A workflow fault-tolerant scheduling method of an end-edge-cloud collaborative system based on reliability driving

Abstract

The application discloses a kind of based on reliability drive's end side cloud coordination system's workflow fault-tolerant scheduling method, comprising: step S1, the relevant information of workflow and each server resource information are read;Step S2, using fault-tolerant scheduling method to each task in workflow is scheduled and resource allocation, obtain scheduling scheme;Step S3, each server creates corresponding number of data storage copy, data transmission copy and task execution copy according to scheduling scheme, and executes workflow according to scheduling scheme.The technical problem solved in the application is that, in the end side cloud coordination computing environment, the fault-tolerant mode of creating data storage copy, data transmission copy and task execution copy enables workflow to be executed with high reliability, obtains efficient scheduling scheme of workflow under the constraints of considering the dependency relationship between tasks of workflow application and deadline, the computing resources of edge server and system total cost, effectively improves the reliability of end side cloud coordination system.

Description

Technical Field

[0001] This invention relates to the field of data transmission and data backup technology in edge-cloud collaborative computing architecture, and specifically to a workflow fault-tolerant scheduling method in an edge-cloud collaborative system. Background Technology

[0002] With the advent of the Internet of Things era, the number and types of mobile terminal access devices have grown exponentially, and massive amounts of data are being collected on terminal devices with heavy computing loads. However, due to physical size limitations, the computing resources and battery capacity of mobile terminals are finite, and increased latency leads to higher energy consumption. Local computing models struggle to meet the low-latency requirements of applications. To address the shortcomings of traditional local processing methods, new computing models have been proposed, giving rise to mobile cloud computing and mobile edge computing technologies. In a mobile cloud computing (MCC) environment, the computing tasks of terminal devices are offloaded to cloud servers with powerful computing capabilities for centralized processing. Mobile edge computing (MEC) pushes computing, originally centralized in the cloud, down to network edge devices, providing computing services closer to the data source. Compared to edge computing, cloud computing has abundant computing resources, and cloud computing latency is generally lower than edge computing, but the transmission distance is farther and the transmission latency is higher. Edge computing, on the other hand, has advantages such as high bandwidth, low latency, and wide coverage, but is limited by network edge resources. For offloading problems under certain complex conditions, a single computing architecture cannot solve the problem. In such cases, an end-edge-cloud collaborative computing architecture can become an effective solution. By combining the abundant computing resources of cloud servers with the low latency of edge servers, a three-tier architecture can provide higher computing and transmission performance than cloud computing or edge computing alone. However, during workflow scheduling, transmission lines, edge servers, and virtual machines are inevitably subject to various failures. Reliability is also an important Quality of Service (QoS) metric. In a collaborative edge-cloud computing environment, introducing fault-tolerant technology based on proactive replication into workflow scheduling can effectively enhance workflow reliability while meeting latency constraints, edge server resource constraints, and total system cost constraints.

[0003] In existing edge-cloud collaborative scheduling work, the optimization objectives are mostly latency and energy consumption, and there are currently no results that consider user reliability in edge-cloud collaborative systems. For example, the paper "Collaborative Cloud-Edge-EndTask Offloading in Mobile-Edge Computing Networks With Limited CommunicationCapability" discloses a task offloading method in an edge-cloud collaborative computing environment. Its goal is to minimize the total latency of all mobile devices under the constraints of mobile device (MD) deadlines and energy consumption by determining the computation offloading strategy, computing resources, delivery rate, and transmit power allocation. The paper "AI-Driven Energy-Efficient Content TaskOffloading in Cloud-Edge-End Cooperation Networks" discloses a task offloading scheme based on deep reinforcement learning (DRL) in an edge-cloud collaborative network environment. By making collaborative caching and task offloading decisions in each time slot based on content request information in previous time slots and the current network state, it minimizes the total system energy consumption. The paper "Collaborative cloud-edge-end task offloading with task dependency based on deep reinforcement learning" discloses a workflow offloading method in edge-cloud collaborative scenarios, which minimizes the average latency and average energy consumption of all IoT devices under the constraints of limited computing resources of IoT devices and multi-core edge servers, as well as task dependencies. None of the methods mentioned above consider fault-tolerant scheduling of workflows in edge-cloud collaborative architectures.

[0004] Existing task scheduling solutions in edge-cloud collaborative computing environments do not fully consider users' reliability requirements for workflow execution, especially regarding potential failures during data storage, data transmission, and task execution. Given constraints such as workflow latency limitations, edge server resource limitations, and overall system cost limitations, successful workflow execution necessitates improved system reliability. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a reliability-driven workflow fault-tolerant scheduling method for edge-cloud collaborative systems. Its purpose is to effectively improve the reliability of workflow execution while ensuring that the partial order relationship between tasks, deadlines, cost constraints, and edge server computing resource constraints are simultaneously satisfied.

[0006] Technical solution:

[0007] A reliability-driven workflow fault-tolerant scheduling method for edge-cloud collaborative systems is proposed. This method uses a fault-tolerant workflow scheduling algorithm that creates data storage replicas, data transmission replicas, and task execution replicas to obtain a replica allocation strategy. Based on the obtained scheduling scheme, data replicas and task replicas are allocated to different servers and virtual machines (VMs), thereby improving the reliability of the entire workflow execution. The main steps of the method include:

[0008] Step S1: Read relevant information about the workflow (including the data volume of each task in the workflow and the average number of computation cycles required per MB of data, the partial order relationship between tasks in the workflow, deadlines, cost constraints, and reliability requirements in data storage, data transmission, and task execution, etc.) and resource information of each server (including the online rate of data storage and the unit price of data storage of each server, the probability of successful data transmission between edge servers and cloud data centers, the bandwidth of data transmission, the unit price of data transmission, the failure rate of each VM on the server, the computation frequency, the unit price of task execution, etc.).

[0009] Step S2: Use a fault-tolerant scheduling method to schedule and allocate resources for each task in the workflow to obtain a scheduling scheme (the scheme includes a deployment decision scheme for data storage copies, data transmission copies, and task execution copies of the task, as well as the minimum total system cost). The fault-tolerant scheduling method includes a task selection method (see step S21) and a copy creation method (see steps S23-S29).

[0010] In step S3, each server creates a corresponding number of data storage replicas, data transmission replicas, and task execution replicas according to the scheduling scheme, and executes the workflow according to the scheduling scheme.

[0011] Step S2, which uses a fault-tolerant scheduling method to schedule and allocate resources for each task in the task sequence to obtain a scheduling scheme, includes the following main steps:

[0012] Step S21: Select the first task in the task pool to be completed as the current task, and parse the current workflow task information;

[0013] Step S22: Determine whether the current task can be executed. The criterion for determining whether the current task can be executed is that all the previous tasks of the current task have been completed.

[0014] Step S23: Initialize the current task and virtual machine pool;

[0015] Step S24: Calculate the data transmission time, transmission overhead and transmission reliability required for the current task to deploy data transmission copies on each server, and determine whether the current task is the starting task. If it is the starting task, the starting data transmission copy is the input data generated on the IoT device that is offloaded to the edge server or first offloaded to the edge server and then transmitted to the cloud data center. Otherwise, the data transmission copy of the current task is the data transmission copy of the predecessor task required to execute on each server.

[0016] Step S25: Calculate the preparation time for deploying task copies on each server for the current task;

[0017] Step S26: Calculate the storage overhead and storage reliability of deploying data storage replicas on each server for the current task;

[0018] Step S27: Calculate the replica deployment budget for the current task;

[0019] Step S28: Determine whether the number of task replicas deployed for the current task exceeds the maximum number of replicas. If it does not exceed the maximum number of replicas, traverse the virtual machine pool, select the VM that meets the task budget constraint and deadline constraint and has the highest reliability to deploy the task execution replicas, and update the virtual machine pool, task budget and task scheduling reliability, and then go to step S28; if it exceeds the maximum number of replicas, the current task scheduling is completed, and go to step S29.

[0020] Step S29: The current task scheduling is completed, and the current task is removed from the task pool to be completed;

[0021] Step S210: Determine whether the task pool to be completed after scheduling is empty. If it is empty, the task scheduling ends and proceeds to step S211; otherwise, proceed to step S21.

[0022] Step S211: Output the scheduling scheme (replica deployment scheme).

[0023] Step S21 involves selecting the first task in the task pool of the workflow to be completed as the current task t. ij Parse the current workflow task information, including but not limited to the status information of all predecessor tasks of the current task, and the task data volume L. ij The average number of computation cycles C required for each bit of data in the task. ij wait;

[0024] In step S21, the task selection method involves the user randomly submitting multiple workflows to the edge-cloud collaborative system. The edge-cloud collaborative system includes a set of |m| terminal mobile devices, MD = {MD1, MD2, ..., MD...}. |m| The set of |b| edge base stations BS = {BS1, BS2, ..., BS} |b|}, and a remote cloud data center DC d , the workflow application set W = {W1, W2,..., W |W|}, and the unfinished tasks are put into the to-be-completed workflow task pool. Each workflow can be described as Each workflow can be described as W i = {T i , E i , D i}, representing the task set, the edge set between tasks, and the deadline requirement of workflow W i respectively. Among them, the task set of the i-th application with n tasks can be represented by ;

[0025] The executable task judgment method in step S22 is to judge whether the current task can be executed. It is necessary to traverse whether all its predecessor tasks have been executed. When all predecessor tasks are completed, the current task can be executed and go to step S23. If the current task has unfinished predecessor tasks, the current task enters the end of the to-be-completed task pool and goes to step S21;

[0026] In step S23, the reliability and cost of the current task are both initialized to 0, and the virtual machine pool is initialized so that the virtual machines of all servers can deploy task execution replicas;

[0027] In step S24, calculate the data transfer time, transfer overhead, and transfer reliability required for the current task to deploy data transfer replicas on each server. The present invention studies the workflow application of a single data source. In the workflow model, only one node is used as the starting node, that is, in workflow W i t i1 is the starting task. Among them, the input data of the task is divided into two types: the input data of the starting task is the data generated by the Internet of Things device, and the non-starting task uses the output data of its direct predecessor task as the input data;

[0028] The starting task t i of workflow W i1 The transmission time of the input data d i1 generated on the Internet of Things device is related to the deployment decision of its data storage replica ; indicates that the unloading of the data replica fails through the base station BS b , indicates that the data replica is unloaded to BS b and the data replica is deployed at BS b ; indicates that the data replica performs task unloading through BS b and transmits data to the cloud data center DC dDeploy data replicas;

[0029] If the data copy in step S24 is deployed at the base station BS b The required data transmission time is calculated according to Shannon's law, and the data offloading rate is... and time Calculated separately as

[0030]

[0031] Among them B m,b θ is the channel bandwidth, g0 is the path loss constant, d0 is the reference distance, d is the actual distance from the user terminal device to the edge server, θ is the path loss exponent, p is the transmission power of the terminal device, and N0 is the noise power spectral density between the terminal device and the edge server.

[0032] Data transmission overhead and data transmission reliability They are respectively represented as

[0033]

[0034] in, and They represent MD respectively m To BS b The unit price of transmission and the probability of successful transmission.

[0035] If the data copy in step S24 is deployed at the base station DC d Its data transmission also includes data from the BS. b To DC d The transmission time, overhead, and reliability are calculated as follows:

[0036]

[0037] Among them, BW b,d , and They represent from BS b To DC d Bandwidth, transmission unit price, and probability of successful transmission;

[0038] Workflow W in step S24 i Non-starting task t ij The data storage is the output data of its direct predecessor task, where the predecessor task... Output data Transmission and task t ij Data storage replication decision The relevant data transmission time, overhead, and probability of successful transmission are expressed as follows:

[0039]

[0040] If any one of the data copies successfully transmits the data, then the transmission of that data is reliable. The transmission reliability is calculated as follows:

[0041]

[0042] Where |dt| represents The total number of deployed data storage replicas;

[0043] Data d ij In server s The overall transport reliability of deploying data replicas is calculated as the product of the transport reliability of all predecessor tasks:

[0044]

[0045] Compute data storage replicas are deployed on the server s Transmission overhead Including the overhead of deploying all predecessor task transport replicas, specifically calculated as follows:

[0046]

[0047] The calculation in step S25 is related to the preparation overhead and preparation time of deploying task copies on each server and the deployment of data transmission copies.

[0048] The initial task t in step S25 i1 In server s Preparation time for execution The calculation is as follows:

[0049]

[0050] Step S25 Non-starting task t ij In server s Preparation time for execution Completion time of previous quest instances Related to the transmission time of the copy:

[0051]

[0052] Step S26 calculates the storage overhead and reliability of deploying data storage replicas on each server in the current task. The data storage overhead is calculated as follows:

[0053]

[0054] in For server s The unit price of storage;

[0055] The reliability of data storage in step S26 is defined as the probability of successful data storage. Data storage reliability is related to the server selected for data replica deployment.

[0056] d ij Data replicas are deployed on edge servers (BS). b In this context, the reliability of data storage is related to the reliability of the edge server. The edge server changes its state over time, and its reliability follows an exponential distribution, decreasing with increasing working time, denoted by λ. b BS b The online rate and data storage reliability are calculated as follows:

[0057]

[0058] d ij Data replicas are deployed on cloud server DC d At that time, the reliability of data storage is related to the reliability of cloud servers, and cloud data center (DC) is used. d Online rate Let represent the probability of its accessibility. Then, the reliability of data storage is expressed as:

[0059]

[0060] Step S27 calculates the replica deployment budget for the current task, which is the budget for the workflow application to ensure that the initial task can be successfully offloaded to the base station. i Divided into minimum unload data budget and disposable budget Right now The task scheduling budget for the current task is calculated as follows:

[0061]

[0062] Among them, Act (Budget) i ) for workflow W i The actual remaining budget, NS(L) i ) for W i The total amount of data for tasks that have not yet been assigned;

[0063] Step S28: Task Execution Copy On the server s The kth virtual machine (VM) s,k Execution time and execution overhead Calculated as

[0064]

[0065] Among them, F s,k and respectively vm s,k The calculation frequency and task execution unit price;

[0066] The reliability calculation of the task execution replica in step S28 is defined as the probability of successful task execution. This invention considers the transient failure of the VM. ij The probability of a transient failure during execution follows a Poisson distribution, denoted by λs. ,k Indicates VM s,k Failure rate, task t ij In VM s,k The reliability calculation for deploying task execution replicas is as follows:

[0067]

[0068] Step S28 updates the state of the VM of the currently deployed execution replica, including updating the idle time slot of the VM and removing the VM of the currently deployed task execution replica from the virtual machine pool of deployable task replicas;

[0069] The overhead of updating the current task scheduling in step S28 includes the overhead of deploying all data transfer replicas, data storage replicas, and task execution replicas;

[0070] t in step S28 ij The reliability of task scheduling is related to the reliability of data transmission replicas, data storage replicas, and task execution replicas. A task replica can only execute successfully if the data replica is successfully transmitted and stored; therefore, task replicas... The probability of a successful scheduling can be expressed as:

[0071]

[0072] The probability of a mission instance scheduling failure is:

[0073]

[0074] This invention uses an active replication scheme to achieve fault tolerance for task execution replicas. ij If any one of the replicas completes successfully, then the task execution has not failed. A task t with |te| task scheduling replicas... ij Reliability calculation is

[0075]

[0076] The output scheduling scheme (replica deployment scheme) of step S211 includes the deployment decision scheme of task data storage replicas, data transmission replicas, task execution replicas, and the reliability of the edge-cloud collaborative system;

[0077] The reliability of the edge-cloud collaborative system in step S211 is the sum of the reliability of all workflow scheduling in the system, and the reliability of a single workflow scheduling is the product of the reliability of all tasks in that workflow.

[0078] Beneficial effects of the present invention

[0079] The technical problem solved by this invention is to enable highly reliable execution of workflows in an edge-cloud collaborative computing environment by creating fault-tolerant methods such as data storage replicas, data transmission replicas, and task execution replicas. Under the constraints of task dependencies and deadlines in workflow applications, edge server computing resources, and total system cost, an efficient workflow scheduling scheme is obtained, which effectively improves the reliability of the edge-cloud collaborative system. Attached Figure Description

[0080] Figure 1 This is a schematic diagram of the overall process of a workflow fault-tolerant scheduling method for an edge-cloud collaborative system in a specific embodiment of the present invention.

[0081] Figure 2 This is a schematic diagram illustrating the specific process of the fault-tolerant scheduling method in a particular embodiment of the present invention. Detailed Implementation

[0082] The following is combined Figure 1 The embodiment is described in detail. In this embodiment, there are two user upload workflow applications. Each user workflow has 3 tasks to be processed. The requirement is to complete the workflow scheduling of these 6 tasks in an edge-cloud collaborative system with 2 edge servers, 1 cloud data center and 9 VMs, and obtain a scheduling scheme with the minimum total system cost.

[0083] The basic information about the task is as follows:

[0084]

[0085] The cost constraint for workflow application #0 is $10.33, and the cost constraint for workflow application #1 is $17.65.

[0086] The basic information of the server is as follows:

[0087]

[0088]

[0089] The transmission cost from the terminal device to the edge server is $0.035 / GB / s, and the transmission cost from the edge server to the cloud data center is $0.5 / GB / s. The reliability information for transmission between servers is as follows:

[0090] Transmission reliability Server 0 Server 1 Server 2 Server 0 0.99 0.98 0.95 Server 1 0.98 0.99 0.91 Server 2 0.95 0.91 0.99

[0091] The basic information of the VM is as follows:

[0092]

[0093] Combination Figure 2 The flowchart shown illustrates a workflow fault-tolerant scheduling method for a reliability-driven edge-cloud collaborative system, which includes the following steps:

[0094] Step S1: Read relevant information about the workflow, including the data volume of each task in the workflow and the average number of computation cycles required per MB of data, the partial order relationship between tasks in the workflow, etc. Read resource information of each server, including the online rate and unit price of data storage of the server, the probability of successful data transmission between servers, the bandwidth of data transmission, the unit price of data transmission, the failure rate, computation frequency, and task execution unit price of each VM on the server, etc. Define the task set as π = {π0, π1, π2, π3, π4, π5}, and establish the relevant scheduling model.

[0095] Step S2: Use a fault-tolerant scheduling method to schedule and allocate resources for each task in the workflow to obtain a scheduling scheme (the scheme includes a deployment decision scheme for data transmission copies, data storage copies, task execution copies, and maximum system reliability). The fault-tolerant scheduling method includes features such as task selection methods and copy creation methods.

[0096] The reliability-driven fault-tolerant scheduling method in step S2 schedules and allocates resources for each task in the workflow to obtain a scheduling scheme, including the following main steps:

[0097] In step S21, the task selection method involves selecting the first task in the task pool of workflow tasks to be completed as the current task and parsing the current workflow task information.

[0098] The executable task determination method in step S22 is based on the criterion that all predecessor tasks of the current task have been completed.

[0099] Step S23: Initialize the current task and virtual machine pool;

[0100] Step S24: Calculate the data transmission time, transmission overhead, and transmission reliability required for the current task to deploy data transmission copies on each server. Determine whether the current task is the starting task. If it is the starting task, the starting data transmission copy is the input data generated on the IoT device that is offloaded to the edge server or first offloaded to the edge server and then transmitted to the cloud data center. Calculate the data transmission time, overhead, and reliability according to Formula 2-7. Otherwise, the data transmission copy of the current task is the data transmission copy of the predecessor task required to execute on each server. Calculate the data transmission reliability and overhead according to Formula 12 and Formula 13 respectively.

[0101] Step S25: Calculate the preparation time for deploying task copies on each server according to Formulas 14 and 15.

[0102] Step S26: Calculate the storage overhead and storage reliability of deploying data storage copies on each server for the current task according to formulas 16-18.

[0103] Step S27: Calculate the replica deployment budget for the current task according to Formula 19;

[0104] Step S28: Determine whether the number of task replicas deployed for the current task exceeds the maximum number of replicas. If it does not exceed the maximum number of replicas, traverse the virtual machine pool, select the VM that meets the task budget constraint and deadline constraint and has the highest reliability to deploy the task execution replicas, and update the virtual machine pool and task budget. Calculate the scheduling reliability of the task according to Formula 25, and then go to step S28. If it exceeds the maximum number of replicas, the current task scheduling is completed, and go to step S29.

[0105] Step S29: The current task scheduling is completed, and the current task is removed from the task pool to be completed;

[0106] Step S210: Determine whether the task pool to be completed after scheduling is empty. If it is empty, the task scheduling ends and proceeds to step S211; otherwise, proceed to step S21.

[0107] Step S211: Output the scheduling scheme (replica deployment scheme and maximum system reliability 1.9706);

[0108] In step S3, each server creates a corresponding number of data storage replicas, data transmission replicas, and task execution replicas according to the scheduling scheme, and executes the workflow according to the scheduling scheme.

[0109] This application proposes a reliability-driven workflow fault-tolerant scheduling method for edge-cloud collaborative systems, ensuring maximum system reliability while meeting the constraints of partial order relationships between tasks, deadlines, cost limitations, and edge server computing resources. Existing task scheduling schemes in edge-cloud collaborative computing environments do not fully consider users' reliability requirements for workflow execution, particularly regarding potential failures during data storage, data transmission, and task execution, which can prevent workflow completion. The method described in this application comprehensively considers the reliability requirements of workflows during data storage, data transmission, and task execution. It not only uses data storage and task execution copies but also creates data transmission copies through proactive replication, thus solving the workflow fault-tolerant scheduling problem in edge-cloud collaborative environments.

[0110] This invention provides a workflow fault-tolerant scheduling method for a reliability-driven edge-cloud collaborative system. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A workflow fault-tolerant scheduling method for a reliability-driven edge-cloud collaborative system, characterized in that... Includes the following steps: Step S1: Read the relevant information of the workflow and the resource information of each server. The relevant information of the workflow includes, but is not limited to, the data volume of each task in the workflow and the average number of computation cycles required per MB of data, the partial order relationship between tasks in the workflow, the deadline constraints, and the reliability requirements in data storage, data transmission, and task execution. The server resource information includes, but is not limited to, the online rate of data storage and the unit price of data storage for each server, the probability of successful data transmission between the edge server and the cloud data center, the bandwidth of data transmission, the unit price of data transmission, the failure rate of each VM on the server, the computation frequency, and the unit price of task execution. Step S2: Use a fault-tolerant scheduling method to schedule and allocate resources for each task in the workflow to obtain a scheduling scheme. The scheduling scheme includes a deployment decision scheme for data storage copies, data transmission copies, and task execution copies of the tasks, as well as the minimum total system cost. The fault-tolerant scheduling method includes a task selection method and a copy creation method. The fault-tolerant scheduling algorithm includes the following steps: Step S21, Task selection method: Select the first task in the task pool to be completed as the current task, and parse the current workflow task information; Step S22, Executable task determination method, determines whether the current task can be executed; Step S23: Initialize the current task and virtual machine pool; Step S24: Calculate the data transmission time, transmission overhead, and transmission reliability required to deploy data transmission copies of the current task on each server. Determine if the current task is the starting task. If it is the starting task, the starting data transmission copy is the input data generated on the IoT device, which is either offloaded to the edge server or first offloaded to the edge server and then transmitted to the cloud data center. Otherwise, the data transmission copy of the current task is the data transmission copy of the preceding task required for execution on each server. In step S24: The input data for tasks is divided into two types: the input data for the starting task of the workflow is the data generated by the IoT device, and the input data for non-starting tasks is the output data of their direct predecessor task. ① If the data storage copy of the initial task is deployed at the edge base station, then data transmission only requires offloading the task to the edge base station. In this case: Data offloading rate to edge base stations Calculate according to the following formula: in, It is the channel bandwidth. It is the path loss constant. For reference distance, θ represents the actual distance from the user terminal device to the edge server, where θ is the path loss exponent. For the transmission power of the terminal device, The noise power spectral density between the terminal device and the edge server; Calculate according to the following formula: in, Represents a data storage copy From mobile devices Offloaded to edge base station Data transmission time, Indicates the amount of data for the current task; and Obtained through the following formula: in, and They represent The unit price of transmission and the probability of successful transmission; Represents a data storage copy From mobile devices Offloaded to edge base station Data transmission overhead; Represents a data storage copy From mobile devices Offloaded to edge base station Data transmission reliability; ② If the data storage copy of the initial task is deployed in the cloud data center, then data transmission requires the task to be offloaded to the edge base station first, and then transmitted to the cloud data center through the edge base station. In this case: in, , and They represent from arrive Bandwidth, transmission unit price, and probability of successful transmission; Represents a data storage copy From mobile devices Offloaded to edge base station Then to cloud data centers Data transmission time; Represents a data storage copy From mobile devices Offloaded to edge base station Then to cloud data centers Data transmission overhead; Represents a data storage copy From mobile devices Offloaded to edge base station Then to cloud data centers Data transmission reliability; ③ Non-starting task The data storage is the output data of all its direct predecessor tasks. In this case, the transmission of the output data of a single predecessor task is related to the data storage copy decision of the current task. If any copy of this data successfully transmits the data, then the transmission of this data is reliable. = in, Indicates data transfer copy By server Transmitted to server Data transmission time, This indicates the transmission overhead of data transfer copies. Indicates the reliability of data transmission copies. , and These represent data transmission bandwidth, transmission unit price, and probability of successful transmission, respectively. data Transmission reliability is obtained by the following formula: in, express The total number of deployed data storage replicas; The total transport reliability of deploying data replicas for non-initial tasks is the product of the transport reliability of all preceding tasks, with the total transport overhead being... The overhead of deploying copies of the transfer for all predecessor tasks is obtained by the following formula: ； Step S25: Calculate the preparation time for deploying task copies on each server for the current task; Step S26: Calculate the storage overhead and storage reliability of deploying data storage replicas on each server for the current task; Step S27: Calculate the replica deployment budget for the current task; Step S28: Determine whether the number of task replicas deployed for the current task exceeds the maximum number of replicas. If it does not exceed the maximum number of replicas, traverse the virtual machine pool, select the VM that meets the task budget constraint and deadline constraint and has the highest reliability to deploy the task execution replicas, and update the virtual machine pool, task budget and task scheduling reliability, and then go to step S28; if it exceeds the maximum number of replicas, the current task scheduling is completed, and go to step S29. Step S29: The current task scheduling is completed, and the current task is removed from the task pool to be completed; Step S210: Determine whether the task pool to be completed after scheduling is empty. If it is empty, the task scheduling ends and proceeds to step S211; otherwise, proceed to step S21. Step S211: Output the scheduling scheme; In step S3, each server creates a corresponding number of data storage replicas, data transmission replicas, and task execution replicas according to the scheduling scheme, and executes the workflow according to the scheduling scheme.

2. The method according to claim 1, characterized in that, The parsing information in step S21 includes, but is not limited to, parsing the current workflow task information and the information of all virtual machines in the virtual machine pool. The workflow task information includes, but is not limited to, the status information of all predecessor tasks of the current task and the data volume of the task. The number of CPU cycles required per MB of data for the task .

3. The method according to claim 1, characterized in that, In step S22, to determine whether the current task can be executed, it is necessary to traverse all its predecessor tasks to see if they have completed preparation. If all predecessor tasks are completed, the current task can be executed and proceed to step S23. If the current task has unfinished predecessor tasks, the current task enters the tail of the task pool and proceeds to step S21.

4. The method according to claim 1, characterized in that, In step S23, the reliability and cost of the current task are initialized to 0, and the virtual machine pool is initialized so that virtual machines on all servers can deploy task execution replicas.

5. The method according to claim 1, characterized in that, In step S25: ① The initial task of the workflow Preparation time for deploying data storage replicas on the server The time taken to transmit input data to this server is obtained using the following formula: in, Indicates that the data copy is unloaded to And in Deploy data replicas, Indicates that the data copy has been passed. Perform task unloading and transfer data to cloud data center. Deploy data replicas; ② Non-starting task Preparation time for deploying data storage replicas on the server Completion time of previous quest instances The transmission time of the transferred copy is related to the transmission time and can be obtained using the following formula: 。 6. The method according to claim 5, characterized in that, In step S26: ① Storage overhead is the product of storage unit price and data volume; ② The reliability of data storage is obtained by the following formula: in, express Data replicas are deployed on edge servers Reliability, express Online rate; ③ The data storage reliability of a cloud data center is obtained by the following formula: in, express Data replicas are deployed in cloud data centers Reliability, Indicates cloud data center Online rate.

7. The method according to claim 1, characterized in that, In step S27, the task scheduling budget for the current task is obtained using the following formula: in, To minimize the data unloading budget, For disposable budget, For workflow The actual remaining budget, for The total amount of data for tasks that have not yet been assigned.

8. The method according to claim 1, characterized in that, In step S28: ①Task Execution Copy On the server The kth virtual machine Execution time and execution overhead The calculation is as follows: in, and They are respectively The calculation frequency and task execution unit price; ② Task execution reliability is defined as the probability of a task successfully executing. The probability of transient failures during execution follows a Poisson distribution. exist Reliability of deployed task execution replicas Calculated using the following formula: in, express The failure rate; ③has Task scheduling copy of the task reliability This can be expressed by the following formula: in, Indicates a mission instance The probability of a successful scheduling is obtained by the following formula: in, and They represent in Deploy data storage replicas Data transmission reliability and data storage reliability, Indicates a mission instance exist Reliability of execution.

Images