Dynamic resource optimization system for virtualized cloud platform

Abstract

The invention relates to a dynamic resource optimization system for a virtualized cloud platform, and belongs to the field of cloud computing. The system comprises an interface and interaction API service, a decision maker and a strategy application. The interface and interaction API service is used for providing a RESTAPI interface for the outside, so that an administrator can interact with the system through a CLI and a webUI; the decision maker comprises a strategy evaluation module and an optimization action planning module, stores target information, strategy information and data source information at the same time, and is used for generating an action plan; and the strategy application component executes each optimization action in all the workflows in the generated action plan in turn or in parallel, and performs optimization adjustment on the resources. According to the method, a cloud platform administrator can be helped to automatically identify and implement system optimization measures, the overall efficiency and cost effectiveness of cloud environment resource use of the cloud platform are improved, and the method is a very valuable automatic service.

Description

Technical Field

[0001] The present invention belongs to the field of cloud computing, and in particular relates to a dynamic resource optimization system for a virtualized cloud platform. Background Art

[0002] Kubernetes autoscaling is one of the features provided by the Kubernetes platform, which is used to dynamically adjust the number of containers based on application load. This enables Kubernetes to automatically manage resources in response to changes in workloads. One of its shortcomings is its limited scope. Kubernetes autoscaling is mainly aimed at containerized applications, not the entire infrastructure layer. Therefore, it may not provide comprehensive optimization capabilities for non-containerized virtual machines or physical servers. Secondly, there are restrictions on its resource types. Kubernetes autoscaling is mainly used to manage application-level resources, such as Pods and Deployments, rather than underlying computing, storage, and network resources. This means that Kubernetes autoscaling may not be the best choice for situations where deeper infrastructure optimization is required.

[0003] CloudWatch Auto Scaling is a service provided by Amazon Web Services that can automatically scale resources based on demand. It can automatically increase or decrease the number of instances based on monitoring indicators to cope with changing workloads. Its disadvantage is its proprietary nature. CloudWatch Auto Scaling only works on the AWS platform, which means it cannot be used across different cloud environments, its integration capabilities often do not meet requirements, and it is not suitable for private cloud environments.

[0004] The above technologies all have their advantages, but they also have certain limitations, especially when it comes to the comprehensive management and optimization of underlying infrastructure resources. This system aims to provide a comprehensive approach that is infrastructure-oriented, easier to customize, and more suitable for private clouds to optimize resources in the entire virtualized cloud platform deployment.

[0005] As time goes by, virtual machines are constantly created, destroyed, modified, and migrated, and the resource usage of data centers is often unbalanced or even extremely unbalanced. It often happens that some servers, storage and other resources are under heavy load for a long time, while other servers and storage resources are always running under light load. The unbalanced load not only wastes some resources, but also causes performance degradation of some servers, resulting in performance bottlenecks when the virtual machines above are running.

[0006] At the same time, unnecessary resource usage also leads to increased energy and hardware costs, and even causes unreasonable budget increases for capacity expansion.

[0007] Sometimes, for the purpose of energy conservation and emission reduction, when the energy consumption of a data center is high and does not meet safety or environmental protection requirements, it is required to shut down certain virtual machines or servers.

[0008] Sometimes the server needs to be shut down for a short period of time due to hardware failure, software upgrade, or hardware replacement. This may result in a longer recovery time when the physical server fails, causing a greater impact on the business.

[0009] In addition, frequent manual migration operations for certain purposes may cause system instability and service interruption.

[0010] Therefore, cloud platform administrators hope to automatically improve resource utilization efficiency, reduce data center operating costs, and provide stability through some optimization measures.

[0011] The system provides a complete optimization loop chain: from metric receiver, to optimization processor and action plan application. Its goal is to provide a framework that can achieve a wide range of cloud optimization goals, including reducing data center operating costs, improving system performance through intelligent virtual machine migration, improving energy efficiency, etc. Summary of the invention

[0012] 1. Technical issues to be resolved

[0013] The technical problem to be solved by the present invention is how to provide a dynamic resource optimization system for a virtualized cloud platform to help cloud service providers and enterprise managers optimize the performance and resource utilization of their cloud infrastructure to improve overall efficiency and reduce costs, and more effectively manage their cloud infrastructure.

[0014] (II) Technical solution

[0015] In order to solve the above technical problems, the present invention proposes a dynamic resource optimization system for a virtualized cloud platform, the system comprising: an interface and interactive API service, a decision maker and a policy application;

[0016] Interface and interactive API services are used to provide REST API interfaces to the outside world, allowing administrators to interact with the system through CLI and webUI;

[0017] The decision maker includes a strategy evaluation module and an optimization action planning module, and stores target information, strategy information, and data source information. The decision maker forms a data source based on the real-time and historical data collected by the monitoring system, selects relevant strategies according to the specified resource optimization goals, and after completing the strategy evaluation preparation, performs strategy evaluation through the strategy evaluation module to calculate a preliminary non-time-sequential solution. The solution then generates a final time-sequential action plan through the optimization action planning module. The action plan generated by the decision maker includes several workflows, and each workflow contains several optimization actions.

[0018] The strategy application component is the execution component in the system; the strategy application component executes each optimization action in all workflows in the generated action plan in turn or in parallel to optimize and adjust the resources.

[0019] (III) Beneficial effects

[0020] The present invention proposes a dynamic resource optimization system for a virtualized cloud platform. The system of the present invention aims to provide automation and continuous optimization capabilities for the use of resources in a virtualized cloud platform environment. It has the following beneficial effects:

[0021] The system is equipped with an energy-saving and emission-reduction strategy to identify and merge virtual machine instances scattered across multiple hosts, merging them onto fewer hosts to achieve the goal of energy saving and emission reduction.

[0022] The system is equipped with performance improvement strategies to help capacity planning. It can help users plan capacity based on historical data and forecast analysis to avoid insufficient or wasted resources.

[0023] The system is equipped with a load balancing strategy to dynamically migrate virtual machine instances to achieve load balancing and resource optimization. By automatically optimizing resource allocation, it reduces idle resources and improves resource utilization. It ensures that server loads are roughly equal and improves overall system performance.

[0024] The system is equipped with a fault migration strategy, which predicts potential fault risks by monitoring abnormal behaviors and early warning information in the system, and takes measures in advance to avoid the impact of system faults on the business.

[0025] The system is equipped with a node maintenance strategy to monitor the status of the host in real time. In the event of failure or maintenance, the system automatically migrates the affected virtual machines according to predefined processing rules, speeding up failure recovery and reducing business interruption time.

[0026] In general, this system can help cloud platform administrators automatically identify and implement system optimization measures, improve the overall efficiency and cost-effectiveness of cloud platform cloud environment resource utilization, and is a very valuable automated service. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a system structure diagram of the present invention;

[0028] Figure 2 A flow chart for evaluating the strategy of the present invention;

[0029] Figure 3 The following is a flow chart of the strategy application of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with the drawings and examples.

[0031] The system belongs to the field of cloud computing, specifically involving the optimization, scheduling and capacity planning of cloud resources. It aims to help cloud service providers and enterprise managers optimize the performance and resource utilization of their cloud infrastructure to improve overall efficiency and reduce costs, and manage their cloud infrastructure more effectively.

[0032] Figure 1 It is a system structure diagram.

[0033] The present invention provides a dynamic resource optimization system for a virtualized cloud platform, which mainly includes three components: an interface and interactive API service, a decision maker, and a policy application;

[0034] The interface and interactive API service is used to provide a REST API interface to the outside world, allowing administrators to interact with the system through CLI and webUI.

[0035] The decision maker includes a strategy evaluation module and an optimization action planning module, and stores target information, strategy information, and data source information.

[0036] The policy application component is the execution component in the system.

[0037] Figure 2 It is the decision maker flow chart.

[0038] The decision maker process is as follows: the decision maker forms a data source based on the real-time and historical data collected by the monitoring system, selects relevant strategies according to the specified resource optimization goals, and after completing the strategy evaluation preparation, performs strategy evaluation through the strategy evaluation module to calculate a preliminary non-time-sequential solution. The solution then generates a final time-sequential action plan through the optimization action planning module. Among them, the action plan generated by the decision maker includes several workflows, and each workflow contains several executable optimization actions.

[0039] Figure 3 It is a flow chart of the strategy application component.

[0040] The process of the strategy application component is as follows: the strategy application component executes each optimization action in all workflows in the generated action plan in turn or in parallel to optimize and adjust the resources.

[0041] Embodiment 1:

[0042] In the cloud platform, when a user proposes a resource optimization goal to the system, the system first collects real-time and historical data from various components and services of the cloud platform. These data include the status and performance indicators of computing nodes, cloud platform services, virtual machines, storage and network devices. Then, the system uses pre-defined strategies to perform policy evaluation on the collected data. Each strategy has a specific goal, such as improving resource utilization or reducing costs. The system will generate a series of optimization suggestions based on these strategies, and evaluate and score these suggestions to select the best solution.

[0043] Finally, in the policy application phase, the system applies the best policy obtained during the policy evaluation process to the actual environment. This may include migrating virtual machines, adjusting resource allocation, or shutting down unused resources. The system also continuously monitors the results of the optimization to ensure that the optimization goals are achieved.

[0044] The main steps include:

[0045] S1. Data Collection

[0046] S101. Collect operation information of servers, cloud platform services, virtual machines, storage, etc. through the monitoring system.

[0047] S2. Strategy evaluation preparation includes the following steps:

[0048] S201. The administrator selects optimization goals on the interface, including energy saving and emission reduction goals, performance improvement goals, load balancing goals, fault migration goals, node maintenance goals, etc.

[0049] S202, the administrator selects a strategy on the interface, including energy saving and emission reduction strategy, performance improvement strategy, load balancing strategy, fault migration strategy, node maintenance strategy, etc. Configure a default strategy for the target, and each target can choose to use one or more strategies.

[0050] Energy-saving and emission-reduction strategies are used to identify and merge virtual machine instances scattered across multiple hosts and merge them onto fewer hosts to achieve energy-saving and emission-reduction goals.

[0051] Performance improvement strategies are used for capacity planning. They can help users plan capacity based on historical data and forecast analysis to avoid insufficient or wasted resources.

[0052] Load balancing strategy is used to dynamically migrate virtual machine instances to achieve load balancing and resource optimization. It automatically optimizes resource allocation, reduces idle resources, and improves resource utilization. It ensures that server loads are roughly equal and improves overall system performance.

[0053] Fault migration strategy is used to predict potential fault risks by monitoring abnormal behaviors and warning information in the system, and take measures in advance to avoid the impact of system failures on the business.

[0054] Node maintenance policy is used to monitor the status of the host in real time. In the event of failure or maintenance, the affected virtual machines are automatically migrated according to predefined processing rules to accelerate failure recovery and reduce business interruption time.

[0055] S203. The administrator creates a policy evaluation template, and can specify optimization targets and policies, or only optimize targets without specifying optimization policies, and use the configured default policies.

[0056] S204: The administrator sends a policy assessment template creation instruction to the interactive API service, and writes the created object and initial status into the database.

[0057] S205: The administrator creates and starts a policy evaluation object. The policy evaluation object includes an optimization target and can be created using an existing evaluation template.

[0058] When creating an assessment object, two types are supported: single assessment mode and continuous assessment mode. The single assessment mode is started once and will eventually provide an action plan. The continuous assessment mode will continuously create new action plans at given time intervals.

[0059] S206. The administrator sends the assessment creation instruction to the interactive API service, and the created object and initial status are written into the database.

[0060] S207: Start policy evaluation and send the message to the message queue bus.

[0061] S208: The decision maker receives the message and prepares to perform evaluation.

[0062] S209, strategy selection, directly using the specified strategy, or automatically selecting one or more strategies when only the target is specified.

[0063] S3. Execution strategy evaluation

[0064] S301. Check whether the policy meets the dependency conditions.

[0065] S302, strategy calculation starts: first, cluster data models are obtained. These models are processed and standardized input data.

[0066] S303 , the strategy calculation continues, and the required data is acquired cyclically until the strategy calculation requirements are met.

[0067] S304: The strategy calculation continues, and a list of actions required to meet the solution of the optimization goal is calculated.

[0068] S305 . The strategy calculation continues to calculate various efficacy scores of the solution.

[0069] S306 , the strategy calculation continues, and the total efficacy score of the solution is calculated based on various efficacy scores.

[0070] S307. Return the solution with unsorted actions.

[0071] S4. Action Planning

[0072] S401, the action planning module receives the calculated solution, starts planning the solution, and generates a time-ordered action plan.

[0073] S5. Strategy Application

[0074] S501, automatically executing the recommended action plan according to the configuration schedule, executing according to the custom event trigger, or manually executing. When there are multiple strategies, if executing manually, you can refer to the efficacy score of the solution calculated by each strategy to select which action plan to adopt.

[0075] S502: Send the execution plan instruction to the computing API, storage API, and bare metal management API.

[0076] S503: Write the relevant status into the database.

[0077] S504: Send to the message queue bus.

[0078] S505: After receiving the message, the strategy application module starts executing the optimization action plan.

[0079] S506: Circularly execute each optimization action of all workflows in the action plan. These actions include but are not limited to the following operations:

[0080] Execute the computing management module API, migrate virtual machines, etc.

[0081] Execute the computing management module API to turn off / on the virtual machine monitor, etc.

[0082] Execute bare metal management module API to start and shut down the machine;

[0083] Execute the storage management module API to migrate cloud hard disks, etc.;

[0084] Execute the storage management module API, migrate storage pools, etc.;

[0085] Execute network management module API, load balancing and other operations; as well as customize other actions.

[0086] Key Points:

[0087] The key points of the present invention are mainly reflected in the following aspects:

[0088] 1. Policy-driven resource optimization: The system provides a policy-driven optimization framework that allows administrators to define optimization policies to guide resource optimization behavior. This policy-driven approach enables the system to automatically perform optimization tasks such as virtual machine migration and resource allocation adjustment according to pre-set rules.

[0089] 2. Decision Maker: The system contains a decision maker that can make decisions based on real-time monitoring data and predefined strategies. The decision maker can evaluate multiple optimization strategies and select the best one to execute.

[0090] 3. Continuous optimization capability: The system can continuously monitor resource usage and automatically perform optimization operations in real time or at regular intervals as needed. This continuous optimization capability helps maintain efficient resource usage.

[0091] These key points make the system a powerful tool that can help administrators optimize resource usage, reduce costs, and improve data center operational efficiency.

[0092] Beneficial effects:

[0093] The present invention proposes a dynamic resource optimization system for a virtualized cloud platform. The system of the present invention aims to provide automation and continuous optimization capabilities for the use of resources in a virtualized cloud platform environment. It has the following beneficial effects:

[0094] The system is equipped with an energy-saving and emission-reduction strategy to identify and merge virtual machine instances scattered across multiple hosts, merging them onto fewer hosts to achieve the goal of energy saving and emission reduction.

[0095] The system is equipped with performance improvement strategies to help capacity planning. It can help users plan capacity based on historical data and forecast analysis to avoid insufficient or wasted resources.

[0096] The system is equipped with a load balancing strategy to dynamically migrate virtual machine instances to achieve load balancing and resource optimization. By automatically optimizing resource allocation, it reduces idle resources and improves resource utilization. It ensures that server loads are roughly equal and improves overall system performance.

[0097] The system is equipped with a fault migration strategy, which predicts potential fault risks by monitoring abnormal behaviors and early warning information in the system, and takes measures in advance to avoid the impact of system faults on the business.

[0098] The system is equipped with a node maintenance strategy to monitor the status of the host in real time. In the event of failure or maintenance, the system automatically migrates the affected virtual machines according to predefined processing rules, speeding up failure recovery and reducing business interruption time.

[0099] In general, this system can help cloud platform administrators automatically identify and implement system optimization measures, improve the overall efficiency and cost-effectiveness of cloud platform cloud environment resource utilization, and is a very valuable automated service.

[0100] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A dynamic resource optimization system for a virtualized cloud platform, characterized in that: The system includes: interface and interactive API services, decision makers and policy applications; Interface and interactive API services are used to provide REST API interfaces to the outside world, allowing administrators to interact with the system through CLI and webUI; The decision maker includes a strategy evaluation module and an optimization action planning module, and stores target information, strategy information, and data source information. The decision maker forms a data source based on the real-time and historical data collected by the monitoring system, selects relevant strategies according to the specified resource optimization goals, and after completing the strategy evaluation preparation, performs strategy evaluation through the strategy evaluation module to calculate a preliminary non-time-sequential solution. The solution then generates a final time-sequential action plan through the optimization action planning module. The action plan generated by the decision maker includes several workflows, and each workflow contains several optimization actions. The strategy application component is the execution component in the system; the strategy application component executes each optimization action in all workflows in the generated action plan in turn or in parallel to optimize and adjust the resources.

2. The dynamic resource optimization system for a virtualized cloud platform according to claim 1, characterized in that: In the cloud platform, when the user proposes a resource optimization goal to the system, the system first collects real-time and historical data from various components and services of the cloud platform. These data include the status and performance indicators of computing nodes, cloud platform services, virtual machines, storage and network devices; then, the system uses pre-defined strategies to perform policy evaluation on the collected data; each strategy has a specific goal, and the system will generate a series of optimization suggestions based on these strategies, and evaluate and score these suggestions to select the best solution; finally, in the policy application stage, the system applies the best strategy obtained in the policy evaluation process to the actual environment, including migrating virtual machines, adjusting resource allocation or shutting down unused resource operations. The system will also continue to monitor the optimization results to ensure that the optimization goals are achieved.

3. The dynamic resource optimization system for a virtualized cloud platform according to claim 1 or 2, characterized in that: The strategic assessment preparation includes: S201, the administrator selects optimization goals on the interface, including: energy saving and emission reduction goals, performance improvement goals, load balancing goals, fault migration goals and node maintenance goals; S202, the administrator selects a strategy on the interface, including an energy-saving and emission-reduction strategy, a performance-enhancing strategy, a load-balancing strategy, a fault-migration strategy, and a node-maintenance strategy, configures a default strategy for the target, and selects one or more strategies for each target; S203, the administrator creates a policy evaluation template, specifies the optimization target and policy, or only optimizes the target without specifying the optimization policy and adopts the configured default policy; S204, the administrator sends a policy assessment template creation instruction to the interactive API service, and writes the created object and initial status into the database; S205, the administrator creates and starts a policy evaluation object, the policy evaluation object includes an optimization target, and is created using an existing evaluation template; S206. The administrator sends the assessment creation instruction to the interactive API service, and the created object and initial status are written into the database; S207, start policy evaluation and send the message to the message queue bus; S208, the decision maker receives the message and prepares to perform the evaluation; S209, strategy selection, directly using the specified strategy, or automatically selecting one or more strategies when only the target is specified.

4. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: Energy conservation and emission reduction policies are used to identify and consolidate virtual machine instances that are scattered across multiple hosts onto fewer hosts.

5. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: Performance improvement strategies are used for capacity planning. Based on historical data and predictive analysis, they help users plan capacity and avoid resource shortages or waste.

6. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: Load balancing strategies are used to dynamically migrate virtual machine instances to achieve load balancing and resource optimization; by automatically optimizing resource allocation, ensuring that server loads are roughly equal.

7. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: The fault migration strategy is used to predict potential fault risks by monitoring abnormal behaviors and warning information in the system, and take measures in advance to avoid the impact of system failures on the business.

8. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: The node maintenance policy is used to monitor the status of the host in real time. In the event of failure or maintenance, the affected virtual machines are automatically migrated according to predefined processing rules to accelerate failure recovery and reduce business interruption time.

9. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: In S205, two types of evaluation objects are supported when they are created: a single evaluation mode and a continuous evaluation mode. The single evaluation mode is activated once, and an action plan is finally provided after activation. The continuous evaluation model continuously creates new action plans at given intervals.

10. The dynamic resource optimization system for a virtualized cloud platform according to claim 3, characterized in that: The policy application component performs the following steps: S501, automatically executing the recommended action plan according to the configuration schedule, executing according to the custom event trigger, or manually executing; when there are multiple strategies, if executing manually, refer to the efficacy score of the solution calculated by each strategy to select which action plan to adopt; S502, sending the execution plan instruction to the computing API, storage API, and bare metal management API; S503, writing the relevant status into the database; S504, sending to the message queue bus; S505, after receiving the message, the strategy application module starts to execute the optimization action plan; S506, looping and executing each optimization action of all workflows in the action plan; These actions include: Execute the computing management module API to migrate virtual machines; Execute the computing management module API to turn off / on the virtual machine monitor; Execute the bare metal management module API to start and shut down the machine; Execute the storage management module API to migrate the cloud hard disk; Execute the storage management module API to migrate the storage pool; Execute the network management module API to perform load balancing operations; and customize other actions.

Images