Cloud computing resource dynamic scheduling optimization method and system

Abstract

The invention relates to the field of data processing, in particular to a cloud computing resource dynamic scheduling optimization method and system. The method comprises the following steps: acquiring a task which is being processed by each server at the current moment and recording the task as a current task; obtaining the relevance of every two tasks and the priority of each task; screening out association-constrained tasks from the current tasks according to the relevance of every two tasks, and obtaining the subsequent constraint duration of the association-constrained tasks; constructing a scheduling objective function according to the subsequent constraint duration of the task subjected to association constraint; and the value of the scheduling target function is maximized to obtain the corresponding subsequent task allocation mode, so that the reasonability of task allocation is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and particularly to a method and system for optimizing dynamic scheduling of cloud computing resources. Background Art

[0002] In the wide application scenarios of cloud computing technology, dynamic resource scheduling is the core link to ensure the efficient operation of the system and improve resource utilization. Traditional cloud computing resource scheduling methods often perform static allocation based on a single attribute of tasks (such as task duration, resource requirements, etc.), ignoring the potential relevance and priority differences between tasks, resulting in low resource utilization efficiency and task execution delays.

[0003] With the increasing complexity of cloud computing application scenarios, there are a large number of tasks with dependencies in many business processes. For example, in the data analysis scenario, data cleaning tasks need to be completed before data modeling and analysis can be carried out. If such related tasks are not reasonably planned and scheduled, it may lead to the computing resources being idle during the waiting for related tasks to be issued, resulting in resource waste. Existing scheduling strategies usually only focus on the allocation of currently issued tasks and do not consider possible future related tasks, and cannot effectively utilize idle resources before related tasks are issued, resulting in a lack of foresight and flexibility in resource scheduling.

[0004] On the other hand, in the process of task scheduling, task priority is a key factor affecting the quality of system services. In scenarios with extremely high requirements for response speed, such as financial transaction settlement and medical image diagnosis, the timely processing of high-priority tasks is directly related to the success or failure of the business and the health and safety of users' lives. This makes it very likely that when effectively utilizing idle resources, if short-term tasks are only allocated based on task duration or resource matching degree, high-priority tasks may not be able to obtain resources in time due to idle resources being occupied by low-priority short-term tasks.

[0005] In summary, how to balance task relevance and priority in cloud computing resource scheduling, reasonably utilize idle resources to execute short-term tasks before related tasks are issued, and ensure the timely allocation of high-priority tasks, and avoid task execution delays and resource waste caused by priority imbalance, has become a technical problem to be solved urgently. Summary of the Invention

[0006] In order to solve the problem of how to pre-plan related tasks while taking into account priorities, the present invention provides a method and system for optimizing dynamic scheduling of cloud computing resources.

[0007] In a first aspect, the present invention provides a method for optimizing dynamic scheduling of cloud computing resources, adopting the following technical solutions: A method for optimizing dynamic scheduling of cloud computing resources, comprising the steps of: Obtain the tasks being processed by each server at the current moment as the current tasks; Obtain the relevance between every two tasks and the priority of each task according to the historical processing data of the tasks; screen out the tasks subject to association constraints from the current tasks according to the relevance between every two tasks, and obtain the subsequent constraint duration of the tasks subject to association constraints, where the subsequent constraint duration is the expected arrival interval duration of the task with a greater relevance to the task subject to association constraints; construct a scheduling objective function, where the scheduling objective function is negatively correlated with the difference between the required predicted processing duration and the subsequent constraint duration of the subsequent tasks arranged on the server where the task subject to association constraints is located, and is also negatively correlated with the weighted waiting duration of all subsequent arranged tasks; the weighted waiting duration is obtained by weighting the predicted waiting duration of all subsequent arranged tasks with the priority as the weight; Maximize the value of the scheduling objective function to obtain the corresponding subsequent task allocation method to achieve task allocation.

[0008] The present invention takes into account the relevance of tasks, leaves time for the tasks subject to association constraints to wait for the arrival of associated tasks, prevents the processing of associated tasks from being interrupted, and improves the coherence of task processing; further, when arranging tasks considering association constraints, the difference between the required predicted processing duration and the subsequent constraint duration of the subsequent tasks arranged on the server where the task subject to association constraints is located is introduced, so as to limit the task arrangement within the constraint duration, so that the server can be effectively utilized before waiting for the arrival of associated tasks, and at the same time, prevent associated tasks from being interrupted. Further, when arranging tasks considering association constraints, the weighted waiting time is also introduced, so that when arranging tasks, not only the association situation of tasks is considered but also the priority of tasks is considered, effectively preventing the situation where some tasks with high priority cannot be arranged preferentially only considering the association situation of tasks, and improving the rationality of task arrangement.

[0009] Preferably, the constructing of the scheduling objective function includes: ; wherein, represents the subsequent constraint duration of the i-th task subject to association constraints, represents the required predicted processing duration of the j-th task subsequently arranged on the server where the i-th task subject to association constraints is located, represents a preset anti-zero coefficient, represents the number of tasks subsequently arranged on the server where the i-th task subject to association constraints is located, represents the number of tasks subject to association constraints, represents the predicted waiting duration of the z-th subsequently arranged task, represents the priority of the z-th subsequently arranged task, U represents the number of subsequently arranged tasks, represents the occupancy rate of the server, Represents the scheduling objective function.

[0010] When constructing the scheduling objective function of the present invention, the difference between the subsequent constraint duration and the predicted processing duration of the subsequent arranged tasks is introduced, so that the processing time of the tasks arranged within the constraint duration is as close as possible to the constraint duration, thereby preventing the processing process of associated tasks from being interrupted due to too long processing duration of subsequent arranged tasks, or preventing the server from being idle due to too short processing duration of subsequent arranged tasks; further, when constructing the objective function, the priority and the predicted waiting duration are introduced, so that the priority situation can be considered when arranging tasks, thereby preventing only the arrangement problem of associated tasks from being concerned when arranging tasks, resulting in tasks with high priority being arranged relatively backward, thus improving the rationality of task arrangement.

[0011] Preferably, the screening of the tasks subject to association constraints from the current tasks according to the association between every two tasks includes: Obtaining the time interval between the tasks subject to association constraints and other tasks for the issued time according to the historical processing data of the tasks, and obtaining the issued duration of the tasks subject to association constraints; If the association between the current task and other tasks is greater than the preset association threshold, the current task is determined as a task suspected of being subject to association constraints; If the time interval between the task suspected of being subject to association constraints and other tasks is greater than the issued duration of the current task, then the task suspected of being subject to association constraints is determined as a task subject to association constraints; Conversely, the task suspected of being subject to association constraints is not a task subject to association constraints.

[0012] The present invention screens out the tasks suspected of being subject to association constraints by comparing the association with the preset threshold, and at the same time compares the issued time interval with the issued duration to screen out the tasks that have exceeded the issued time interval, thereby preventing unnecessary waiting due to the non-issuance of associated tasks.

[0013] Preferably, the subsequent constraint duration of the tasks subject to association constraints includes: Obtaining the other task with the greatest association with the task subject to association constraints and recording it as the target task; Taking the difference between the issued time interval and the issued duration as the subsequent constraint duration of the tasks subject to association constraints.

[0014] The present invention excludes the issued duration of the current task by taking the difference between the issued time interval and the issued time length to obtain the relatively accurate issued duration of the associated task.

[0015] Preferably, the obtaining of the association between every two tasks according to the historical processing data of the tasks includes: Determine whether there is a data inflow and outflow relationship between two tasks. If there is a data inflow and outflow relationship, set the basic value of the task association to 1; otherwise, set it to 0. Set a time window, and calculate the confidence of the association rule for each pair of tasks using the Apriori algorithm based on the data within the time window. Multiply the confidence of the association rule by the basic value of the association to obtain the association between the two tasks. <>

[0016] The present invention comprehensively and accurately reflects the association between tasks based on whether there is a data inflow and outflow relationship between tasks and the confidence of the association rule of the tasks.

[0017] Preferably, before obtaining the corresponding subsequent task allocation method by maximizing the value of the scheduling objective function, a solution space needs to be constructed. The construction method of the solution space includes: For any task to be allocated, select a server as the target server according to any selection method, allocate the target server to each task to be allocated, and regard the corresponding relationship between all tasks to be allocated and the server as a feasible solution. The space composed of all feasible solutions is used as the solution space.

[0018] Preferably, the method for obtaining the predicted waiting duration includes: Obtain the predicted processing duration of each task, obtain the tasks to be processed arranged before each task, which are recorded as the pre - tasks to be processed. The predicted waiting duration of each task is obtained by adding the cumulative sum of the predicted processing durations of all the pre - tasks to be processed before each task and the predicted remaining processing duration of the current task on the corresponding server.

[0019] Preferably, the method of maximizing the value of the scheduling objective function to obtain the corresponding subsequent task allocation method includes: Maximize the scheduling objective function, use an optimization algorithm to solve the optimal solution of the scheduling objective function, and obtain the task allocation method corresponding to the optimal solution.

[0020] Preferably, the implementation of task allocation includes: Allocate servers to each task to be allocated according to the task allocation method corresponding to the optimal solution.

[0021] In a second aspect, the present invention provides a cloud computing resource dynamic scheduling optimization system, adopting the following technical solution: A cloud computing resource dynamic scheduling optimization system includes: a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above - mentioned cloud computing resource dynamic scheduling optimization method is implemented.

[0022] By adopting the above technical solution, a computer program is generated from the above-mentioned method for optimizing dynamic scheduling of cloud computing resources and stored in a memory to be loaded and executed by a processor, thereby manufacturing a terminal device according to the memory and the processor, which is convenient to use.

[0023] The present invention has the following technical effects: The present invention takes into account the relevance of tasks, leaves time for the subsequent tasks subject to association constraints to wait for the issuance of associated tasks, prevents the processing of associated tasks from being interrupted, and improves the coherence of task processing; Furthermore, when arranging tasks considering association constraints, the difference between the predicted processing duration and the subsequent constraint duration of the subsequent tasks arranged on the server where the tasks subject to association constraints are located is introduced, so as to limit the task arrangement within the constraint duration, so that the server can be effectively utilized before waiting for the issuance of associated tasks and at the same time prevent associated tasks from being interrupted.

[0024] Furthermore, when arranging tasks considering association constraints, a weighted waiting time is also introduced, so that when arranging tasks, not only the association of tasks is considered but also the priority of tasks is considered, effectively preventing the situation where some tasks with high priority cannot be arranged first only considering the association of tasks, and improving the rationality of task arrangement. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flowchart of the method in an embodiment of the method for optimizing dynamic scheduling of cloud computing resources of the present invention. DETAILED DESCRIPTION

[0026] An embodiment of the present invention discloses a method for optimizing dynamic scheduling of cloud computing resources, referring to Figure 1 , including steps S1 - S3: S1: Obtain the tasks being processed by each server at the current moment and record them as current tasks.

[0027] Specifically, obtain the tasks being processed by each server at the current moment and record them as current tasks.

[0028] S2: Obtain the correlation between every two tasks and the priority of each task based on the historical processing data of the tasks; filter out the tasks subject to correlation constraints in the current tasks based on the correlation between every two tasks, and obtain the subsequent constraint duration of the tasks subject to correlation constraints, where the subsequent constraint duration is the expected interval duration of the tasks with greater correlation than the tasks subject to correlation constraints; construct a scheduling objective function, which is negatively correlated with the difference between the required predicted processing duration and the subsequent constraint duration of the subsequent scheduled tasks on the server where the tasks subject to correlation constraints are located, and is also negatively correlated with the weighted waiting time of all subsequent scheduled tasks; the weighted waiting time is obtained by weighted calculation of the predicted waiting time of all subsequent scheduled tasks using priority as the weight.

[0029] It should be noted that in order to solve the problem of pre-planning related tasks while taking into account the priority, it is necessary to first obtain the relevance and priority of the tasks.

[0030] S20: Obtain the correlation between every two tasks and the priority of each task based on the historical processing data of the tasks.

[0031] Preferably, as an example, obtaining the relevance between every two tasks and the priority of each task based on the historical processing data of the tasks includes: Determine whether there is a data inflow and outflow relationship between the two tasks. If there is a data inflow and outflow relationship, set the basic value of the correlation between the two tasks to 1, otherwise set it to 0; Set a time window of preset length and use the Apriori algorithm to calculate the confidence of the association rule between each two tasks based on the data in the time window; The correlation between the two tasks is obtained by multiplying the confidence of the association rule by the correlation basis value.

[0032] The priority of each task is set manually, or the priority of the task is set using a priority setting network, where the priority setting network can use an existing network, such as an XGBoost network.

[0033] It's understandable that only when there's a data flow between two tasks, and they frequently occur together within a short timeframe, can they be considered interconnected. In other words, the processing of one task requires the data of the other, and the two tasks cannot be processed independently. Therefore, these tasks should be assigned to the same server, and there shouldn't be too many other tasks between them to ensure continuity between them.

[0034] It should be noted that the use of the Apriori algorithm to calculate the confidence of the association rule between each two tasks is an existing technology and will not be described in detail here.

[0035] S21: Screen out the tasks subject to association constraints from the current tasks according to the association between every two tasks, and obtain the subsequent constraint duration of the tasks subject to association constraints.

[0036] It should be noted that sometimes the associated tasks of the current task have not been issued yet. Therefore, in order to prevent the associated tasks from being interrupted when the subsequent associated tasks are issued due to the lack of idle processors. Therefore, it is necessary to predict and analyze whether there are associated tasks in the future of the current task, so as to reserve some time for the subsequent current tasks with a high probability of having associated tasks to wait for the issuance of the associated tasks. First of all, it is necessary to analyze the historical data to predict whether there are associated tasks in the current task and set an appropriate reserved time according to the issuance interval between two tasks in history.

[0037] Preferably, as an example, screening out the tasks subject to association constraints from the current tasks according to the association between every two tasks, and obtaining the subsequent constraint duration of the tasks subject to association constraints, including: Obtain the issuance time interval between the tasks subject to association constraints and other tasks according to the historical processing data of the tasks, and obtain the issued duration of the tasks subject to association constraints; If the association between the current task and other tasks is greater than the preset association threshold, the current task is determined as a task suspected of being subject to association constraints; If the issuance time interval between the task suspected of being subject to association constraints and other tasks is greater than the issued duration of the current task, then the task suspected of being subject to association constraints is determined as a task subject to association constraints; On the contrary, the task suspected of being subject to association constraints is not a task subject to association constraints.

[0038] Obtain the other task with the greatest association with the task subject to association constraints and record it as the target task; Take the difference between the issuance time interval and the issued duration as the subsequent constraint duration of the task subject to association constraints.

[0039] It should be noted that the issuance time interval between the task subject to association constraints and the target task is the issuance time interval between the two tasks predicted by analyzing the historical data. The issued duration reflects the issued duration of the associated task of the task subject to association constraints. The difference between the issuance time interval and the issued duration can reflect the predicted issuance time length of the associated task. Therefore, by setting the predicted issuance time length behind each task subject to association constraints to reserve some time for the associated task, the interruption of the two associated tasks can be prevented.

[0040] It should be added that the method for obtaining the issuance time interval between the tasks subject to association constraints and other tasks according to the historical processing data of the tasks includes: Record the other task being analyzed as the research task; In the historical processing data of the task, the task that is the same as the task subject to the association constraint is obtained and recorded as the benchmark task, and the task that is the same as the research task is obtained and recorded as the reference task; The time interval between each benchmark task and the nearest reference task is obtained, and the average of the time intervals between all benchmark tasks and the nearest reference task is used as the time interval between the task subject to the association constraint and other tasks.

[0041] S22: Construct a scheduling objective function.

[0042] It's important to note that if the server isn't running while waiting for associated tasks to be delivered, resources will be wasted. Therefore, inserting short-duration tasks while waiting for associated tasks to be delivered can effectively improve resource utilization. Furthermore, since tasks generally have certain priorities, forcing low-priority, short-duration tasks to the front to ensure resource utilization will cause higher-priority tasks to wait longer. Therefore, when scheduling tasks while waiting for associated tasks to be delivered, it's important to consider not only resource utilization but also task priorities.

[0043] Preferably, as an example, constructing a scheduling objective function includes: ; in, represents the subsequent constraint duration of the i-th task subject to the associated constraint, It represents the predicted processing time required for the jth task to be subsequently scheduled on the server where the ith task subject to the association constraint is located. Indicates the preset anti-zero coefficient, Indicates the number of tasks subsequently scheduled on the server where the i-th task subject to the associated constraint is located. represents the number of tasks subject to the association constraint, Indicates the predicted waiting time of the zth subsequent scheduled task, Indicates the priority of the zth subsequent scheduled task, U indicates the number of subsequent scheduled tasks, Indicates the server utilization rate.

[0044] It is understandable that It reflects the difference between the processing time of the tasks scheduled in the process of waiting for the associated tasks to be issued and the subsequent constraint time. The larger the value, the greater the difference between the processing time of the tasks scheduled in the process of waiting for the associated tasks to be issued and the subsequent constraint time. Therefore, the tasks scheduled in the process of waiting for the associated tasks to be issued either have a long running time, resulting in a more serious situation where the associated tasks cannot be processed in a connected manner; or have a short running time, resulting in low resource utilization of the server, and thus the more unreasonable this scheduling method is. It reflects the weighted waiting time weighted by priority. The larger this value is, the longer the waiting time of tasks with higher priorities, indicating that this task arrangement method is more unreasonable. is the occupancy rate of the server. The larger this value is, the higher the utilization rate of the server, indicating that this task arrangement method is more reasonable.

[0045] It should be added that the method for obtaining the predicted processing duration includes: Obtain tasks identical to each task in the historical processing data of the tasks as analysis tasks, and use the average value of the processing durations of all analysis tasks as the predicted processing duration of each task.

[0046] It should be further added that the method for obtaining the predicted waiting duration includes: Obtain the predicted processing duration of each task, obtain the tasks to be processed arranged before each task as the pre - pending tasks, and obtain the predicted waiting duration of each task by adding the cumulative sum of the predicted processing durations of all pre - pending tasks before each task and the predicted remaining processing duration of the current task on the corresponding server. The predicted remaining processing duration is equal to the predicted processing duration of the current task minus the processed duration.

[0047] S3: Maximize the value of the scheduling objective function to obtain the corresponding subsequent task allocation method to achieve task allocation.

[0048] Preferably, as an example, maximizing the value of the scheduling objective function to obtain the corresponding subsequent task allocation method to achieve task allocation includes: Maximize the scheduling objective function, use the optimization solution algorithm to solve the scheduling objective function to obtain the optimal solution, and obtain the task allocation method corresponding to the optimal solution. Allocate servers to each task to be allocated according to the task allocation method corresponding to the optimal solution. Existing optimization solution algorithms can be genetic algorithms or simulated annealing algorithms, and this embodiment does not make specific limitations.

[0049] It can be understood that by maximizing the value of the scheduling objective function, the task arrangement is made the most reasonable, and thus the task arrangement is carried out according to the task arrangement method with the most appropriate task arrangement.

[0050] Before the optimization solution, it is necessary to clarify the solution space of the scheduling objective function. Therefore, the construction method of the solution space will be described below.

[0051] It should be added that the construction method of the solution space is: For any task to be allocated, select a server as the target server according to any selection method, allocate the target server to each task to be allocated, and take the corresponding relationship between all tasks to be allocated and the servers as a feasible solution.

[0052] According to different selection methods, the target servers selected for each task to be allocated are different. By using different selection methods to select target servers, several feasible solutions are obtained, and the space composed of all feasible solutions is used as the solution space.

[0053] An embodiment of the present invention also discloses a cloud computing resource dynamic scheduling optimization system, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a cloud computing resource dynamic scheduling optimization method according to the present invention is implemented.

[0054] The above system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface. Their settings and functions are known in the art, so they will not be described in detail here.

[0055] In the present invention, the aforementioned memory can be any tangible medium that contains or stores a program, and this program can be used by or in combination with an instruction execution system, device, or device. For example, a computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory, a dynamic random access memory, a static random access memory, an enhanced dynamic random access memory, a high-bandwidth memory, a hybrid memory cube, etc., or any other medium that can be used to store the required information and can be accessed by an application program, a module, or both. Any such computer storage medium can be part of the device or accessible or connectable to the device.

Claims

1. A dynamic scheduling optimization method for cloud computing resources, characterized in that Including steps: Obtain the task that each server is currently processing and record it as the current task; According to the historical processing data of the tasks, the correlation between each two tasks and the priority of each task are obtained; according to the correlation between each two tasks, the tasks subject to the correlation constraint are screened out in the current tasks, and the subsequent constraint duration of the tasks subject to the correlation constraint is obtained, and the subsequent constraint duration is the expected interval duration of the tasks with greater correlation than the tasks subject to the correlation constraint; a scheduling objective function is constructed, and the scheduling objective function is negatively correlated with the difference between the required predicted processing duration and the subsequent constraint duration of the subsequent scheduled tasks on the server where the tasks subject to the correlation constraint are located, and is also negatively correlated with the weighted waiting time of all subsequent scheduled tasks; the weighted waiting time is obtained by weighted calculation of the predicted waiting time of all subsequent scheduled tasks using the priority as the weight; The value of the scheduling objective function is maximized to obtain the corresponding subsequent task allocation method to achieve task allocation.

2. The dynamic scheduling optimization method of cloud computing resources according to claim 1, wherein, The construction of the scheduling objective function includes: ； Among them, represents the subsequent constraint duration of the i-th task subject to associated constraints, represents the required predicted processing duration of the j-th task subsequently arranged on the server where the i-th task subject to associated constraints is located, represents a preset anti-zero coefficient, represents the number of tasks subsequently arranged on the server where the i-th task subject to associated constraints is located, represents the number of tasks subject to associated constraints, represents the predicted waiting duration of the z-th subsequently arranged task, represents the priority of the z-th subsequently arranged task, U represents the number of subsequently arranged tasks, represents the occupancy rate of the server.

3. The dynamic scheduling optimization method for cloud computing resources according to claim 1, wherein The step of filtering out tasks subject to association constraints from the current tasks based on the association between each two tasks includes: Obtain the time interval between the task subject to the associated constraint and other tasks based on the task's historical processing data, and obtain the duration of the task subject to the associated constraint; If the correlation between the current task and other tasks is greater than the preset correlation threshold, the current task is determined to be a task suspected of being subject to correlation constraints; If the time interval between the release of a task suspected of being subject to association constraints and other tasks is greater than the release duration of the current task, the task suspected of being subject to association constraints will be determined as a task subject to association constraints; On the contrary, a task suspected of being subject to association constraints is not a task subject to association constraints.

4. A cloud computing resource dynamic scheduling optimization method according to claim 3, characterized in that: The subsequent constraint duration of the task subject to the associated constraint includes: The other tasks with the greatest correlation with the task subject to the correlation constraint are recorded as target tasks; The difference between the release time interval and the released duration is used as the subsequent constraint duration of the task subject to the associated constraint.

5. A dynamic scheduling optimization method for cloud computing resources according to claim 1, characterized in that The step of obtaining the correlation between two tasks based on the historical processing data of the tasks includes: Determine whether there is a data inflow and outflow relationship between the two tasks. If there is a data inflow and outflow relationship, set the basic value of the correlation between the two tasks to 1, otherwise set it to 0; Set a time window and use the Apriori algorithm to calculate the confidence of the association rules between each two tasks based on the data in the time window; The correlation between the two tasks is obtained by multiplying the confidence of the association rule by the correlation basis value.

6. The dynamic scheduling optimization method for cloud computing resources according to claim 1, characterized in that Before maximizing the value of the scheduling objective function and obtaining the corresponding subsequent task allocation method, a solution space must be constructed. The method for constructing the solution space includes: For any task to be assigned, a server is selected as the target server according to any selection method, and a target server is assigned to each task to be assigned. The correspondence between all tasks to be assigned and servers is regarded as a feasible solution, and the space formed by all feasible solutions is regarded as the solution space.

7. A method for optimizing dynamic scheduling of cloud computing resources according to claim 1, characterized in that The method for obtaining the predicted waiting time includes: Obtain the predicted processing duration of each task. The to-be-processed tasks arranged before each task are recorded as pre - to - be - processed tasks. The predicted waiting duration of each task is obtained by adding the accumulated sum of the predicted processing durations of all the pre - to - be - processed tasks before each task and the predicted remaining processing duration of the current task on the corresponding server.

8. A method for optimizing dynamic scheduling of cloud computing resources according to claim 6, characterized in that, The step of obtaining the corresponding subsequent task allocation method by maximizing the value of the scheduling objective function includes: Maximize the scheduling objective function, use an optimization algorithm to solve the optimal solution of the scheduling objective function, and obtain the task allocation method corresponding to the optimal solution.

9. A method for optimizing dynamic scheduling of cloud computing resources according to claim 8, characterized in that, The step of achieving task allocation includes: Allocate servers to each task to be allocated according to the task allocation method corresponding to the optimal solution.

10. A dynamic scheduling optimization system for cloud computing resources, characterized in that, It includes: A processor and a memory, where the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a dynamic scheduling optimization method for cloud computing resources according to any one of claims 1 - 9 is implemented.

Images