A cloud service resource collaborative scheduling method and system based on dynamic pricing

Abstract

The application discloses a cloud service resource collaborative scheduling method and system based on dynamic pricing, and the method comprises the following steps: calculating the unit cloud service resource operation cost of a data center and the maximum number of available virtual machines; combining the number of instances to be allocated to calculate the dynamic price signal of each region at the current time step; calculating the computing power demand of the cloud computing task of each region at the current time step under the dynamic price signal; comparing the computing power demand with the number of instances to be allocated, if the difference between the two is within a set range, then taking the current number of instances to be allocated as the planned allocation instance number, selecting the execution region and execution period for each task, allocating cloud service resources according to the number of instances required by each task, outputting a cloud service resource collaborative scheduling scheme, otherwise adjusting the number of instances to be allocated. The application can realize dynamic optimization scheduling of cloud service resources, improve the utilization rate of computing power resources, improve the load operation characteristics of the data center, and reduce the operation cost of the data center.

Description

Technical Field

[0001] This invention belongs to the field of cloud resource scheduling technology, and relates to a method and system for collaborative scheduling of cloud service resources based on dynamic pricing. Background Technology

[0002] With the rapid development of cloud computing, big data, and artificial intelligence, cloud service providers typically deploy multiple data centers in different regions to meet the computing power needs of distributed users. Due to differences in computing resource allocation, electricity prices, energy structures, and server operating status across different regions, the energy consumption costs and resource utilization of the same cloud computing task vary significantly depending on the region and time period. Furthermore, cloud computing tasks differ in terms of computational scale, execution time, response latency requirements, and resource utilization duration. Without an effective scheduling and control mechanism, some data centers may operate at high loads or during periods of high electricity prices, while computing resources in other regions or during periods of low electricity prices may not be fully utilized.

[0003] Existing technologies primarily rely on fixed computing power pricing and static resource allocation mechanisms. In real-world applications, these methods have significant limitations, mainly in the following two aspects: First, fixed pricing mechanisms lack responsiveness to changes in cloud computing task demand and resource operating status. During peak computing power demand periods, user requests increase rapidly, while service prices and resource scheduling strategies remain unchanged, easily leading to localized computing resource shortages, increased task queuing times, and higher service response latency. Conversely, during low-load periods, some servers and computing instances have low utilization rates, resulting in resource idleness and energy waste. Second, existing technologies do not fully utilize the guiding role of temporal and spatial differences in electricity prices on cloud service resource scheduling. In actual data center operation scenarios, electricity prices fluctuate significantly with time and region, especially during peak electricity consumption periods, where price increases typically coincide with peak grid load. If cloud computing tasks are still executed centrally according to fixed rules, it will lead to higher power loads in data centers during periods of high electricity prices, increasing operating costs and exacerbating grid pressure, hindering the achievement of peak shaving and valley filling and optimal resource allocation for data centers. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a cloud service resource collaborative scheduling method and system based on dynamic pricing. By comprehensively considering the spatiotemporal differences in electricity prices, real-time data center load, and the availability of computing resources, a dynamic price signal for resource scheduling is generated. Based on this signal, the execution area, execution time period, and resource instance allocation of cloud computing tasks are controlled. This enables dynamic matching between cloud computing tasks and data center resources, taking into account the spatiotemporal characteristics of electricity prices and the cross-time period occupation of computing resources. This achieves dynamic optimization scheduling of cloud service resources, improves the utilization rate of computing resources, enhances the load operation characteristics of data centers, and reduces data center operating costs.

[0005] The present invention adopts the following technical solution.

[0006] A cloud service resource collaborative scheduling method based on dynamic pricing includes: Step 1: Calculate the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region; Step 2: Calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used in each historical time step and the virtual machine release rate; Step 3: Calculate the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated; Step 4: Calculate the computing power requirement of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. Step 5: Compare the computing power requirement with the number of virtual machines to be allocated. If the difference is within the set range, the current number of virtual machines to be allocated is taken as the planned number of virtual machines to be allocated, and proceed to Step 6. Otherwise, adjust the number of virtual machines to be allocated and return to Step 3. Step 6: Based on the planned number of virtual machines and the dynamic price signal of each region at the current time step, select the execution region and execution time period for each task, allocate cloud service resources according to the number of virtual machines required by each task, and output the cloud service resource collaborative scheduling scheme.

[0007] Preferably, step 1 involves calculating the unit cloud service resource operating cost of each regional data center at the current time step based on the time-of-use electricity price and data center parameters, using the following formula: (4) in, It is the unit cloud service resource operating cost of the data center in region r at the current time step t; It is the time-of-use electricity price for region r at the current time step t; and These are the average power of the data center in region r under idle and full-load conditions, respectively. This is the maximum number of virtual machines available in the data center of region r.

[0008] Preferably, step 2 involves calculating the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used in each historical time step and the virtual machine release rate, using the following formula: (6) in, It represents the maximum number of available virtual machines in the data center of region r at the current time step t; This represents the maximum number of virtual machines available in the data center of region r; This represents the number of virtual machines used by the data center in region r at time step a; The data center in region r at time step Virtual machine release rate.

[0009] Preferably, step 3 involves calculating the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated, using the following formula: (8) in, It is the dynamic price signal of the data center in region r at the current time step t; , , It represents the unit cloud service resource operating cost, maximum number of available virtual machines, and number of virtual machines to be allocated for the data center in region r at the current time step t. The highest price acceptable to the user.

[0010] Preferably, step 4 involves calculating the computing power requirement of cloud computing tasks in each region at the current time step based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region under the dynamic price signal, using the following formula: (9) in, and These represent the computing power requirement and maximum potential requirement for cloud computing tasks in region r at the current time step t, respectively. It is the price elasticity coefficient of cloud computing tasks in region r; It is the dynamic price signal of the data center in region r at the current time step t.

[0011] Preferably, step 5, comparing the computing power requirement with the number of virtual machines to be allocated, and if the difference is within a set range, then the current number of virtual machines to be allocated is taken as the planned number of virtual machines to be allocated, and proceeding to step 6; otherwise, adjusting the number of virtual machines to be allocated and returning to step 3, includes: Compare the number of virtual machines to be allocated Computing power requirements : If | If |≤δ, where δ is the preset error, then the number of virtual machines to be allocated will be... As a planned allocation of virtual machines Proceed to step 6; If | |>δ, and > Then increase And return to step 3; If | |>δ, and < Then decrease Then return to step 3.

[0012] Preferably, in step 6, based on the planned number of virtual machines and the dynamic price signal of each region at the current time step, an execution region and execution time period are selected for each task, and cloud service resources are allocated according to the number of virtual machines required by each task, outputting a cloud service resource collaborative scheduling scheme, as follows: With the constraint that the total number of virtual machines required for all tasks executed in the same region and at the same time step does not exceed the planned number of virtual machines allocated for that region and time step, the execution region is selected for each task according to the following formula. and execution period :

[0013] in, , For the task j The set of regions and time periods that can be scheduled; It is the dynamic price signal of the data center in region r at the current time step t; In the selected execution region and execution period The corresponding number of virtual machines are allocated to the corresponding task according to the number of virtual machines required for each task. Output a cloud service resource collaborative scheduling scheme, including: the execution area, execution time period, and number of virtual machines allocated for each task.

[0014] A cloud service resource collaborative scheduling system based on dynamic pricing includes: The operating cost calculation module calculates the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region. The available cloud service resource calculation module is used to calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used at each historical time step and the virtual machine release rate. The dynamic price signal calculation module is used to calculate the dynamic price signal of each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated. The computing power demand calculation module is used to calculate the computing power demand of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. The optimization and adjustment module is used to compare the computing power requirement with the number of virtual machines to be allocated. If the difference between the two is within a set range, the current number of virtual machines to be allocated is used as the planned number of virtual machines to be allocated, and the module enters the collaborative scheduling module. Otherwise, the number of virtual machines to be allocated is adjusted and the module returns to the dynamic price signal calculation module. The collaborative scheduling module is used to select an execution region and execution time period for each task based on the planned number of virtual machines and the dynamic price signal of the current time step of each region, and to allocate cloud service resources according to the number of virtual machines required by each task, and output a collaborative scheduling scheme for cloud service resources.

[0015] A terminal includes a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.

[0016] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0017] Compared with the prior art, the beneficial effects of the present invention include at least the following: This invention uses dynamic price signals as the basis for task scheduling and resource control, guiding delayed or migrated cloud computing tasks to shift to time periods and regions with low electricity prices, low loads, or abundant available resources. By coordinating the control of task execution time periods, execution regions, and resource instance allocation, it avoids the concentrated execution of a large number of tasks during high electricity price periods. Under the premise of ensuring user service quality and task execution constraints, it realizes dynamic optimization scheduling of cloud service resources in time and space dimensions, reduces data center operating costs, improves computing resource utilization efficiency, and improves the load distribution status of cloud service systems.

[0018] This invention generates dynamic price signals for cloud service resources based on the electricity price level, real-time computing load, and available resource status of each data center's region. This ensures that the price always lies between the unit operating cost and the highest acceptable price, avoiding abnormal price fluctuations and proactively increasing the price signal when electricity prices are high and resource occupancy is high. This reflects changes in electricity costs and the scarcity of cloud service resources, improving the price signal's ability to characterize spatiotemporal energy consumption differences. This helps reduce the operating costs of cloud service providers and improves the balance of resource utilization across data centers in different regions. Attached Figure Description

[0019] Figure 1 This is a flowchart of a cloud service resource collaborative scheduling method based on dynamic pricing according to the present invention.

[0020] Figure 2 This is a schematic diagram illustrating the principle of a cloud service resource collaborative scheduling method based on dynamic pricing according to the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0022] Embodiment 1 of the present invention provides a cloud service resource collaborative scheduling method based on dynamic pricing, such as... Figures 1-2 As shown, it includes: Step 1: Calculate the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region; Step 2: Calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used in each historical time step and the virtual machine release rate; Step 3: Calculate the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated; Step 4: Calculate the computing power requirement of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. Step 5: Compare the computing power requirement with the number of virtual machines to be allocated. If the difference is within the set range, the current number of virtual machines to be allocated is taken as the planned number of virtual machines to be allocated, and proceed to Step 6. Otherwise, adjust the number of virtual machines to be allocated and return to Step 3. Step 6: Based on the planned number of virtual machines and the dynamic price signal of each region at the current time step, select the execution region and execution time period for each task, allocate cloud service resources according to the number of virtual machines required by each task, and output the cloud service resource collaborative scheduling scheme.

[0023] The specific principle is as follows: First, by analyzing the spatiotemporal variation characteristics of electricity prices in different regions where distributed data centers are located (spatiotemporal differences in electricity prices). Availability of computing resources and server load level Analysis and modeling of cloud service resource operation status yields a cloud service resource operation status perception model, which is used to characterize the supply capacity of cloud service resources in different regions and at different time steps. and operating costs Specifically, the cloud service resource operation status perception model is based on data center resource utilization. Model, power operation cost (energy consumption cost) Model, server load level (energy consumption) Model and computing resource usage characteristics across time periods The design takes into account the energy consumption of the data center under different load levels. Calculation constraints and regional time-of-use electricity pricing energy costs Constraints and current schedulable resource constraints Secondly, based on the resource / computing power requirements of the cloud computing tasks submitted by users. And the price elasticity coefficient, which corresponds to the dynamic price response characteristics. A cloud computing task demand model is generated by modeling the cloud computing task demand process. This model describes the submission, delayed execution, or cross-regional migration characteristics of different types of cloud computing tasks under the influence of dynamic computing power price signals. The cloud computing task demand model considers the maximum potential demand of cloud computing tasks. Finally, based on the aforementioned resource operation status awareness model and cloud computing task demand model, a dynamic price signal for resource scheduling is generated to characterize the scheduling cost differences of executing cloud computing tasks in different regions and at different time steps. This dynamic price signal is then transformed into task scheduling and resource control strategies. Based on the dynamic price signal, the execution areas of cloud computing tasks in each region and at each time step (period) are determined. and execution period This includes resource instance allocation schemes to achieve dynamic optimization and scheduling of cloud service resources. Details are as follows: 1. Cloud service resource operation status awareness model: On the cloud service resource supply side, the current computing power resource usage status is first characterized based on the number of available instances in the data center, forming resource availability status information; Furthermore, by combining the energy consumption characteristics of data centers under different load levels, the energy consumption cost corresponding to executing cloud computing tasks in different regions and at different time steps is calculated; At the same time, regional time-of-use electricity price information is introduced to convert data center energy consumption costs into electricity operating costs, and to serve as an important basis for dynamic price signal generation and subsequent resource scheduling and control.

[0024] The above processing can identify differences in the operating costs of cloud service resources in different regions and at different times, providing a basis for selecting task execution regions, adjusting execution times, and allocating resource instances.

[0025] (1) Cloud service resource usage status of the data center at the current time step: (1) It represents the resource utilization rate of the data center located in region r at time step t. This represents the number of cloud service instances used by the data center located in region r at time step t. This is the maximum number of available instances in the data center located in region r.

[0026] (2) Under this cloud service resource utilization level, the energy consumption of the data center can be estimated by linear interpolation between idle power and peak power: (2) This is the energy consumption of the data center located in region r at time step t. and These are the average power of the data center located in region r under idle and full-load conditions, respectively.

[0027] (3) Further considering the time-of-use electricity price in the region, the short-term power operating cost of the data center performing tasks at the current time step is as follows: (3) This is the time-of-use electricity price for region r at time step t. It is the short-term energy consumption cost of a data center located in region r at time step t.

[0028] (4) According to the definition of incremental cost per unit resource, the unit resource operating cost corresponding to the allocation of a new cloud service instance in the current time step of the data center can be obtained through the relationship between the power operating cost and the number of allocated instances. Combining formulas (1)-(3), the following can be derived:

[0029] Right now (4) It is the unit cloud service resource operating cost of the data center located in region r at time step t.

[0030] (5) To describe the degree of response of cloud computing tasks to dynamic computing power price signals, a dynamic price response coefficient is further introduced.

[0031] When the dynamic computing power price signal is low, the system prioritizes the execution of delayed or migrated tasks in this region and at this time step. When the dynamic computing power price signal is high, the system reduces the task scheduling priority of that region and time step, and transfers some tasks to low-cost time periods or resource-rich regions.

[0032] (5) It is the price elasticity coefficient of a user submitting a cloud computing task in region r at time step t. It represents the highest acceptable price for cloud computing services for users submitting cloud computing tasks.

[0033] (6) Since user-submitted cloud computing tasks often need to be executed across multiple time steps, some cloud service instances cannot be released before the task is completed, resulting in a limitation on the resources available for new task scheduling in the current time step.

[0034] Therefore, the maximum schedulable cloud service resources of the data center at the current time step can be expressed as: (6) It is the maximum available cloud service resource of the data center located in region r at time step t. It is the instance release rate of the data center located in region r at time step t.

[0035] (7) Based on the above, the present invention constructs a dynamic computing power price signal model based on the unit resource operation cost, which is used to characterize the scheduling cost differences of cloud service resources in different regions and at different time steps, and uses the dynamic computing power price signal as the control basis for task execution time selection, execution region selection and resource instance allocation.

[0036] (7) (8) Combine equations (5) and (7), substitute equation (5) into equation (7), and eliminate the error. And solve it in reverse. Then, the formula for calculating dynamic price signals for resource scheduling can be obtained: (8) 2. Cloud computing task requirement model: On the demand side of cloud computing tasks, a cloud computing task demand model is constructed based on the resource requirements, execution time limits, and response latency requirements of cloud computing tasks.

[0037] The model introduces a dynamic price response parameter to describe the characteristics of cloud computing tasks in terms of submission, delayed execution, or cross-regional migration under the influence of price signals. It also combines maximum potential demand and service quality constraints to limit the schedulable task size at each time step, providing a basis for subsequent resource scheduling.

[0038] (9) but ,in and These represent the computing power requirement and maximum potential requirement of cloud computing tasks in region r at time step t, respectively. It is the price elasticity coefficient of cloud computing tasks in region r.

[0039] 3. Resource scheduling mechanism based on dynamic pricing: Based on the cloud service resource operation status perception model and the cloud computing task demand model, dynamic price signals for resource scheduling are generated.

[0040] Within each region and time step, based on data center operating costs, available resource status, and server load levels, the execution region, execution period, and resource instance allocation scheme for cloud computing tasks are generated to achieve dynamic optimization and scheduling of cloud service resources.

[0041] (10) The cloud service resource scheduling capability function of the data center located in region r at time step t is based on dynamic price signals, available resource status and server load level, specifically as formula (8). The task requirement response function formed by a cloud computing task located in region r at time step t based on resource requirements, execution time limits, and response latency requirements is expressed as formula (9). It is a dynamic computing power price signal used in resource scheduling decisions.

[0042] This resource scheduling mechanism incorporates factors such as the spatiotemporal differences in electricity prices, cross-period occupation of computing resources, server load status, and cloud computing task response characteristics during the modeling process, in order to characterize the operating costs and schedulable capabilities of cloud service resources in different regions and at different time steps.

[0043] By calculating dynamic computing power price signals and generating task scheduling strategies, the number of tasks allocated and the resource instance configuration scheme of each regional data center at each time step can be determined.

[0044] Based on the above principles, the implementation steps of the cloud service resource collaborative scheduling method based on dynamic pricing in this invention are as follows: Step 1: Calculate the unit cloud service resource operating cost of each data center in each region at the current time step based on the time-of-use electricity price and data center parameters. The formula is as follows: (4) in, It is the unit cloud service resource operating cost of the data center in region r at the current time step t; It is the time-of-use electricity price for region r at the current time step t; and These are the average power of the data center in region r under idle and full-load conditions, respectively. This is the maximum number of virtual machines available in the data center of region r.

[0045] Step 2: Calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service instances used and the instance release rate at each historical time step. The formula is as follows: (6) in, It represents the maximum number of available virtual machines in the data center of region r at the current time step t; This represents the maximum number of virtual machines available in the data center of region r; This represents the number of virtual machines used by the data center in region r at time step a; The data center in region r at time step Instance release rate.

[0046] Step 3: Calculate the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of instances to be allocated. The formula is as follows: (8) in, It is the dynamic price signal of the data center in region r at the current time step t; , , This refers to the unit cloud service resource operating cost, maximum number of available virtual machines, and number of instances to be allocated for the data center in region r at the current time step t. The highest price acceptable to the user.

[0047] Step 4: Calculate the computing power requirement of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. The formula is as follows: (9) in, and These are the computing power requirements and maximum potential requirements of cloud computing tasks in region r at the current time step t (based on the maximum number of tasks arriving at each time step in the historical data of data center r in region r). This is the price elasticity coefficient for cloud computing tasks in region r (which can be set according to the price sensitivity of users in each region; for example, it can be set to 0.2, 0.5, and 0.8 for low-price-sensitive, medium-price-sensitive, and high-price-sensitive users, respectively). It is the dynamic price signal of the data center in region r at the current time step t.

[0048] Step 5: Compare the computing power requirement with the number of instances to be allocated. If the difference is within a set range, use the current number of instances to be allocated as the planned number of instances to be allocated, and proceed to Step 6; otherwise, adjust the number of instances to be allocated and return to Step 3, including: Compare the number of instances to be allocated Computing power requirements : If | If |≤δ, where δ is the preset error, then the number of instances to be allocated will be... As the number of planned allocation instances Proceed to step 6; If | |>δ, and > Then increase And return to step 3; If | |>δ, and < Then decrease Then return to step 3.

[0049] In practice, the adjustment step size for increasing / decreasing can be set according to the actual scheduling accuracy requirements.

[0050] Step 6: Based on the planned number of instances and the dynamic price signal of each region at the current time step, select the execution region and execution time period for each task, allocate cloud service resources according to the number of instances required by each task, and output the cloud service resource collaborative scheduling scheme, as follows: With the constraint that the total number of virtual machines required for all tasks executed in the same region and at the same time step does not exceed the planned number of virtual machines allocated for that region and time step, the execution region is selected for each task according to the following formula. and execution period :

[0051] in, , For the task j The set of regions and time periods that can be scheduled; It is the dynamic price signal of the data center in region r at the current time step t; In the selected execution region and execution period The corresponding number of instances are allocated to the corresponding tasks according to the number of instances required for each task. Output a cloud service resource collaborative scheduling solution, including: the execution area, execution time period, and number of instances allocated for each task.

[0052] This method can guide deferred or migrated tasks to regions and time steps with low electricity prices, low loads, or abundant resources. Under the premise of ensuring user service quality and task execution constraints, it can achieve dynamic optimization scheduling of cloud service resources in time and space dimensions, reduce data center operating costs, and improve the utilization efficiency of computing resources.

[0053] Embodiment 2 of the present invention provides a cloud service resource collaborative scheduling system based on dynamic pricing, the system comprising: The operating cost calculation module calculates the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region. The available cloud service resource calculation module is used to calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service instances used and the instance release rate at each historical time step. The dynamic price signal calculation module is used to calculate the dynamic price signal of each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of instances to be allocated. The computing power demand calculation module is used to calculate the computing power demand of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. The optimization and adjustment module is used to compare the computing power requirement with the number of instances to be allocated. If the difference between the two is within a set range, the current number of instances to be allocated is used as the planned number of instances to be allocated, and the module enters the collaborative scheduling module. Otherwise, the number of instances to be allocated is adjusted and the module returns to the dynamic price signal calculation module. The collaborative scheduling module is used to select the execution region and execution time period for each task based on the planned number of allocated instances and the dynamic price signal of the current time step of each region, and to allocate cloud service resources according to the number of instances required by each task, and output a collaborative scheduling scheme for cloud service resources.

[0054] Embodiment 3 of the present invention provides a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.

[0055] Embodiment 4 of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0056] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0057] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0058] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0059] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A cloud service resource collaborative scheduling method based on dynamic pricing, characterized in that, include: Step 1: Calculate the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region; Step 2: Calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used in each historical time step and the virtual machine release rate; Step 3: Calculate the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated; Step 4: Calculate the computing power requirement of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. Step 5: Compare the computing power requirement with the number of virtual machines to be allocated. If the difference is within the set range, the current number of virtual machines to be allocated is taken as the planned number of virtual machines to be allocated, and proceed to Step 6. Otherwise, adjust the number of virtual machines to be allocated and return to Step 3. Step 6: Based on the planned number of virtual machines and the dynamic price signal of each region at the current time step, select the execution region and execution time period for each task, allocate cloud service resources according to the number of virtual machines required by each task, and output the cloud service resource collaborative scheduling scheme.

2. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: Step 1 involves calculating the unit cloud service resource operating cost of each regional data center at the current time step based on the time-of-use electricity price and data center parameters. The formula is as follows: （4） in, It is the unit cloud service resource operating cost of the data center in region r at the current time step t; It is the time-of-use electricity price for region r at the current time step t; and These are the average power of the data center in region r under idle and full-load conditions, respectively. This is the maximum number of virtual machines available in the data center of region r.

3. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: Step 2 involves calculating the maximum number of available virtual machines (VMs) for each region's data center in the current time step based on the number of cloud service VMs used in each historical time step and the VM release rate. The formula is as follows: （6） in, It represents the maximum number of available virtual machines in the data center of region r at the current time step t; This represents the maximum number of virtual machines available in the data center of region r; This represents the number of virtual machines used by the data center in region r at time step a; The data center in region r at time step Virtual machine release rate.

4. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: Step 3 involves calculating the dynamic price signal for each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated. The formula is as follows: （8） in, It is the dynamic price signal of the data center in region r at the current time step t; , , It represents the unit cloud service resource operating cost, maximum number of available virtual machines, and number of virtual machines to be allocated for the data center in region r at the current time step t. The highest price acceptable to the user.

5. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: Step 4 involves calculating the computing power requirement of cloud computing tasks in each region at the current time step based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region under the dynamic price signal. The formula is as follows: （9） in, and These represent the computing power requirement and maximum potential requirement for cloud computing tasks in region r at the current time step t, respectively. It is the price elasticity coefficient of cloud computing tasks in region r; It is the dynamic price signal of the data center in region r at the current time step t.

6. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: Step 5 compares the computing power requirement with the number of virtual machines to be allocated. If the difference is within a set range, the current number of virtual machines to be allocated is used as the planned number of virtual machines to be allocated, and the process proceeds to step 6. Otherwise, the number of virtual machines to be allocated is adjusted and the process returns to step 3, including: Compare the number of virtual machines to be allocated Computing power requirements : If | If |≤δ, where δ is the preset error, then the number of virtual machines to be allocated will be... As a planned allocation of virtual machines Proceed to step 6; If | |>δ, and > Then increase And return to step 3; If | |>δ, and < Then decrease Then return to step 3.

7. The cloud service resource collaborative scheduling method based on dynamic pricing according to claim 1, characterized in that: In step 6, based on the planned number of virtual machines and the dynamic price signal of each region at the current time step, an execution region and execution time period are selected for each task, and cloud service resources are allocated according to the number of virtual machines required by each task. The cloud service resource collaborative scheduling scheme is then output, as follows: With the constraint that the total number of virtual machines required for all tasks executed in the same region and at the same time step does not exceed the planned number of virtual machines allocated for that region and time step, the execution region is selected for each task according to the following formula. and execution period : in, , For the task j The set of regions and time periods that can be scheduled; It is the dynamic price signal of the data center in region r at the current time step t; In the selected execution region and execution period The corresponding number of virtual machines are allocated to the corresponding task according to the number of virtual machines required for each task. Output a cloud service resource collaborative scheduling scheme, including: the execution area, execution time period, and number of virtual machines allocated for each task.

8. A cloud service resource collaborative scheduling system based on dynamic pricing, comprising the method described in any one of claims 1-7, characterized in that, The system includes: The operating cost calculation module calculates the unit cloud service resource operating cost of each data center in the current time step based on the time-of-use electricity price and data center parameters of each region. The available cloud service resource calculation module is used to calculate the maximum number of available virtual machines in each region's data center at the current time step based on the number of cloud service virtual machines used at each historical time step and the virtual machine release rate. The dynamic price signal calculation module is used to calculate the dynamic price signal of each region at the current time step based on the unit cloud service resource operating cost, the maximum number of available virtual machines, and the number of virtual machines to be allocated. The computing power demand calculation module is used to calculate the computing power demand of cloud computing tasks in each region at the current time step under the dynamic price signal, based on the maximum potential demand and price elasticity coefficient of cloud computing tasks in each region at the current time step. The optimization and adjustment module is used to compare the computing power requirement with the number of virtual machines to be allocated. If the difference between the two is within a set range, the current number of virtual machines to be allocated is used as the planned number of virtual machines to be allocated, and the module enters the collaborative scheduling module. Otherwise, the number of virtual machines to be allocated is adjusted and the module returns to the dynamic price signal calculation module. The collaborative scheduling module is used to select an execution region and execution time period for each task based on the planned number of virtual machines and the dynamic price signal of the current time step of each region, and to allocate cloud service resources according to the number of virtual machines required by each task, and output a collaborative scheduling scheme for cloud service resources.

9. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-7.

Images