dynamic adjustments cannot be made, and the greatest advantages of resource virtualization cannot be brought into play.

[0012] 1. Does not support resource cross-cloud scheduling;

[0013] 2. Does not support SLA (service level agreement);

[0014] 3. It does not support historical statistics of resource usage, and the information on changes in resource usage is not accurate enough to provide a basis for long-term decision-making

[0016] 1. Does not support resource cross-cloud scheduling;

[0017] 2. Does not support SLA (service level agreement)

Images