A hierarchical service unloading method and system based on data perception

Abstract

The present invention relates to the technical field of layered service unloading, and in particular to a layered service unloading method and system based on data perception, wherein the unloading method comprises the following steps: S1: initializing a cloud-edge collaborative elastic optical network; S2: generating a group of service requests based on the cloud-edge collaborative elastic optical network; S3: dividing the service requests into indivisible services and divisible services; S4: performing resource unloading processing on the indivisible services and the divisible services respectively. The present invention reduces the end-to-end time delay problem in the service request process, improves the network resource utilization efficiency in the cloud-edge collaborative elastic optical network, and reduces the blocking rate.

Description

Technical Field

[0001] The present invention relates to the technical field of layered service unloading, and in particular to a layered service unloading method and system based on data perception. Background Art

[0002] Recently, with the rapid development of the Internet of things (IoTs), a large number of latency-sensitive, computationally intensive applications have emerged, such as smart cities, face recognition, and intelligent transportation systems, which require a large amount of computing resources to meet user demands for low latency and low blocking rate. Cloud-edge collaborative elastic optical networks (CCEONs) are a promising technology to meet the needs of mobile users.

[0003] In this field, some technicians have proposed a dependency-aware cooperative offloading (DA-CO) based on a genetic algorithm (GA) and applied it to urban optical networks. This method can reduce the average response time of different services. In addition, some technicians have introduced the concepts of decision-making services and computing services to reduce the latency of collaborative edge computing according to the processing type of the service. However, no one has combined data perception and dependency perception to optimize the end-to-end latency and network resource allocation of flexible optical networks. Summary of the invention

[0004] To this end, the technical problem to be solved by the present invention is to overcome the problems of end-to-end delay in business processing in the prior art and low efficiency of network resource utilization in the cloud-edge collaborative elastic optical network.

[0005] To solve the above technical problems, the present invention provides a data-aware layered service unloading method, which includes the following steps:

[0006] S1: Initialize the cloud-edge collaborative elastic optical network;

[0007] S2: Generate a set of service requests based on the cloud-edge collaborative elastic optical network;

[0008] S3: Divide the service request into non-dividable services and dividable services;

[0009] S4: Perform resource unloading processing on the indivisible business and the divisible business respectively, including: for the indivisible business, processing the indivisible business at the server node; for the divisible business, dividing the divisible business into at least two sub-businesses, establishing a hierarchical business model based on the sub-businesses, obtaining a target sub-business set for each divisible business, and performing resource unloading on the target sub-business in the target sub-business set.

[0010] In one embodiment of the present invention, the method for processing the indivisible service at the server node includes:

[0011] Determine whether the available computing resources of the local server node are greater than the computing resource requirements of the indivisible service: if so, process the indivisible service at the local server node; otherwise, offload the indivisible service to a server node in an adjacent area of ​​the edge network for processing;

[0012] Determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, offload the indivisible business to the cloud server nodes for processing.

[0013] In one embodiment of the present invention, the method for establishing the hierarchical service model is:

[0014] According to the data flows and dependencies between the sub-businesses, a directed acyclic graph is established for the sub-businesses to obtain an initial business model;

[0015] Simplifying the dependency relationships between the sub-businesses in the initial business model, traversing the directed acyclic graph in the initial business model, determining whether each subtask has a direct dependency relationship and an indirect dependency relationship with the subtask it depends on, and if both exist, simplifying the direct dependency relationship to obtain a business model with simplified dependency relationships;

[0016] The directed acyclic graph in the business model after the dependency relationship is simplified is traversed, and the sub-business with a single continuous dependency and its dependent sub-business are merged to obtain the target business model, that is, the hierarchical business model.

[0017] In one embodiment of the present invention, a method for performing resource unloading on a target sub-service in the target sub-service set includes:

[0018] Classifying the target subtasks in the target sub-business set into calculation subtasks and decision-making subtasks;

[0019] When the computing resources of the local server node are sufficient, the decision-making sub-business in the target sub-business is processed at the local server node, and the computing sub-business in the target sub-business is offloaded to the server node in the adjacent area of ​​the edge network for processing;

[0020] When the computing resources of the local server node are insufficient, but the computing resources of the server node in the adjacent area of ​​the edge network are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the server node in the adjacent area for processing;

[0021] When the server node resources in the adjacent area are insufficient but the computing resources of the cloud server are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the cloud server for processing.

[0022] In one embodiment of the present invention, the cloud-edge collaborative elastic optical network G (SN, EN, OL, ES, DC, B) is composed of a set of conversion nodes SN carrying data centers, a set of edge computing nodes EN carrying edge servers, a set of optical fiber links OL of the network, a set of edge servers ES providing edge computing resources, a set of data centers DC providing cloud computing resources, and a set of base stations B;

[0023] The service request UR includes a source point s for receiving the service request locally, a bandwidth requirement b for the service request, a computing resource requirement c for the service request, a generation time t for the service request, and a division type d for the service request.

[0024] In one embodiment of the present invention, the hierarchical service model ST (PS, SR, SB, PT, SE) is composed of a set of dependent prior sub-services PS of a group of sub-services, a set of computing resource requirements SR of a group of sub-services, a set of bandwidth resource requirements SB of a group of sub-services, a set of processing types PT of a group of sub-services and a set of generation times SE of a group of sub-services.

[0025] In one embodiment of the present invention, the computing sub-service is dependent on at least one sub-service, and the decision-making sub-service is dependent on at least one sub-service.

[0026] Based on the same inventive concept, the present invention also provides a data-aware layered service unloading system, which includes the following modules:

[0027] Network initialization module, used to initialize the cloud-edge collaborative elastic optical network;

[0028] A service request generation module, used to generate a set of service requests based on the cloud-edge collaborative elastic optical network;

[0029] A service request classification module, used for classifying the service request into non-dividable services and dividable services;

[0030] The resource unloading module is used to perform resource unloading processing on the indivisible business and the divisible business respectively, including: for the indivisible business, processing the indivisible business at the server node; for the divisible business, dividing the divisible business into at least two sub-businesses, establishing a hierarchical business model based on the sub-businesses, obtaining a target sub-business set for each divisible business, and performing resource unloading on the target sub-business in the target sub-business set.

[0031] In one embodiment of the present invention, the resource unloading module includes a non-divisible service unloading submodule and a divisible service unloading submodule, and the divisible service unloading submodule includes a service partitioning submodule, a hierarchical service model establishment submodule, a task hierarchical submodule and a service unloading submodule;

[0032] Among them, the indivisible business unloading submodule is used to determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, the indivisible business is unloaded to the server nodes of the cloud network for processing.

[0033] In one embodiment of the present invention, the service division submodule is used to divide the divisible service into at least two sub-services;

[0034] The hierarchical business model establishment submodule is used to establish a directed acyclic graph for the sub-businesses according to the data flows and dependencies between the sub-businesses to obtain an initial business model; simplify the dependencies between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, determine whether each subtask has a direct dependency and an indirect dependency on the subtask it depends on, and if both exist, simplify the direct dependency to obtain a business model with simplified dependencies; traverse the directed acyclic graph in the business model with simplified dependencies, merge a single continuously dependent sub-business and its dependent sub-businesses to obtain a hierarchical business model;

[0035] The task stratification submodule is used to classify the subtasks in the target sub-business set into calculation subtasks and decision-making subtasks;

[0036] The service unloading submodule is used to obtain a target sub-service set of each divisible service based on the layered service model, and to perform resource unloading on the target sub-service in the target sub-service set.

[0037] The above technical solution of the present invention has the following advantages compared with the prior art:

[0038] The present invention addresses the problems of end-to-end delay in business request processing and inefficient network resource utilization in a cloud-edge collaborative elastic optical network. It divides business requests into divisible business and non-divisible business, establishes a business model for divisible business, and layers subtasks to better utilize the computing resources of nodes in the network during business processing, greatly reducing processing delays, thereby ensuring that the average end-to-end delay of business requests can be reduced and improving network service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein

[0040] Figure 1 It is a flowchart of the implementation of the layered service unloading method based on data perception according to the present invention;

[0041] Figure 2 This is the architecture diagram of the cloud-edge collaborative elastic optical network;

[0042] Figure 3 is an architectural diagram of the initial business model described in the present invention;

[0043] Figure 4 It is a schematic diagram of processing the dependency relationship between sub-services described in the present invention. DETAILED DESCRIPTION

[0044] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.

[0045] Embodiment 1

[0046] Reference Figure 1 As shown, the present invention provides a layered service unloading method based on data perception, the method comprising the following steps:

[0047] S1: Initialize the cloud-edge collaborative elastic optical network;

[0048] S2: Generate a set of service requests based on the cloud-edge collaborative elastic optical network;

[0049] S3: Divide the service request into non-dividable services and dividable services;

[0050] S4: Perform resource unloading processing on the indivisible business and the divisible business respectively, including: for the indivisible business, processing the indivisible business at the server node; for the divisible business, dividing the divisible business into at least two sub-businesses, establishing a hierarchical business model based on the sub-businesses, obtaining a target sub-business set for each divisible business, and performing resource unloading on the target sub-business in the target sub-business set.

[0051] In order to achieve the goal of minimizing the time slot resources occupied by a group of service requests, the cloud-edge collaborative elastic optical network G(SN, EN, OL, ES, DC, B) is first initialized, including a set of conversion nodes SN carrying data centers, a set of edge computing nodes EN carrying edge servers, a set of optical fiber links OL of the network, a set of edge servers ES providing edge computing resources, a set of data centers DC providing cloud computing resources and a set of base stations B; secondly, a set of service requests UR is generated, including the source point s for local reception of service requests, the bandwidth requirement b of the service requests, the computing resource requirement c of the service requests, the generation time t of the service requests and the division type d of the service requests; finally, based on the data perception of the service, the service is classified into non-divisible service and divisible service, and the service modeling is performed for the divisible service. For the non-divisible service and the service model of the divisible service, the service resources are unloaded to ensure the optimal latency while achieving low blocking and optimized network resource allocation.

[0052] Specifically, the SN set is written as SN = {sn1, sn2, ..., sn |SN|}, the EN set is written as EN = {en1,en2,…,en |EN|}, the OL set is written as OL = {ol1, ol2, ..., ol |OL|}, ES set is written as ES = {es1, es2, ..., es |ES|}, the DC set is written as DC = {dc1, dc2, ..., dc |DC|}, and the B set is written as B={b1,b2,…,b |B|}. Where |SN|, |EN|, |OL|, |ES|, |DC|, and |B| represent the number of switching nodes, edge computing nodes, fiber links, edge servers, data centers, and base stations, respectively.

[0053] like Figure 2As shown in the figure, the design of the cloud-edge collaborative elastic optical network is a 14-node network divided into two layers, one for the edge computing layer and one for the cloud layer; the computing resources of the edge server of the edge computing layer are set to 80 computing units, the computing resources of the conversion node server are set to 100 computing resources, and the computing resources of the cloud server of the cloud layer are set to infinity, which is not considered here. The spectrum slot resources of each optical fiber link are set to 100.

[0054] In step S2, a group of service requests are generated based on the cloud-edge collaborative elastic optical network. The computing resource requirement of each service request is 2 to 5, the spectrum slot resource requirement is 2 to 6, the source point is randomly generated in the server node of the edge layer, and the number of subtask divisions of each divisible service is set to 3 to 6, and the division ratio is used to determine the spectrum slot resource requirement and computing resource requirement of each subtask.

[0055] In step S3, the service request is divided into an indivisible service and a divisible service. The indivisible service may include a binary service, which is defined as: (1) a highly integrated or relatively simple service that cannot be divided; (2) a service that must be executed as a whole. The divisible service can be divided into at least two sub-services, and the dependency between the sub-services cannot be ignored.

[0056] In S4, for the indivisible service, a method for processing the indivisible service at the server node includes:

[0057] Determine whether the available computing resources of the local server node are greater than the computing resource requirements of the indivisible service: if so, process the indivisible service at the local server node; otherwise, offload the indivisible service to a server node in an adjacent area of ​​the edge network for processing;

[0058] Determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, offload the indivisible business to the cloud server nodes for processing.

[0059] In S4, a hierarchical service model is established based on the sub-services, and the hierarchical service model is constructed by:

[0060] Step a: According to the data flow and dependency relationship between the sub-services, a directed acyclic graph is established for the sub-services to obtain an initial service model, such as Figure 3 shown in Figure 3In the figure, boxes are used to represent sub-services, arrows are used to represent the dependency relationship and data flow between sub-services, and the sub-services are numbered from 0 to 6. The dependency relationship between sub-services is mainly manifested as follows: if sub-service i depends on sub-service k, then the input data flow of sub-service i can include the output data flow of sub-service k, and the dependency relationship between sub-service k and sub-service i is represented by an arrow, which is an arrow starting from sub-service k and pointing to sub-service i.

[0061] Step b: Simplify the dependencies between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, and determine whether each subtask has a direct dependency and an indirect dependency on the subtask it depends on. If both exist, simplify the direct dependency, such as Figure 4 As shown in the figure, it is reflected in the business model, that is, the arrows directly pointing between the sub-business and the sub-business it depends on are erased, and the business model with simplified dependency relationships is obtained; Figure 4 As shown, "S1" represents the dependency simplification step, the business model in the middle is the business model after the dependency simplification, and the dependency that can be simplified is marked with a circle;

[0062] Step c: traverse the directed acyclic graph in the business model after the dependency relationship is simplified, and merge the sub-business with a single continuous dependency and its dependent sub-business, such as Figure 4 As shown, "S2" represents the dependency merging step, and the dotted box represents at least two sub-businesses that can be merged to obtain the target business model, that is, the hierarchical business model.

[0063] The hierarchical service model ST (PS, SR, SB, PT, SE) is composed of a set of previous sub-services PS of which the sub-services are dependent, a set of computing resource requirements SR of the sub-services, a set of bandwidth resource requirements SB of the sub-services, a set of processing types PT of the sub-services, and a set of generation time SE of the sub-services. Specifically, the PS set is written as PS = {ps1, ps2, ..., ps |ST|}, the SR set is written as SR = {sr1,sr2,…,sr |ST|}, the SB set is written as SB = {sb1,sb2,…,sb |ST|}, the PT set is written as PT = {pt1, pt2,…, pt |ST|}, the SE set is written as SE = {se1,se2,…,se |ST|}. Among them, |ST| represents the total number of sub-businesses.

[0064] Step S4 also includes: based on the hierarchical service model, obtaining a target sub-service set of each divisible service, and performing resource unloading on the target sub-service in the target sub-service set. The method for performing resource unloading on the target sub-service includes the following steps:

[0065] The target subtasks in the target sub-business set are classified into computing subtasks and decision-making subtasks; the computing sub-business is dependent on at least one sub-business, and the decision-making sub-business is dependent on at least one sub-business.

[0066] When the computing resources of the local server node are sufficient, the decision-making sub-services in at least two target sub-services are processed at the local server node, and the computing sub-services in the target sub-services are offloaded to the server node in the adjacent area of ​​the edge network for processing;

[0067] When the computing resources of the local server node are insufficient, but the computing resources of the server node in the adjacent area of ​​the edge network are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the server node in the adjacent area for processing;

[0068] When the server node resources in the adjacent area are insufficient but the computing resources of the cloud server are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the cloud server for processing.

[0069] Embodiment 2

[0070] Based on the same inventive concept as the data-aware layered service unloading method described in Embodiment 1, the present invention further provides a data-aware layered service unloading system, which includes the following modules:

[0071] Network initialization module, used to initialize the cloud-edge collaborative elastic optical network;

[0072] A service request generation module, configured to generate a set of service requests based on the cloud-edge collaborative elastic optical network;

[0073] A service request classification module, used for classifying the service request into non-dividable services and dividable services;

[0074] The resource unloading module is used to perform resource unloading processing on the indivisible business and the divisible business respectively, including: for the indivisible business, processing the indivisible business at the server node; for the divisible business, dividing the divisible business into at least two sub-businesses, establishing a hierarchical business model based on the sub-businesses, obtaining a target sub-business set for each divisible business, and performing resource unloading on the target sub-business in the target sub-business set.

[0075] In this embodiment, the resource unloading module includes a non-divisible service unloading submodule and a divisible service unloading submodule, and the divisible service unloading submodule includes a service division submodule, a hierarchical service model establishment submodule, a task hierarchical submodule and a service unloading submodule;

[0076] Among them, the indivisible business unloading submodule is used to determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, the indivisible business is unloaded to the server nodes of the cloud network for processing.

[0077] In this embodiment, the service division submodule is used to divide the divisible service into at least two sub-services;

[0078] The hierarchical business model establishment submodule is used to establish a directed acyclic graph for the sub-businesses according to the data flows and dependencies between the sub-businesses to obtain an initial business model; simplify the dependencies between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, determine whether each subtask has a direct dependency and an indirect dependency on the subtask it depends on, and if both exist, simplify the direct dependency to obtain a business model with simplified dependencies; traverse the directed acyclic graph in the business model with simplified dependencies, merge a single continuously dependent sub-business and its dependent sub-businesses to obtain a hierarchical business model;

[0079] The task hierarchical submodule is used to classify the target subtasks in the hierarchical business model into calculation subtasks and decision-making subtasks;

[0080] The service unloading submodule is used to obtain a target sub-service set of each divisible service based on the layered service model, and to perform resource unloading on the target sub-service in the target sub-service set.

[0081] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0082] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0083] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0084] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0085] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A data-aware layered service unloading method, characterized in that: include: S1: Initialize the cloud-edge collaborative elastic optical network; S2: Generate a set of service requests based on the cloud-edge collaborative elastic optical network; S3: Divide the service request into an indivisible service and a divisible service; wherein the indivisible service includes a binary service, and the binary service meets two conditions: first, it is a highly integrated service that cannot be divided; second, it is a service that must be executed as a whole; and the divisible service can be divided into at least two sub-services; S4: performing resource unloading processing on the non-divisible service and the divisible service respectively, including: for the non-divisible service, processing the non-divisible service at the server node; for the divisible service, dividing the divisible service into at least two sub-services, establishing a hierarchical service model based on the sub-services, obtaining a target sub-service set of each divisible service, and performing resource unloading on the target sub-service in the target sub-service set; Wherein, in S4, the method for establishing the hierarchical service model is: According to the data flows and dependencies between the sub-businesses, a directed acyclic graph is established for the sub-businesses to obtain an initial business model; Simplify the dependency relationships between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, determine whether each subtask has a direct dependency relationship and an indirect dependency relationship with the subtask it depends on, and if both exist, simplify the direct dependency relationship to obtain a business model with simplified dependency relationships; The directed acyclic graph in the business model after the dependency relationship is simplified is traversed, and the sub-business with a single continuous dependency and its dependent sub-business are merged to obtain the target business model, that is, the hierarchical business model.

2. The data-aware layered service offloading method according to claim 1, characterized in that: The method for processing the indivisible service at the server node includes: Determine whether the available computing resources of the local server node are greater than the computing resource requirements of the indivisible service: if so, process the indivisible service at the local server node; otherwise, offload the indivisible service to a server node in an adjacent area of ​​the edge network for processing; Determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, offload the indivisible business to the cloud server nodes for processing.

3. The data-aware layered service offloading method according to claim 1, characterized in that: The method for performing resource unloading on a target sub-service in the target sub-service set includes: Classifying the target subtasks in the target sub-business set into calculation subtasks and decision-making subtasks; When the computing resources of the local server node are sufficient, the decision-making sub-business in the target sub-business is processed at the local server node, and the computing sub-business in the target sub-business is offloaded to the server node in the adjacent area of ​​the edge network for processing; When the computing resources of the local server node are insufficient, but the computing resources of the server node in the adjacent area of ​​the edge network are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the server node in the adjacent area for processing; When the server node resources in the adjacent area are insufficient but the computing resources of the cloud server are sufficient, both the decision-making sub-business and the computing sub-business in the target sub-business are offloaded to the cloud server for processing.

4. The data-aware layered service offloading method according to claim 1, characterized in that: The cloud-edge collaborative elastic optical network G (SN, EN, OL, ES, DC, B) is composed of a set of conversion nodes SN carrying data centers, a set of edge computing nodes EN carrying edge servers, a set of optical fiber links OL of the network, a set of edge servers ES providing edge computing resources, a set of data centers DC providing cloud computing resources, and a set of base stations B; The service request UR includes a source point s for receiving the service request locally, a bandwidth requirement b for the service request, a computing resource requirement c for the service request, a generation time t for the service request, and a division type d for the service request.

5. The data-aware layered service offloading method according to claim 1, characterized in that: The hierarchical service model ST (PS, SR, SB, PT, SE) is composed of a set of prior sub-services PS on which a group of sub-services depend, a set of computing resource requirements SR of a group of sub-services, a set of bandwidth resource requirements SB of a group of sub-services, a set of processing types PT of a group of sub-services and a set of generation times SE of a group of sub-services.

6. The data-aware layered service offloading method according to claim 3, characterized in that: The computing sub-service is dependent on at least one sub-service, and the decision-making sub-service is dependent on at least one sub-service.

7. A data-aware layered service offloading system, characterized in that: Used to implement the data-aware layered service unloading method according to any one of claims 1 to 6, the system comprises: Network initialization module, used to initialize the cloud-edge collaborative elastic optical network; A service request generation module, configured to generate a set of service requests based on the cloud-edge collaborative elastic optical network; A service request classification module, used for classifying the service request into indivisible services and divisible services; wherein the indivisible services include binary services, and the binary services meet two conditions: first, highly integrated services that cannot be divided; second, services that must be executed as a whole; and the divisible services can be divided into at least two sub-services; The resource unloading module is used to perform resource unloading processing on the non-divisible service and the divisible service respectively, including: for the non-divisible service, processing the non-divisible service at the server node; for the divisible service, dividing the divisible service into at least two sub-services, establishing a hierarchical service model based on the sub-services, obtaining a target sub-service set for each divisible service, and performing resource unloading on the target sub-service in the target sub-service set; Wherein, establishing the layered service model includes: According to the data flows and dependencies between the sub-businesses, a directed acyclic graph is established for the sub-businesses to obtain an initial business model; Simplify the dependency relationships between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, determine whether each subtask has a direct dependency relationship and an indirect dependency relationship with the subtask it depends on, and if both exist, simplify the direct dependency relationship to obtain a business model with simplified dependency relationships; The directed acyclic graph in the business model after the dependency relationship is simplified is traversed, and the sub-business with a single continuous dependency and its dependent sub-business are merged to obtain the target business model, that is, the hierarchical business model.

8. The data-aware layered service offloading system according to claim 7, characterized in that: The resource unloading module includes a non-dividable service unloading submodule and a dividable service unloading submodule, and the dividable service unloading submodule includes a service division submodule, a hierarchical service model establishment submodule, a task hierarchical submodule and a service unloading submodule; Among them, the indivisible business unloading submodule is used to determine whether the available computing resources of the server nodes in the adjacent area of ​​the edge network are greater than the computing resource requirements of the indivisible business: if so, the server nodes in the adjacent area of ​​the edge network process the indivisible business; otherwise, the indivisible business is unloaded to the server nodes of the cloud network for processing.

9. The data-aware layered service offloading system according to claim 8, characterized in that: The service division submodule is used to divide the divisible service into at least two sub-services; The hierarchical business model establishment submodule is used to establish a directed acyclic graph for the sub-businesses according to the data flows and dependencies between the sub-businesses to obtain an initial business model; simplify the dependencies between the sub-businesses in the initial business model, traverse the directed acyclic graph in the initial business model, determine whether each subtask has a direct dependency and an indirect dependency on the subtask it depends on, and if both exist, simplify the direct dependency to obtain a business model with simplified dependencies; traverse the directed acyclic graph in the business model with simplified dependencies, merge a single continuously dependent sub-business and its dependent sub-businesses to obtain a hierarchical business model; The task hierarchical submodule is used to classify the subtasks in the hierarchical business model into computational subtasks and decision-making subtasks; The service unloading submodule is used to obtain a target sub-service set of each divisible service based on the layered service model, and to perform resource unloading on the target sub-service in the target sub-service set.

Images