Service chain deployment method and device for cloud-edge collaborative Internet of Things

Abstract

The present invention provides a service chain deployment method and device for cloud-edge collaborative Internet of Things, the method comprising: after obtaining the first construction scheme of SFC according to the preset arrangement order and preset dependency relationship of VNF in the service chain SFC to be deployed, determining the VNF with parallel function requirements in SFC according to the business type of VNF, and performing parallel processing on some VNFs in SFC to form a second construction scheme; merging some virtual nodes in the second construction schemes of multiple different SFCs to obtain a comprehensive construction scheme; establishing a service chain deployment optimization model of the comprehensive construction scheme, obtaining the deployment scheme of each virtual node and each virtual link in the comprehensive construction scheme according to the optimal solution of the service chain deployment optimization model, and deploying each virtual node and each virtual link on the cloud-edge collaborative network according to the deployment scheme. The present invention reduces the service chain delay and improves the quality, reliability and stability of the network.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular to a service chain deployment method and device for a cloud-edge collaborative Internet of Things. Background Art

[0002] With the development of 5G and cloud computing, data on the edge side is growing rapidly and needs to be transmitted to the data center network for processing. The traditional cloud-centric computing model of the central network can support these services with its powerful computing power and abundant resources, but it will cause a huge load on the core network and bring a large end-to-end delay to the service. In order to meet this challenge, MEC (Mobile edge computing) technology was proposed to reduce the overall delay by providing computing power at the edge of the network. However, the limited resources and computing power of the edge server cannot completely replace the role of cloud computing. Therefore, it is very necessary to effectively combine the advantages of cloud computing and edge computing and build a CEC (Cloud-edge-collaborative) network architecture that effectively utilizes cloud and edge resources.

[0003] However, under the CEC architecture, cloud-edge network resources are difficult to be integrated and / or managed. Therefore, the CECIoT (Cloud-edge-collaborative Internet of things) architecture is proposed. Under the CECIoT architecture, by mapping the VNF (Virtual Network Function) in the SFC (Service Function Chain) to the edge server, cloud server and communication link of the CEC network, and allocating corresponding network resources to the SFC, the integrated management and control of cloud-edge computing and communication resources can be achieved.

[0004] At present, in the CECIoT architecture, the deployment process of the service chain is mainly simplified by merging homogeneous services. However, this deployment method can only merge completely identical service chains. In actual applications, multiple service chains have different structures and functions, and there are rarely completely identical service chains. Therefore, it is difficult to obtain a reasonable service chain deployment solution using this method, and it is also difficult to effectively reduce the end-to-end delay and resource consumption costs in the SFC mapping process. Summary of the invention

[0005] The present invention provides a service chain deployment method and device for a cloud-edge collaborative Internet of Things, which is used to solve the defect that the prior art can only merge completely identical service chains, making it difficult to obtain a reasonable service chain deployment plan, and further unable to reduce the end-to-end delay and resource consumption costs in the service chain mapping process. The method and device can realize reasonable deployment of the service chain and reduce the end-to-end delay and resource consumption costs in the service chain mapping process.

[0006] The present invention provides a service chain deployment method for cloud-edge collaborative Internet of Things, comprising:

[0007] Obtaining a first construction plan for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each virtual network function VNF in each service chain to be deployed;

[0008] According to the service type of each VNF, determine the VNFs with parallel function requirements in each service chain to be deployed, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain the second construction scheme of each service chain to be deployed;

[0009] If there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, then at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged to obtain a comprehensive construction scheme;

[0010] Establish a service chain deployment optimization model for the comprehensive construction plan, obtain the deployment plan for each virtual node and each virtual link in the comprehensive construction plan according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on the physical node of the cloud-edge collaborative network and deploy each virtual link on the physical link of the cloud-edge collaborative network according to the deployment plan.

[0011] According to a service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention, the VNFs with parallel function requirements in each service chain to be deployed are determined according to the service type of each VNF, and the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed are processed in parallel to obtain the second construction scheme of each service chain to be deployed, including:

[0012] For any VNF, if the service type of the VNF and the service type of the next VNF ​​adjacent to the VNF are both monitor types, it is determined that the VNF has a parallel function requirement;

[0013] If there is a dependency relationship between any two VNFs with parallel functional requirements in each service chain to be deployed, the any two VNFs are serially connected according to the dependency relationship, and the two serially connected VNFs are used as new VNFs with parallel functional requirements;

[0014] The new VNFs with parallel function requirements in the first construction plan of each service chain to be deployed are processed in parallel to obtain a second construction plan of each service chain to be deployed.

[0015] According to a service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention, if there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged to obtain a comprehensive construction scheme, including:

[0016] If there are at least two virtual nodes in the second construction scheme of different service chains to be deployed, both of which carry the same network function type VNF and have an instance sharing function, then the number of VNFs carried by the at least two virtual nodes is calculated;

[0017] If the number of VNFs carried by the at least two virtual nodes is less than or equal to a first preset value, the at least two virtual nodes are merged.

[0018] According to a service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention, if there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, then at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged, including:

[0019] Determine the service type of each service chain to be deployed according to the maximum tolerable delay index of each service chain to be deployed; wherein the service type of the service chain to be deployed includes delay-sensitive services and non-delay-sensitive services;

[0020] The service chain to be deployed, whose service type is non-delay-sensitive service, is used as the target service chain;

[0021] If there are at least two virtual nodes in the second construction scheme of different target service chains, both of which carry the same network function type VNF and have instance sharing function, the at least two virtual nodes are merged.

[0022] According to a service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention, the service chain deployment optimization model for establishing the comprehensive construction solution includes:

[0023] Calculate the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed in the comprehensive construction solution, and use the product as the reliability index of each service chain to be deployed;

[0024] Backing up the VNF with the smallest reliability index in the service chain to be deployed whose reliability index is less than or equal to the second preset value;

[0025] After updating the reliability index of the virtual node corresponding to the backed-up VNF, recalculating the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed until the reliability index of each service chain to be deployed is greater than the second preset value;

[0026] According to the comprehensive construction plan after the last backup, the service chain deployment optimization model is established.

[0027] According to a service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention, the service chain deployment optimization model for establishing the comprehensive construction solution includes:

[0028] According to the attribute information of each physical node and each physical link in the cloud-edge collaborative network and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme are obtained;

[0029] Constructing the service chain deployment optimization model according to the total load rate and the total delay time;

[0030] Among them, the attribute information of the physical node includes the total CPU resources, the remaining CPU resources, the processed data packet size and the processing capacity; the attribute information of the physical link includes the total bandwidth resources, the remaining bandwidth resources, the propagation delay time and the processed data packet size; the attribute information of the virtual node includes the CPU resources required for deployment; the attribute information of the virtual link includes the bandwidth resources required for deployment.

[0031] According to a service chain deployment method for a cloud-edge collaborative Internet of Things provided by the present invention, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme are obtained according to the attribute information of each physical node and each physical link in the cloud-edge collaborative network and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, including:

[0032] Obtaining a first load rate of each physical node according to the total CPU resources and the remaining CPU resources of each physical node of the cloud-edge collaborative network;

[0033] Obtaining a second load rate of each physical node according to the total CPU resources of each physical node and the CPU resources required for deploying the virtual nodes carried on each physical node;

[0034] Obtaining a first load rate of each physical link according to the total bandwidth resources and the remaining bandwidth resources of each physical link of the cloud-edge collaborative network;

[0035] Obtaining a second load rate of each physical link according to the total bandwidth resources of each physical link and the bandwidth resources required for the deployment of the virtual links carried on each physical link;

[0036] Obtaining the total load rate according to the first load rate and the second load rate of all physical nodes in the cloud-edge collaborative network and the first load rate and the second load rate of all physical links in the cloud-edge collaborative network;

[0037] According to the propagation delay time of each physical link, the propagation delay time of each service chain to be deployed is obtained;

[0038] According to the data packet size processed by each physical link and the bandwidth resources required for the deployment of each virtual link, the transmission delay time of each service chain to be deployed is obtained;

[0039] According to the data packet size and processing capacity processed by each physical node, as well as the CPU resources required for the deployment of each virtual node, the processing delay time of each service chain to be deployed is obtained;

[0040] The total delay time is obtained according to the propagation delay time, the transmission delay time, the processing delay time and the queuing delay time of all the service chains to be deployed in the comprehensive construction solution.

[0041] The present invention also provides a service chain deployment device for cloud-edge collaborative Internet of Things, including:

[0042] The first construction module is used to obtain a first construction scheme for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each virtual network function VNF in each service chain to be deployed;

[0043] The second construction module is used to determine the VNFs with parallel function requirements in each service chain to be deployed according to the service type of each VNF, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain a second construction scheme of each service chain to be deployed;

[0044] A third construction module is configured to merge at least two virtual nodes in the second construction schemes of different service chains to be deployed to obtain a comprehensive construction scheme if at least two virtual nodes in the second construction schemes of different service chains to be deployed both carry VNFs of the same network function type and have an instance sharing function;

[0045] A deployment module is used to establish a service chain deployment optimization model for the comprehensive construction plan, obtain the deployment plan of each virtual node and each virtual link in the comprehensive construction plan according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on the physical node of the cloud-edge collaborative network and deploy each virtual link on the physical link of the cloud-edge collaborative network according to the deployment plan.

[0046] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, the steps of the service chain deployment method for cloud-edge collaborative Internet of Things as described in any one of the above-mentioned methods are implemented.

[0047] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the service chain deployment method for cloud-edge collaborative Internet of Things as described in any of the above are implemented.

[0048] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned service chain deployment methods for cloud-edge collaborative Internet of Things.

[0049] The service chain deployment method and device for cloud-edge collaborative Internet of Things provided by the present invention divide each service chain to be deployed into multiple NVFs on the one hand, and sequentially arrange and parallelize the VNFs in each service chain to be deployed according to the preset arrangement order, preset dependency and business type of each VNF in each service chain to be deployed, so as to form a second construction plan for each service chain to be deployed, and merge the virtual nodes carrying VNFs of the same network function type and having instance sharing functions in the second construction plans of multiple service chains to be deployed, so as to realize the merging of heterogeneous service chains, so that the construction plan of the service chain to be deployed is more reasonable, and the virtual nodes can be effectively reduced, and the end-to-end delay and resource consumption cost in the mapping process of the service chain to be deployed can be reduced; on the other hand, a service chain deployment optimization model is constructed with the goal of minimizing the total load rate of the cloud-edge collaborative network and minimizing the delay time of all service chains to be deployed in the comprehensive construction plan, and according to the optimal solution of the service chain deployment optimization model, each virtual node and each virtual link in the comprehensive construction plan are reasonably deployed, which can effectively reduce the service delay of the tasks to be processed and improve the reliability and stability of the cloud-edge collaborative network service. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 This is one of the flow diagrams of the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0052] Figure 2 It is a structural schematic diagram of a first construction scheme of a service chain in a service chain deployment method for a cloud-edge collaborative Internet of Things provided by the present invention;

[0053] Figure 3 It is a structural schematic diagram of a second construction scheme of a service chain in a service chain deployment method for a cloud-edge collaborative Internet of Things provided by the present invention;

[0054] Figure 4 It is a structural schematic diagram of a comprehensive construction solution of a service chain in a service chain deployment method for a cloud-edge collaborative Internet of Things provided by the present invention;

[0055] Figure 5 It is a structural schematic diagram of a comprehensive construction scheme of a backed-up service chain in a service chain deployment method for a cloud-edge collaborative Internet of Things provided by the present invention;

[0056] Figure 6 It is a structural diagram of service chain construction in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0057] Figure 7 It is one of the structural schematic diagrams of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0058] Figure 8 This is the second structural diagram of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0059] Fig. 9 This is the third structural diagram of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0060] Fig.10 This is the fourth structural diagram of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0061] Fig.11 This is the fifth structural diagram of the simulation result in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0062] Fig.12 This is the sixth structural diagram of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0063] Fig.13 This is the seventh structural diagram of the simulation results in the service chain deployment method for cloud-edge collaborative Internet of Things provided by the present invention;

[0064] Fig.14 It is a structural schematic diagram of a service chain deployment device for cloud-edge collaborative Internet of Things provided by the present invention;

[0065] Fig.15 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] At present, some scholars have proposed a service chain deployment method. This deployment method improves the resource utility of service chain deployment by sharing VNF instances to achieve the superposition of network functions, while reducing the resource consumption of the backup instance process, greatly improving the availability and resource utilization of NFV. However, this method aims to find a balance between availability and resource utility. Each deployment only considers the deployment of a single SFC in the request, without integrating multiple requests and multiple SFCs in the request for deployment, which is prone to cause excessive local load.

[0068] In addition, some scholars have proposed a service function chain deployment method, which can be applied to the SFC deployment of task requests in edge computing network scenarios, aiming to reduce the execution time overhead of each SFC and minimize the total service time to complete the task set in edge computing network scenarios. By reducing the time overhead of executing SFC for each task request, the total service time to complete the task set in edge computing network scenarios is minimized. However, this method only emphasizes reducing business latency without comprehensively considering the load size that the edge network can carry, and also ignores the sensitivity of different requests to latency, which may lead to unnecessary network overhead in order to shorten latency.

[0069] Some scholars have also proposed a method for merging and deploying service function chains. When merging service function chain requests, the service function chain requests are classified according to whether they are homogeneous service function chains, and then homogeneous service function chain requests are merged. This method simplifies the service chain deployment process by merging homogeneous services and optimizes resource consumption and reduces the blocking rate during the deployment process. However, this method does not flexibly consider the merging of heterogeneous service chains. It requires exactly the same service chains to be merged. It is difficult to obtain a reasonable service chain deployment plan, and it is difficult to effectively reduce the end-to-end delay and cost during the SFC mapping process.

[0070] On the premise of ensuring the reliability and resource requirements of IoT terminals for services, this embodiment provides a reasonable and effective graphic combination method for multiple service chains according to different business needs, forming a SFG (Service Function Graph) that can perform the same function, that is, a comprehensive construction plan. The processing delay of the service chain is shortened by parallel processing and merging of service chains, and then the VNF is backed up based on instance sharing to form a final comprehensive construction plan to improve the software reliability of the service chain. Finally, a service chain deployment optimization model containing an objective function and multiple constraints is constructed; and the optimization model is solved based on the taboo search heuristic algorithm to deploy the service function graphs of multiple service chains to be deployed on CECIoT, thereby realizing the merging of heterogeneous service chains, improving the service quality and reliability of the cloud-edge collaborative network, and reducing resource costs.

[0071] In order to achieve the above purpose, the following steps are mainly included:

[0072] 1. In the face of different needs of IoT services under the cloud-edge collaborative architecture, multiple service chains to be deployed SFC are merged to form a comprehensive construction solution. First, the service chain to be deployed is divided into multiple subtasks, namely VNFs, for processing; then, according to the dependencies between VNFs, some VNFs in the heterogeneous SFC are parallelized to reduce end-to-end processing delays; then, multiple SFCs are aggregated into PSFG (Parallelized SFG) to coordinate the resource consumption of multiple SFCs;

[0073] 2. In order to improve the software reliability of VNF, first, some VNFs in PSFG are backed up to reduce resource consumption in backup; then, the constructed service chain deployment optimization model is optimized and solved through heuristic algorithms to deploy SFG on CECIoT, and then the communication and computing resources of cloud nodes and edge nodes are comprehensively utilized to jointly optimize the QoS (Quality of service) and resource consumption cost of the cloud-edge collaborative network.

[0074] Combine the following Figure 1 The service chain deployment method for cloud-edge collaborative Internet of Things of the present invention is described, and the method includes: step 101, according to the preset arrangement order and preset dependency relationship of each virtual network function VNF in each service chain to be deployed, obtaining a first construction plan for each service chain to be deployed;

[0075] Optionally, the execution subject of this embodiment is a device with computing function such as a processor.

[0076] When the service chain needs to be built and deployed, the terminal sends a request message to the processor; wherein the request message includes one or more service chains to be deployed, which can be expressed as SR = {SFC1, SFC2, ..., SFC |SR|}, where |SR| is the number of service chains to be deployed in the request information.

[0077] Each service chain to be deployed contains multiple VNFs, and the specific number can be set according to actual needs. s The attribute information can be used with the tuple SFC s = <VF s , rdelay s , P s , source s )express.

[0078] Among them, VF s SFC is the service chain to be deployed s The set of all VNFs contained in VF s = {v1, v2, ..., v |VFs|}. For the i-th VNF, that is, v i , its attribute information can be expressed as (type i , rcpu i , rband i , r i , dep i ), type i Indicates v i Business types, including rectifier type (type i =1) and monitor type (type i =0). Among them, the monitor type VNF can be processed in parallel; usually each SFC consists of a preset number and order of VNFs, and each VNF can perform a specific type of network function and requires some computing resources to support its operation, rcpu i v i CPU (Central Processing Unit) resources required for deployment; rband i vi When deploying, v i Precursor VNF to v i Required link bandwidth resources; r i v i reliability; because VNFs are organized in a predefined order in SFC and process data flows in that order, multiple VNFs in SFC have dependencies; dep i Indicates v i The dependency relationship, that is, v i Needs to be processed after the functions it depends on.

[0079] rdelay s SFC is the service chain to be deployed s The maximum tolerable delay index P s SFC is the service chain to be deployed s The set of packet sizes processed by each VNF in The packet size may change after each VNF processing; source s Indicates the service chain SFC to be deployed s The starting node deployed when accessing the cloud-edge collaborative network.

[0080] Optionally, after receiving the request information sent by the terminal, the processor parses the request information to obtain the preset arrangement order and preset dependency relationship of each VNF in each service chain to be deployed;

[0081] Then, all VNFs in each service chain to be deployed are arranged according to the preset arrangement order and preset dependency of each VNF in each service chain to be deployed to obtain the first construction plan of each service chain to be deployed.

[0082] like Figure 2 As shown in the figure, the service chain SFC to be deployed contains 4 VNFs, and the traffic of SFC will pass through VNF1, VNF2, VNF3 and VNF4 to complete the service. And VNF2, VNF3 and VNF4 are all dependent on VNF1, that is, the traffic input to the three must be processed by VNF1. VNF1, VNF2, VNF3 and VNF4 can be connected in sequence according to the preset arrangement order.

[0083] Step 102: Determine the VNFs with parallel function requirements in each service chain to be deployed according to the service type of each VNF, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain a second construction scheme of each service chain to be deployed;

[0084] Optionally, since the service type of each VNF can be a rectifier type or a monitor type, when the service type of the VNF is a rectifier type, the data traffic can be read and modified; when the service type of the VNF is a monitor type, the VNF only reads and monitors the traffic data and does not modify the traffic data; therefore, there is a parallel function requirement for the rectifier type VNF.

[0085] The VNFs with parallel function requirements in the first construction plan of each service chain to be deployed can be processed in parallel to obtain a second construction plan for each service chain to be deployed.

[0086] Optionally, the parallel processing method can convert a main chain in the first construction scheme into a second construction scheme consisting of a main chain and multiple branch chains.

[0087] like Figure 3 As shown, it is assumed that VNF1 and VNF4 in SFC are rectifier type, VNF2 and VNF3 are monitor type; and VNF2, VNF3 and VNF4 are all dependent on VNF1; in order to shorten the end-to-end delay of the entire SFC, the first construction scheme of sequential connection can be parallelized into a second construction scheme with a main chain and multiple branch chains, that is, a directed acyclic graph.

[0088] Optionally, each branch chain is merged into the main chain at a rectifier type or the exit of the main chain. It should be noted that for each VNF on the main chain after parallel processing, there may be multiple predecessor VNFs; in this case, the VNF will not immediately process the incoming data packet until all predecessor VNFs have completed processing the data packet.

[0089] When the first construction scheme is converted into a second construction scheme consisting of a main chain and multiple branch chains, not only will each data packet of each data stream propagate along the main chain and be processed by the virtual nodes on the main chain, but a copy of the data packet will also be forwarded to each branch chain and processed by the virtual nodes on the branch chain.

[0090] It should be noted that during the parallelization process, it is necessary to ensure that the order of the rectifier-type VNFs in the service chain to be deployed and the dependencies between all VNFs are not changed.

[0091] like Figure 3 As shown, Figure 2 The second construction scheme of the service chain to be deployed is formed by parallel processing of the service chain to be deployed in sequence in the first construction scheme. Figure 2 In the example, VNF1, VNF2, VNF3, and VNF4 are executed sequentially. The total service delay includes the processing delay of the four VNFs, the transmission delay caused by VNF1, VNF2, and VNF3 sending data, and the three-hop propagation delay.

[0092] Figure 3 After the data flow is processed by VNF1, the parallelized service chain to be deployed can be forwarded to VNF2 and VNF3 at the same time. Since VNF2 and VNF3 are not rectifier-type VNFs, the traffic passing through them will not be modified, and VNF4 can successfully obtain the traffic data processed by VNF1. The total delay of the original service is shortened to the maximum delay of the two parallel service chains from VNF1 to VNF2 to VNF4 and from VNF1 to VNF3 to VNF4. The path with a larger delay can be used as the target link of the service chain to be deployed, that is, the set of all VNFs on the main chain of the service chain to be deployed can be expressed as The latency of the main chain is the latency of the service chain to be deployed. The latency of the service chain to be deployed after parallel processing is significantly smaller than Figure 2 The latency of the unparallelized service chain to be deployed.

[0093] Step 103: If there are at least two virtual nodes in the second construction schemes of different service chains to be deployed that both carry VNFs of the same network function type and have an instance sharing function, merge the at least two virtual nodes in the second construction schemes of the different service chains to be deployed to obtain a comprehensive construction scheme;

[0094] The virtual node is used to carry the network function of the VNF in the service chain to be deployed. The instance sharing function is used to characterize that the virtual node can be used to carry the network function of multiple VNFs.

[0095] Optionally, for the service chain SFC to be deployed s The set of virtual nodes that each VNF passes through on the main chain can be expressed as The set of virtual links connecting virtual nodes on the main chain is in, Can write Indicates that the two ends of the virtual link are connected to virtual nodes and

[0096] In order to reduce resource consumption, some virtual nodes in the second construction schemes of multiple different service chains to be deployed may be merged to form a comprehensive construction scheme of multiple different service chains to be deployed, so as to minimize the number of virtual nodes in the comprehensive construction scheme.

[0097] It should be noted that in the process of merging some virtual nodes in the second construction scheme, it is also necessary to ensure that the order of the rectifier-type VNFs on each service chain to be deployed is not changed, and to ensure that the dependencies between all VNFs are not changed.

[0098] Optionally, after obtaining the second construction schemes of all the service chains to be deployed in the request information, it can be determined whether there are at least two virtual nodes in the second construction schemes of the multiple service chains to be deployed that both carry VNFs of the same network function type and have an instance sharing function;

[0099] If so, merge at least two virtual nodes so that all VNFs carried on at least two virtual nodes are deployed on the same virtual node.

[0100] After merging the virtual nodes, a comprehensive construction plan is formed; the comprehensive construction plan can be directly used as the final comprehensive construction plan for multiple service chains to be deployed; or after backing up some VNFs in the parallel service function diagram formed after the merger, the final comprehensive construction plan is formed, which is not specifically limited in this embodiment.

[0101] Step 104, establish a service chain deployment optimization model for the comprehensive construction plan, obtain the deployment plan of each virtual node and each virtual link in the comprehensive construction plan according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on the physical node of the cloud-edge collaborative network and deploy each virtual link on the physical link on the cloud-edge collaborative network according to the deployment plan.

[0102] Optionally, the attribute information set of virtual nodes and virtual links in the comprehensive construction scheme is represented as G V (N V ,L V ),in A collection of virtual nodes to be deployed in the comprehensive construction solution, each of which carries one or more VNF instances with the same network function type;

[0103] To simplify the deployment problem, it can be assumed that each virtual node only carries one VNF of the network function type, and the mapping between virtual nodes and physical nodes corresponds one to one, that is, one virtual node is deployed on each physical node. s ,The virtual node sequence corresponding to the VNF sequence that constitutes SFC can be expressed as

[0104] Virtual Node The attribute information can be expressed as Among them, v k For virtual nodes The VNFs carried; For virtual nodes CPU resources required for deployment to physical nodes; For virtual nodes The sum of CPU resources required for the deployment of all VNFs carried by the SR. Because in the same request process, the service chains to be deployed through each virtual node are different. j Through the virtual node The set of service chains SFC to be deployed. j is a subset of SR, that is The set of service chains to be deployed to which the VNF carried on the virtual node belongs is The size of the data packet processed by the SFC carried by the virtual node is the sum of the sizes of data packets processed by all VNFs carried by the virtual node, that is, the size of the data packet processed by the SFC s = <VF s ,rdelay s ,P s ,source s > Calculate acquisition. For virtual nodes The VNF instance v k Software reliability, that is, virtual nodes reliability.

[0105] is a collection of virtual links to be deployed; The two virtual nodes connected by this link are and Virtual Link Required bandwidth resources Virtual Link Virtual Node and virtual nodes Required link bandwidth resources.

[0106] Optionally, after determining the comprehensive construction plan of all the service chains to be deployed in the request information, the service chains to be deployed are specifically deployed according to the comprehensive construction plan. First, based on the attribute information of each VNF in the comprehensive construction plan and the attribute information of the cloud-edge collaborative network, a service chain deployment optimization model is established with the total load rate of the cloud-edge collaborative network and the total delay time of all the service chains to be deployed in the comprehensive construction plan as the minimum.

[0107] Then, an optimization algorithm is used to optimize and solve the service chain deployment optimization model. The optimization algorithm may be a taboo search algorithm, etc., which is not specifically limited in this embodiment.

[0108] Since the taboo search algorithm has the characteristic of avoiding falling into a local optimum, in order to quickly and accurately obtain the optimal solution of the service chain deployment optimization model, this embodiment uses the taboo search algorithm to solve the service chain deployment optimization model.

[0109] It should be noted that all solutions in the service chain deployment optimization model are subject to multiple constraints in the service chain deployment optimization model.

[0110] The solution of the service chain deployment optimization model is the deployment location of each virtual node and each virtual link in the comprehensive construction solution; in the solution process, each solution component is continuously updated, that is, the physical node selected by each virtual node on the cloud-edge collaborative network is updated. k The update formula can be expressed as (n k )→(n k' ), where n k' is the kth solution component after update.

[0111] The taboo table in the taboo search algorithm is used to indicate that restoration to the original state is prohibited, that is, all selected solutions must be put into the taboo table to avoid selecting duplicate solutions and improve the efficiency of solving the problem.

[0112] For the amnesty rule in the taboo search algorithm, if S * is the current optimal solution. When a forbidden neighbor solution S satisfies goal(S)≤goal(S * ), the banned solution S is pardoned. Among them, goal(S * ) is called the amnesty value; where goal(·) is the objective function of the service chain deployment optimization model.

[0113] For the termination criterion in the tabu search algorithm, when the number of iterations reaches the maximum allowed number of iterations, the tabu search algorithm terminates.

[0114] The brief steps of optimizing and solving the service chain deployment optimization model using the taboo search algorithm include:

[0115] Step 1: Initialization; specifically, importing the attribute information of the base network of the cloud-edge collaborative network, and comprehensively constructing the attribute information set G of the virtual nodes and virtual links in the solution. V (N V ,L V ), attribute information of the service chain to be deployed in the original request information; initialize the taboo table, set the parameters of the taboo search algorithm, such as the maximum number of iterations max_iter=x; randomly generate an initial solution s0, and initialize the current number of iterations iter=0;

[0116] Step 2: Check the constraints. Specifically, the initial solution is judged according to the constraints in the service chain deployment optimization model. If the constraints are met, go to step 3. Otherwise, the deployment fails.

[0117] Step 3, checking the termination criteria; specifically, if the termination criteria are met, outputting the current optimal solution and terminating the algorithm, otherwise proceeding to step 4;

[0118] Step 4: The epsilon-greedy algorithm explores the neighborhood solutions of the current deployment solution and selects candidate solutions. Specifically, first, a set of neighborhood solutions of the current solution is generated, and then several neighborhood solutions that meet the constraints are selected as candidate solutions;

[0119] Step 5, checking the amnesty criteria; specifically, if the candidate solution meets the amnesty criteria, the candidate solution that meets the best state of the amnesty criteria is used as the current solution, and the taboo table and the amnesty value are updated, and the current solution is used to update the current optimal solution, and then the process goes to step 7, otherwise it goes to step 6;

[0120] Step 6, checking the taboo table; specifically, determining the taboo attributes of the candidate solutions, taking the best non-taboo candidate solution as the current solution, and updating the taboo table;

[0121] Step 7, updating the number of iterations; specifically, updating the number of iterations to iter=iter+1, and proceeding to step 3;

[0122] When deploying each request, there are one or more comprehensive construction solutions to be deployed in the SFG set. In summary, each time a comprehensive construction solution is deployed, a service chain deployment optimization model needs to be built and the service chain deployment optimization model needs to be optimized and solved.

[0123] The specific steps of optimizing and solving the service chain deployment optimization model using the taboo search algorithm include:

[0124] Step 1, initialize the taboo table, randomly generate the initial solution s0, and initialize the current iteration number iter=0;

[0125] Step 2, check G V (N V ,L V ) in each virtual node Randomly select a physical node n k As the kth component of the solution S; if the current virtual node Predecessor virtual node If deployed on a cloud node, the currently selected physical node is the cloud node;

[0126] Step 3, Check All service chains to be deployed in the original request information deployed on the network are used to determine the bandwidth resources required for the virtual link;

[0127] Step 4, Check Go to the node that has been solved The virtual links that exist between them constitute the currently deployed virtual link set

[0128] Step 5: For the virtual link set Each virtual link in The physical nodes corresponding to the virtual nodes that have been solved are known, and the shortest path is used to find The path of the underlying network mapped to the cloud-edge collaborative network, that is, the sequence set of physical nodes; if G V (N V ,L V ) If there are no deployed virtual nodes, jump to step 2, otherwise jump to step 6;

[0129] Step 6: If If yes, jump to step 7, otherwise jump to step 8; where rd.random() is a function that generates a random decimal between 0 and 1;

[0130] Step 7: For each component n of the solution S k Update, that is (n k )→(n k' ) and solve the deployment path of the corresponding virtual link. k' Select by traversing all physical nodes so that goal(S')<goal(S), and update the current solution to S=S'; update the taboo table and jump to step 9;

[0131] Step 8: Randomly select a component n of the solution k , generating n k A candidate set; each element in the candidate set is required to be non-taboo or pardon; randomly select n k' , respectively update the current solution component and the current solution as (n k )→(n k' ) and S=S', update the taboo table;

[0132] Step 9, if goal(S)<goal(S * ), then S*=S. Determine whether the number of iterations has not reached the maximum number of iterations; if not, jump to step 6; if so, jump to step 10;

[0133] Step 10, output the optimal solution S*.

[0134] Optionally, when the optimal solution is obtained, the deployment scheme of each virtual node and each virtual link in the comprehensive construction scheme can be obtained according to the optimal solution. The deployment scheme is to deploy each virtual node on a specific physical node of the cloud-edge collaborative network, and to deploy each virtual link on a specific physical link of the cloud-edge collaborative network;

[0135] Finally, each virtual node and each virtual link are deployed on the cloud-edge collaborative network according to the deployment plan.

[0136] On the one hand, this embodiment divides each service chain to be deployed into multiple NVFs, and according to the preset arrangement order, preset dependency and business type of each VNF in each service chain to be deployed, sequentially arranges and parallelizes the VNFs in each service chain to be deployed to form a second construction plan for each service chain to be deployed, and merges the virtual nodes that carry VNFs of the same network function type and have instance sharing functions in the second construction plans of multiple service chains to be deployed, so as to realize the merging of heterogeneous service chains, so that the construction plan of the service chain to be deployed is more reasonable, and can effectively reduce virtual nodes, reduce end-to-end delay and resource consumption costs in the mapping process of the service chain to be deployed; on the other hand, a service chain deployment optimization model is constructed with the goal of minimizing the total load rate of the cloud-edge collaborative network and minimizing the delay time of all service chains to be deployed in the comprehensive construction plan, and according to the optimal solution of the service chain deployment optimization model, each virtual node and each virtual link in the comprehensive construction plan are reasonably deployed, which can effectively reduce the service delay of the tasks to be processed and improve the reliability and stability of the cloud-edge collaborative network service.

[0137] On the basis of the above-mentioned embodiments, the present embodiment determines the VNFs with parallel functional requirements in each service chain to be deployed according to the service type of each VNF, and performs parallel processing on the VNFs with parallel functional requirements in the first construction scheme of each service chain to be deployed to obtain the second construction scheme of each service chain to be deployed, including: for any VNF, if the service type of the VNF and the service type of the next VNF ​​adjacent to the VNF are both monitor types, then it is determined that the VNF has parallel functional requirements; if there is a dependency relationship between any two VNFs with parallel functional requirements in each service chain to be deployed, then the any two VNFs are serially connected according to the dependency relationship, and the two serially connected VNFs are used as new VNFs with parallel functional requirements; the new VNFs with parallel functional requirements in the first construction scheme of each service chain to be deployed are parallelized to obtain the second construction scheme of each service chain to be deployed.

[0138] Optionally, the step of obtaining a second construction plan for each service chain to be deployed includes:

[0139] Step 1: Input the attribute information SFC of each service chain to be deployed in the first construction scheme formed by sequential connection s =〈VF s ,rdelay s ,P s ,source s >And the preset arrangement order and preset dependency relationship of each VNF; among them, VFs ={v1,v2,...,v |VFs|},v i =(type i ,rcpu i ,rband i ,r i ,dep i ) and initialize the service chain SFC to be deployed s The main chain M of SFC s The first VNF ​​is added as the first element a to the main chain M, the set U of monitor-type VNFs is initialized, and the set B of branch chains is initialized; where a is used to track the last VNF ​​in the main chain, and U tracks the monitor-type VNFs that have not been built in the service chain to be deployed;

[0140] Step 2: Service chain SFC to be deployed s VNF instance v in i and v i Is the next VNF ​​of each VNF a monitor type? If so, determine v i If parallel function is required, jump to step 3, otherwise jump to step 4;

[0141] Step 3: Create a branch chain b with a starting point a, then B = B∪{b}, U = U∪v i ; Jump to step 5;

[0142] Step 4: Set v i Connect to the main chain M and update a to a=v i ;

[0143] Step 5: Check each VNF in U. If v i If there is a dependent VNF, jump to step 6, otherwise jump to step 7;

[0144] Step 6: Connect the dependent VNF ​​to v i , and delete the dependent VNF ​​from U; if there are still VNFs that have not been built in U, jump to step 5, otherwise jump to step 7;

[0145] Step 7: If the service chain SFC is to be deployed s If there are still unbuilt VNFs, jump to step 2; otherwise, jump to step 8;

[0146] Step 8: Connect a to the service chain SFC to be deployed s The next node of a in the first construction scheme of ;

[0147] Step 9, check the remaining VNFs in U and connect the remaining VNFs to the next node of a;

[0148] Step 10: Return the main chain M and the branch chain set B, and obtain the service chain SFC to be deployed consisting of the main chain M and the branch chain set B. s The second construction plan.

[0149] This embodiment adopts the above steps to obtain the second construction plan of each service chain to be deployed, with SFC s As input, the second construction scheme consisting of the main chain M and the branch chain set B is output. U is used to track the monitor VNF that is not built in the first construction scheme of the service chain to be deployed, and a is used to track the last VNF ​​in the main chain. Initialize the main chain M and use SFC s The first VNF ​​is used as the first element of the main chain. Then, the service chain SFC to be deployed is traversed in a loop. s For each current VNF ​​instance v i , if v i If it is a monitor type, a new branch chain is created, which starts from the last VNF ​​in the main chain (i.e., a) and points to v i , add this branch chain to B, and add v i Add to U. If v i If it is a rectifier VNF, it is attached to the main chain M and a is updated to v i Then, search for v in U i For each dependent monitor VNF, there is a branch chain that currently ends at the dependent monitor VNF; then, the dependent monitor VNF is connected to the v i , thereby extending the branch chain to v i , and remove the dependent monitor VNF from U. By generating a parallel second construction scheme for the service chain to be deployed in this way, the end-to-end delay time of the service chain to be deployed can be shortened as much as possible while ensuring that the order between the rectifier VNFs and the dependency relationship between the VNFs in the service chain to be deployed are not changed.

[0150] On the basis of the above embodiments, in this embodiment, if there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have instance sharing functions, then at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged to obtain a comprehensive construction scheme, including: if there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have instance sharing functions, then the number of VNFs carried by the at least two virtual nodes is calculated; if the number of VNFs carried by the at least two virtual nodes is less than or equal to a first preset value, then the at least two virtual nodes are merged.

[0151] Optionally, the step of obtaining a comprehensive construction plan of multiple different service chains to be deployed includes:

[0152] Step 1: Input the attribute information set of the virtual nodes and virtual links that each VNF passes through in the second construction scheme of the service chain to be deployed And the attribute information set G of the virtual nodes and virtual links passed by each VNF in the comprehensive construction solution obtained historically V (N V ,L V );and n p Initialize the service chain SFC to be deployed s The first virtual node of the network; wherein the historically acquired comprehensive construction scheme may be arbitrarily selected from the second construction schemes of multiple service chains to be deployed, or may be pre-merged and acquired from the second construction schemes of some of the multiple service chains to be deployed; this embodiment does not specifically limit this.

[0153] Step 2, for Each of Find G V (N V ,L V ) has been carried and If there is a virtual node with VNF of the same network function type and instance sharing function, jump to step 3; otherwise, jump to step 4;

[0154] Step 3: The virtual node found is represented as n C , n C The number of times the instance is shared plus 1, that is, n C The number of carried VNFs is increased by 1, and the process goes to step 5.

[0155] Step 4: Add a new virtual node

[0156] Step 5: Set virtual node n C With n p Connect and update n p n p =n c ;

[0157] Step 6: If you want to deploy the service chain SFC s There are unchecked Jump to step 2, otherwise jump to step 7;

[0158] Step 7: According to the preset order and preset dependencies, p Connect to the virtual node in the second construction scheme of the service chain to be deployed to obtain the updated GV (N V ,L V ).

[0159] It should be noted that, in order to simplify the complexity of deployment during the merging process, it is assumed that virtual links of the same request information will not cross each other, and different virtual links will not pass through the same virtual node in one deployment.

[0160] In this embodiment, the second construction schemes of multiple service chains to be deployed are combined into a comprehensive construction scheme. V (N V ,L V ) and a second construction plan of an unprocessed service chain to be deployed are used as the input for the next comprehensive construction plan update, and the two are combined to update G V (N V ,L V ) as output. By checking each virtual node in the second construction scheme of the service chain to be deployed in turn If G V (N V ,L V ) already exists with If a virtual node with the same VNF network function and the number of VNF instances carried by the virtual node does not reach the preset number, that is, the maximum number of instance sharing, the current virtual node Merge with the found virtual node. If there is no virtual node that meets the conditions, add a new virtual node in the second construction plan of the service chain to be deployed. n c Connect to the second construction plan of the service chain to be deployed and maintain the original VNF order and dependency. By merging some virtual nodes in the second construction plans of multiple service chains to be deployed, the number of virtual nodes in the formed comprehensive construction plan is reduced, effectively saving resource costs.

[0161] like Figure 4 As shown, in the second construction scheme of the two service chains to be deployed, there are virtual nodes including VNF1, VNF2, and VNF3 that carry VNFs of the same network function type and have instance sharing functions. All VNF1s can be deployed on the same virtual node, all VNF2s can be deployed on the same virtual node, and all VNF3s can be deployed on the same virtual node.

[0162] In the prior art, in order to reduce resource consumption during redundant backup, multiple identical service chains to be deployed are usually merged. However, this merger is only applicable to scenarios where multiple service chains to be deployed are identical, and has poor applicability, and it is difficult to achieve the purpose of reducing resource consumption.

[0163] In this embodiment, the virtual nodes corresponding to different service chains to be deployed share the same type of VNF instances or service chain sub-segments to reduce the number of virtual nodes deployed. Under the premise that the number of VNF instances carried by the virtual node does not exceed the maximum number of instance sharing, VNFs with the same network functions on different service chains to be deployed are combined into the same virtual node as much as possible. By parallelizing the processing of the service chains to be deployed and combining multiple service chains to be deployed, a more reasonable comprehensive construction plan can be obtained.

[0164] On the basis of the above embodiments, in this implementation, if there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNF of the same network function type and have an instance sharing function, then at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged, including: determining the service type of each service chain to be deployed according to the maximum tolerant delay indicator of each service chain to be deployed; wherein the service type of the service chain to be deployed includes delay-sensitive services and non-delay-sensitive services; taking the service chain to be deployed whose service type is non-delay-sensitive services as the target service chain; if there are at least two virtual nodes in the second construction schemes of different target service chains, both of which carry VNF of the same network function type and have an instance sharing function, then the at least two virtual nodes are merged.

[0165] Optionally, since the combination of multiple different service chains to be deployed may lead to an increase in virtual links, additional bandwidth resource consumption may be generated; therefore, in order to simplify the complexity of deployment, it is limited to assume that virtual links of the same request will not cross, and different virtual links will not pass through the same physical node in one deployment. With the increase of virtual links, the number of hops between virtual nodes may increase, which in turn generates additional bandwidth resource consumption, increasing the end-to-end delay of the service chain to be deployed. Therefore, it is determined whether to merge the service chain to be deployed based on its business type.

[0166] In order to maximize the network performance and service quality of the cloud-edge collaborative network, this example first determines the service type of the service chain to be deployed, and only merges the virtual nodes in the second construction scheme of the service chain to be deployed whose service type is non-delay-sensitive services.

[0167] Optionally, if the maximum tolerant delay index of any service chain to be deployed is less than a set value, the service chain to be deployed is determined to be a delay-sensitive service; if it is not less than the set value, the service chain to be deployed is determined to be a non-delay-sensitive service; wherein the set value can be set according to actual needs.

[0168] During the deployment phase, the virtual nodes corresponding to the service chains to be deployed for latency-sensitive services tend to be deployed on the edge. The virtual nodes corresponding to the service chains to be deployed for non-latency-sensitive computing-intensive services tend to be deployed on the cloud platform. Latency-sensitive services tend to have short cycles and high request frequencies; latency-insensitive services, on the contrary, have low request frequencies and long request intervals.

[0169] Due to the service chain SFC to be deployed for delay-sensitive services s = <VF s ,rdelay s ,P s ,source s >, its maximum tolerable delay index rdelay s If the virtual nodes are merged, it is difficult to meet the delay requirements of the service chain to be deployed, and it is difficult to ensure the network performance and service quality of the cloud-edge collaborative network. Therefore, the virtual nodes in the second construction scheme of the service chain to be deployed are not combined.

[0170] Optionally, since the request information SR={SFC1, SFC2, ..., SFC |SR| All service chains to be deployed contained in} may be combined into multiple service chain sets. After the delay-sensitive service chains to be deployed in the same request information are processed in parallel, they will not be merged into virtual nodes, but will be directly used as the second construction plan of the service chain to be deployed as the final comprehensive construction plan; the non-delay-sensitive service chains to be deployed in the same request information will be parallelized and then merged into virtual nodes to form the final comprehensive construction plan.

[0171] On the basis of the above embodiments, a service chain deployment optimization model of a comprehensive construction plan is established in this embodiment, including: establishing a service chain deployment optimization model of a comprehensive construction plan, including: calculating the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed in the comprehensive construction plan, and using the product as the reliability index of each service chain to be deployed; backing up the VNF with the smallest reliability index in the service chain to be deployed whose reliability index is less than or equal to the second preset value; after updating the reliability index of the virtual node corresponding to the backed-up VNF, recalculating the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed, until the reliability index of each service chain to be deployed is greater than the second preset value; establishing a service chain deployment optimization model according to the comprehensive construction plan after the last backup.

[0172] Optionally, the VNF sequence of the sth service chain to be deployed in the request information SR can be written as Each VNF has a corresponding software reliability index, namely v i =(type i,rcpu i ,rband i ,r i ,dep i ) i If any network function on the service chain to be deployed is unavailable, the entire service on the service chain to be deployed will fail. Therefore, the product of the reliability indicators of all VNFs on the service chain to be deployed can be used as the reliability indicator of the service chain to be deployed. For any service chain to be deployed, the VNF sequence that constitutes the service chain to be deployed can correspond to a virtual node sequence. Therefore, the product of the reliability indicators of the virtual nodes corresponding to all VNFs in the service chain to be deployed can be used as the SFC of the service chain to be deployed. s Reliability s , the formula is:

[0173]

[0174] in, Virtual nodes in the service chain to be deployed Reliability index, VF s is the number of VNFs included in the sth service chain to be deployed.

[0175] After obtaining the reliability index of each service chain to be deployed, determining whether the reliability index of each service chain to be deployed is less than or equal to a second preset value;

[0176] If it is less than or equal to the second preset value, the reliability of the service chain to be deployed is poor, and the VNF with the lowest reliability on the service chain to be deployed is backed up, that is, a new virtual node is established to carry the backup VNF. When the virtual node carrying the VNF fails, it is considered that the current virtual node has lost the function of processing the VNF, and the new virtual node is used to carry the current function of the VNF. The comprehensive construction plan after the backup is used as the final comprehensive construction plan.

[0177] like Figure 5 As shown in the figure, the service chains SFC A and SFC B to be deployed both need to back up an instance of VNF1 to ensure its reliability, that is, a total of two VNF1 instances need to be backed up. The service chains SFC A and SFC B to be deployed are merged into SFG C. By jointly backing up a VNF1 instance, it can ensure that both SFC A and SFC B meet the software reliability requirements, effectively reducing the resource consumption generated by the backup and reducing the network load.

[0178] Optionally, the backed up virtual node The reliability index of the new backup virtual node will be updated. The newly added backup virtual node has the same connection relationship as the source virtual node. Therefore, the reliability index of the source virtual node is updated to:

[0179]

[0180] Among them, backuptimes is the number of VNFs backed up by the source virtual node; the reliability index of the source virtual node is recalculated after each update until the reliability index of each service chain to be deployed is greater than the preset value lsreliability, and the formula is:

[0181]

[0182] In addition, the initial reliability index may be subtracted from the final reliability index of each service chain to be deployed to obtain the degree of improvement of the reliability index of each service chain to be deployed.

[0183] Then, based on the comprehensive construction plan after the last backup, a service chain deployment optimization model is established to obtain a more reasonable and reliable deployment plan.

[0184] In summary, the main steps to determine the final comprehensive construction plan for multiple target service chains include:

[0185] Step 1: after sorting the VNFs in order according to the preset arrangement order and preset dependency of each VNF in each service chain to be deployed, the VNFs of the service chain to be deployed are processed in parallel according to the service type of the VNF, and a second construction plan for each service chain to be deployed is obtained;

[0186] Step 2: determining the service type of each service chain to be deployed according to the maximum tolerable delay index of each service chain to be deployed; selectively merging the virtual nodes in the second construction scheme of some service chains to be deployed for different service types to form a comprehensive construction scheme;

[0187] Step 3: Check the reliability index of each service chain to be deployed in the comprehensive construction plan, perform redundant backup on the service chain to be deployed that does not meet the reliability requirements, and form a final comprehensive construction plan.

[0188] The specific steps to determine the final comprehensive construction plan for multiple target service chains include:

[0189] Step 1: Input request information SR = {SFC1, SFC2, ..., SFC |SR|}; Initialize the set G of the comprehensive construction scheme PBKSFG to be empty; Initialize the second construction scheme set p-set to be empty;

[0190] Step 2: for each service chain to be deployed, obtain a second construction scheme of the service chain to be deployed, and add it to the second construction scheme set p-set;

[0191] Step 3: Use the second construction scheme of the first service chain to be deployed in the second construction scheme set to PBKSFG set G V (N V , L V ) to initialize;

[0192] Step 4: determine whether the service chain to be deployed after parallel processing is a non-delay-sensitive service; if so, jump to step 5; otherwise, jump to step 6;

[0193] Step 5: merge multiple parallelized service chains to be deployed to form a comprehensive construction plan; if there are unchecked parallelized service chains to be deployed, return to step 4; otherwise, jump to step 7;

[0194] Step 6, directly add the second construction scheme of the service chain to be deployed to the set G; if there is an unchecked parallelized service chain to be deployed, return to step 4, otherwise jump to step 8;

[0195] Step 7, adding the comprehensive construction scheme to set G;

[0196] Step 8: Check the reliability index of each service chain to be deployed in the comprehensive construction plan. If any reliability index is less than or equal to the set value, jump to step 9; otherwise, jump to step 10.

[0197] Step 9: Redundant backup is performed on the service chain to be deployed whose reliability index is less than or equal to the set value; specifically, the virtual node corresponding to the VNF with the lowest reliability in the service chain to be deployed is found and the virtual node is backed up. Return to step 8;

[0198] Step 10, returning the PBKSFG set G, which includes the attribute information of the virtual node set and the attribute information of the virtual link set.

[0199] Optionally, in this embodiment, the second construction plan set p-set is first initialized to be empty. A parallel processing algorithm is performed on each service chain to be deployed in the service chain set to be deployed, and the obtained parallel post-processing second construction plan is added to the second construction plan set p-set. For non-delay-sensitive service chains to be deployed, they are combined to obtain a comprehensive construction plan. The specific steps include initializing the set G of comprehensive construction plans PBKSFG as the first service chain to be deployed in the second construction plan set p-set. For each remaining service chain to be deployed in p-set, they are combined and added to the set G of comprehensive construction plans.

[0200] Through the above method, after each request information processing is completed, a set G of comprehensive construction plans will be generated; each service chain to be deployed in the set is a single delay-sensitive business, and the second construction plan of the service chain to be deployed is directly used as the comprehensive construction plan; or, for non-delay-sensitive computing-intensive businesses, the second construction plans of multiple service chains to be deployed are combined to form a comprehensive construction plan.

[0201] Finally, evaluate the reliability index of each service chain to be deployed, and perform redundant backup for the service chains to be deployed that do not meet the reliability requirements; the specific steps are to find the virtual node corresponding to the VNF with the lowest reliability in the service chain to be deployed, add a new virtual node to the comprehensive construction plan, and carry the VNF instance with the same network function as the lowest VNF. The connection method of the newly added virtual node is consistent with the source node, and the bandwidth request of the newly added virtual link is consistent with the corresponding bandwidth request of the source virtual link. At the same time, update the comprehensive construction plan. Execute the above steps to get the final comprehensive construction plan to be deployed.

[0202] On the basis of the above embodiments, a service chain deployment optimization model of a comprehensive construction plan is established in this embodiment, including: according to the attribute information of each physical node and each physical link in the cloud-edge collaborative network, and the attribute information of each virtual node and each virtual link in the comprehensive construction plan, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction plan are obtained; the service chain deployment optimization model is constructed according to the total load rate and the total delay time; wherein the attribute information of the physical node includes the total CPU resources, the remaining CPU resources, the processed data packet size and the processing capacity; the attribute information of the physical link includes the total bandwidth resources, the remaining bandwidth resources, the propagation delay time and the processed data packet size; the attribute information of the virtual node includes the CPU resources required for deployment; the attribute information of the virtual link includes the bandwidth resources required for deployment.

[0203] Optionally, after obtaining the final comprehensive construction plan, each virtual node and virtual link in the comprehensive construction plan needs to be mapped to the cloud-edge collaborative network.

[0204] Optionally, the service is generated from the terminal and connected to a specific physical node to perform VNF function processing, but the terminal does not carry the VNF function. Therefore, the base network of the cloud-edge collaborative network can be designed as a graph G(N, L), where N = {N E , N C}={n1,n2,...,n |N|} is a set of physical nodes, L = {l1, l2, ..., l |L|} is a collection of physical links connecting physical nodes. Physical nodes include two types: edge nodes and E∈N E and cloud node n C ∈N C .

[0205] The attribute set of each physical node is in, and pr k They are physical nodes n k The total CPU resources and processing power; processing power refers to the rate at which data is processed; is the physical node n k type, when When physical node n k is an edge node; when When physical node n k Edge nodes are closer to the terminal, have less propagation delay in processing services, and have fewer resources; cloud nodes are farther away from the terminal, have greater propagation delay in processing services, have more CPU resources, and can carry services with high computing volume.

[0206] The attribute set of each physical link is and d m They are respectively physical link l m Total bandwidth resources and propagation delay time; Indicates physical link l m Connecting physical nodes and therefore and d m Represents physical nodes and The total bandwidth resources between nodes and the propagation delay time between nodes.

[0207] It should be noted that there are three types of physical links: (1) If the physical link is m Both ends of the connection and If they are all edge nodes, then the physical link l m It is a link between edge nodes, with low propagation delay and small total bandwidth; (2) If the physical link l m If the edge nodes and cloud nodes are connected separately, the link propagation delay is higher and the total bandwidth is larger; (3) If the physical link l m When connecting two cloud nodes, the link propagation delay is higher and the total bandwidth is larger;

[0208] The following describes the service chain deployment method under a single cloud-edge collaborative network as an example.

[0209] Each received request contains multiple service chains to be deployed. By checking the latency requirements of the service chains to be deployed, the service type of the service chains to be deployed can be obtained. For latency-sensitive services, in order to meet latency requirements, they need to be processed as quickly as possible and tend to be deployed on edge nodes for execution. For non-latency-sensitive computing-intensive services, they tend to be deployed on cloud nodes for processing to avoid excessive consumption of limited resources on edge server nodes.

[0210] In addition, combining multiple service chains to be deployed to form a comprehensive construction plan will cause additional bandwidth consumption and increase latency. Therefore, in the specific deployment process, for latency-sensitive services, only the service chains to be deployed are processed in parallel and then deployed to the base network; for non-latency-sensitive services, the amount of calculation and resource consumption are large, so multiple service chains to be deployed are first processed in parallel and then combined to form a comprehensive construction plan, and then deployed to the base network to reduce the consumption of redundant resources during the backup process.

[0211] Before deployment, it is necessary to obtain the set of comprehensive construction solutions G to be deployed corresponding to the request information. During the deployment process, comprehensive indicators of latency consumption, computing resource consumption, and bandwidth consumption of the optimized service are comprehensively considered.

[0212] Optionally, in order to reflect the deployment of virtual nodes and virtual links in the comprehensive construction solution, two binary variables are defined: and when Indicates the jth sub-request information Virtual nodes shared by the service chain to be deployed Deployed on physical nodes On. When Indicates the jth request information Virtual links shared by the service chains to be deployed Deployed on physical links superior.

[0213] It should be noted that when processing a request for SR, each virtual node and virtual link corresponds to a different service chain to be deployed, so each virtual node or virtual link only processes one service of part of the service chain to be deployed. The mapping of virtual nodes to physical nodes is one-to-one, and the mapping of virtual links may include multi-hop continuous physical links.

[0214] The total load rate of the cloud-edge collaborative network is the sum of the load rates of all physical nodes and physical links in the cloud-edge collaborative network; the load rate of the physical node can be calculated and obtained through the attribute information of the VNF and the attribute information of the physical node; the load rate of the physical link can be calculated and obtained through the attribute information of the VNF and the attribute information of the physical link;

[0215] It should be noted that the main chain is the path with the longest delay in the parallelized service chain to be deployed, so the delay of the main chain is the delay of the service chain to be deployed.

[0216] The delay time of the service chain to be deployed includes but is not limited to propagation delay, transmission delay, processing delay and queuing delay.

[0217] Optionally, during the VNF deployment process, the optimization goal is to optimize the weighted cost of the load rate and latency of the cloud-edge collaborative network while ensuring the software reliability of the service chain to be deployed; wherein the formula of the service chain deployment optimization model is:

[0218]

[0219]

[0220]

[0221]

[0222]

[0223] Where goal(·) is the objective function of the service chain deployment optimization model; where, according to the resource allocation optimization problem model, the optimal solution of the service chain deployment optimization model is:

[0224]

[0225] Among them, S is a virtual node Deployed on physical node n k The objective function goal(·) will generate a corresponding value goal(S) in the solution S.

[0226] Among them, C1, C2, C3, C4 and C5 are the constraints of the service chain deployment optimization model; SR is the request information, SR j is the jth sub-request information in the request information; w l and w d is the weight coefficient, w l +w d =1; For virtual nodes On physical nodes deployment decisions; For virtual nodes Deployed on physical nodes When the physical node Load factor; For physical nodes The first load rate, For physical nodes The second load rate, For virtual nodes Deploy the required CPU resources, For physical nodes The total CPU resources of the edge-cloud collaboration network are as follows: N and L are the number of physical nodes and physical links in the cloud-edge collaboration network, respectively; N V and L V are the number of virtual nodes and virtual links in the comprehensive construction scheme, respectively; Virtual Link In the physical link deployment decisions; Virtual Link Deployed on physical links , the load rate of the physical link; For physical link The first load rate, For physical link The second load rate, Virtual Link Deploy the required link bandwidth resources, For physical link Total bandwidth resources; is the delay of the service chain SFCs to be deployed, totdelay s is the delay time of the service chain SFCs to be deployed in the comprehensive construction solution, stdelay is the standard delay time of the service chain to be deployed in advance; rdelay s is the maximum tolerable delay time of the service chain SFCs to be deployed.

[0227] The total load rate in the brackets of the objective function is intended to reduce the load level of the network by optimizing the load levels of physical nodes and physical links after deployment.

[0228] Constraints C1 and C2 are used to limit the sum of physical resources required by virtual nodes and virtual links deployed on physical nodes and physical links of the cloud-edge collaborative network to not exceed their total physical resources;

[0229] Constraint C3 is used to limit each virtual node to be hosted by only one physical node;

[0230] Constraint C4 is used to prevent the load from being concentrated on a part of the physical nodes, so as to better balance the load of the network, that is, at most one virtual node can be deployed on the same physical node in the same request information;

[0231] Constraint C5 is used to limit the delay time of each service chain to be deployed to meet the delay requirement.

[0232] On the basis of the above-mentioned implementation embodiments, in this embodiment, according to the attribute information of each physical node and each physical link in the cloud-edge collaborative network, and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme are obtained, including: according to the total CPU resources and the remaining CPU resources of each physical node of the cloud-edge collaborative network, the first load rate of each physical node is obtained;

[0233] Among them, the physical node n k The calculation formula of the first load rate is:

[0234]

[0235] Among them, lr k is the physical node n k The first load rate, and They are physical nodes n k The total CPU resources and remaining CPU resources.

[0236] Obtaining a second load rate of each physical node according to the total CPU resources of each physical node and the CPU resources required for deploying the virtual nodes carried on each physical node;

[0237] Optionally, a ratio between the CPU resources required for deployment of the virtual nodes carried on each physical node and the total CPU resources of each physical node is calculated, and the ratio is used as the second load rate of each physical node.

[0238] Obtaining a first load rate of each physical link according to the total bandwidth resources and the remaining bandwidth resources of each physical link of the cloud-edge collaborative network;

[0239] Among them, the physical link The calculation formula of the first load rate is:

[0240]

[0241] Among them, lr m is the physical link l m The first load rate; and They are respectively physical link l m The total bandwidth resources and remaining bandwidth resources.

[0242] Obtaining a second load rate of each physical link according to the total bandwidth resources of each physical link and the bandwidth resources required for the deployment of the virtual links carried on each physical link;

[0243] Optionally, a ratio between bandwidth resources required for deployment of virtual links carried on each physical link and total bandwidth resources of each physical link is calculated, and the ratio is used as the second load rate of each physical link.

[0244] Obtain a total load rate according to the first load rate and the second load rate of all physical nodes in the cloud-edge collaborative network and the first load rate and the second load rate of all physical links;

[0245] Add the first load rate and the second load rate of all physical nodes in the cloud-edge collaborative network as the total load rate of all physical nodes; add the first load rate and the second load rate of all physical links as the total load rate of all physical links; then add the total load rate of all physical nodes and the total load rate of all physical links to obtain the total load rate.

[0246] According to the propagation delay time of each physical link, the transmission delay time of each service chain to be deployed is obtained;

[0247] Optionally, the propagation delay is the delay caused by the length of the physical link, which is determined by the physical link attribute information The propagation delay time d m Confirmed; Service chain SFC to be deployed s The propagation delay time propdelay s The calculation formula is:

[0248]

[0249] According to the data packet size processed by each physical link and the bandwidth resources required for the deployment of each virtual link, the transmission delay time of each service chain to be deployed is obtained;

[0250] Among them, the transmission delay includes the wireless transmission delay at the terminal layer and the wired transmission delay at the edge layer and cloud layer.

[0251] At the terminal layer, service traffic is transmitted wirelessly, such as 5G (5th Generation Mobile Communication Technolog), to provide user terminals with guaranteed performance services, so as to adapt to the transmission bandwidth required by the terminal. The transmission bandwidth of the 5G physical link can meet the data rate requirements of the terminal by adjusting the transmission power and frequency band.

[0252] At the edge layer and cloud layer, the wired transmission delay is the time required for a virtual node to transmit a packet of a VNF to one of its output links. It is assumed that the transmission delay mainly depends on the bandwidth resources allocated to the service chain to be deployed and the length of the packet. Assuming that the deployment on the physical node The virtual node is Service chain SFC to be deployed sTransmission delay trandelay s can be written as:

[0253]

[0254] in, Physical nodes on physical links The size of the data packets processed, For this virtual node To the successor virtual node The bandwidth resources required by the virtual link Deploy the required bandwidth resources.

[0255] According to the data packet size and processing capacity processed by each physical node, as well as the CPU resources required for the deployment of each virtual node, the processing delay time of each service chain to be deployed is obtained;

[0256] Among them, the processing delay represents a physical node The time required to process the data packets deployed on the node. It is assumed that the processing delay is related to the length of the arriving data packets, the computing resources allocated to the VNF, and the processing capacity of the physical node. s Processing delay procdelay s The formula is:

[0257]

[0258] in, is the physical node n k2 Handling virtual nodes The packet size, is the physical node n k2 Virtual nodes deployed at Required CPU resources, is the physical node n k2 processing capabilities.

[0259] The total delay time is obtained according to the propagation delay time, transmission delay time, processing delay time and queuing delay time of all service chains to be deployed in the comprehensive construction plan.

[0260] The queuing delay time indicates the waiting time for data packets to be processed on the physical node, that is, the queuing delay time is the sum of the processing time of all VNFs deployed on the physical node before the current virtual node is deployed. s The calculation steps of the queuing delay time are as follows: before the request SR arrives, z requests have been processed, then the current service chain SFC to be deployed s The queuing delay of is the sum of the queuing delays of each virtual node it passes through, and the calculation formula is:

[0261]

[0262] in, is the physical node n k2 Virtual nodes hosted on The size of the processed data packet, rcpu k′ For virtual nodes Deploy the required CPU resources, Virtual node in historical request information At physical node n k2 Deployment decision on SR u request information for history;

[0263] Service chain SFC to be deployed s The total delay time is calculated as:

[0264] totdelay s =propdelay s +trandelay s +procdelay s +queudelay s ;

[0265] This example comprehensively considers multiple delay times, so that the total delay time obtained is more accurate, thereby minimizing the delay time of the service chain to be deployed.

[0266] In order to further verify the effectiveness of the service chain deployment method of this embodiment, the service chain deployment method of this embodiment is evaluated in combination with specific examples.

[0267] The service chain deployment method in this embodiment can be simulated and verified based on Python language or the like.

[0268] In order to simulate the architecture of the cloud-edge collaborative network, a network topology consisting of 40 physical nodes is used, of which 30 physical nodes are cloud nodes and 10 physical nodes are edge nodes. The physical nodes in the network are connected by 204 physical links.

[0269] The request information is generated according to a certain arrival frequency, such as a five-time window period. The first four time windows arrive at delay-sensitive services, which consist of one service chain to be deployed; the fifth time window arrives at computationally intensive services, which consist of four service chains to be deployed.

[0270] In order to simulate the differentiated service requirements of different types of service chains to be deployed, a set of 10 types of VNFs is selected, and the type range of VNFs is [0, 9]; v1, v2 and v3 are monitor-type VNFs, and the others are rectifier-type VNFs. In addition, it is assumed that each service chain to be deployed consists of 2 to 5 different VNFs, and the required bandwidth resources of each VNF are 700 Mbps to 900 Mbps.

[0271] The reliability index of each VNF is between [0.95, 0.99]. The reliability index value of each service request is one of {0.95, 0.98, 0.99, 0.995, 0.999}. Assume that the dependency constraint set of the VNF type is {(v0), (v4, v5), (v1, v2, v3), (v6, v7, v8), (v9)*}. The dependency level in the entire dependency constraint set decreases from left to right, that is, the traffic must pass through the VNF in the left bracket before it can be smoothly processed by the VNF in the right bracket. VNFs in the same bracket have the same dependency level, and VNF instances of type v9 can be linked in any order.

[0272] For delay-sensitive services, the delay constraint of each VNF is set to 100ms, and for compute-intensive applications, the delay constraint of each VNF is set to 200ms.

[0273] It should be noted that the parameters of the cloud-edge collaborative network and the service chain to be deployed are randomly set and obey uniform distribution.

[0274] For the tabu search algorithm, the tabu table size is 20, the number of iterations is 50, and the epsilon of the epsilon-greedy method is set to 0.3. The minimum reliability index requirement of the service chain to be deployed is 0.9, and the maximum number of instance sharing is 5.

[0275] In order to verify the performance of the service chain deployment method implemented in this paper, the following methods are used for comparative verification:

[0276] The deployment method in this embodiment is taken as method 1; wherein the weight parameter in the service chain deployment optimization model is w l =0.3, w d =0.7.

[0277] For method 2, only the service chains to be deployed are parallelized and redundantly backed up, but multiple service chains to be deployed are not combined; in the deployment phase, the taboo search algorithm is used to complete the deployment. This method is used to compare the impact of combination processing on service chain deployment. Among them, the weight parameter in the optimization model is w l =0.3, w d =0.7.

[0278] For method 3, multiple service chains to be deployed are directly combined to form a service function graph SFG, and then redundant backup is performed, but the service chains to be deployed are not parallelized; and the business types of the service chains to be deployed are not distinguished, and all service chains to be deployed are combined to form SFG. In the deployment phase, the SFG is deployed to the base network using the taboo search algorithm. This method is used to compare the impact of parallel processing on service chain deployment. Among them, the weight parameter in the optimization model is w l =0.3, w d =0.7.

[0279] For method 4, the service chain to be deployed is not pre-built, that is, no parallel processing and combination are performed. The service chain to be deployed directly meets the software reliability requirements by backing up the VNF instance, and then the service chain is deployed to the base network using the taboo search algorithm. Among them, the weight parameter in the optimization model is w l =0.3, w d =0.7.

[0280] The following takes a non-delay-sensitive service consisting of four service chains SR={SFC1, SFC2, SFC3, SFC4} as an example to describe the effectiveness of the service chain deployment method in this embodiment.

[0281] For service chain SFC1, the included VNFs are v0, v4, v1, v2, and v7, and the reliability indices are 0.99, 0.95, 0.97, 0.98, and 0.99, respectively. The initial software reliability index of service chain SFC1 is 0.885;

[0282] For service chain SFC2, the included VNFs are v0, v4, v1, v2, and v8, and the reliability indices are 0.99, 0.95, 0.97, 0.98, and 0.98, respectively. The initial software reliability index of service chain SFC2 is 0.876;

[0283] For service chain SFC3, the included VNFs are v0, v4, v1, and v6, and the reliability indexes are 0.99, 0.95, 0.97, and 0.99, respectively. The initial software reliability index of service chain SFC3 is 0.903;

[0284] For service chain SFC4, the included VNFs are v0 and v4, with reliability indices of 0.99 and 0.95, respectively. The initial software reliability index of service chain SFC3 is 0.9405;

[0285] From the dependency constraints of VNF, we can see that in the above example, v4 depends on v0; v1 and v2 depend on v4; v6, v7 and v8 depend on v1 and v2. Instances of other types of VNFs can be linked in any order. Figure 6As shown in the figure, the first construction plan of the four service chains is formed according to the dependency relationship and the preset arrangement order. Then, according to the dependency and order between the VNFs in each service chain to be deployed, as well as the type of VNF, the service chains to be deployed are processed in parallel to obtain the second construction plan. Then, since the four service chains are all non-delay sensitive, the four service chains are merged to form a comprehensive construction plan. Finally, it can be seen from the software reliability index that SFC1 and SFC2 need redundant backup, and the virtual node corresponding to v4 with the lowest reliability is found to back it up, and a new virtual node is established to back up the v4 instance.

[0286] After the backup is completed, the reliability index of the virtual node corresponding to v4 is updated to 0.9975, the reliability of SFC1 is updated to 0.929, and the reliability of SFC2 is changed to 0.948, meeting the minimum reliability requirements. The final comprehensive construction plan formed at this time meets the deployment requirements and can be used for the next stage of deployment.

[0287] After the final comprehensive construction plan is generated, the virtual nodes in the comprehensive construction plan are deployed, and the performance of the deployment method in this embodiment is compared and verified from the aspects of performance under different parameters, optimization goals, end-to-end service delay, and reliability of unit cost improvement.

[0288] For different weight parameters, only the weight factor is adjusted, and other parameters remain unchanged; and the service chain deployment optimization model is evaluated using the objective function value cost = goal (S), the acceptance rate of the service chain, and the reliability of the service chain unit cost improvement.

[0289] Among them, the calculation formula of the reliability χ increased by the unit cost of the service chain is:

[0290]

[0291] from Figure 7-Figure 9 It can be seen that different weight factors have a certain impact on the deployment results. l =0,w d = 1, the objective function value is the smallest, but the acceptance rate is too low; because in the case of fully optimizing the delay, the key path with the shortest service delay is occupied early in the first comprehensive construction plan, resulting in the end-to-end delay of the service when the subsequent comprehensive construction plan is deployed. Therefore, it is not suitable for large-scale service chain deployment. l =0.3, w d =0.7, when the objective function value is minimum, the service reception rate can still be effectively guaranteed, and the resource utility rate in the backup phase can be effectively improved.

[0292] like Fig. 9As shown in the figure, when the time is early and the number of integrated construction schemes deployed in the network is small, the reliability of the service chain unit cost improvement χ fluctuates greatly. t=0 is the first time window; at time t=4, when the fifth time window arrives, the first non-delay-sensitive task arrives, forming a comprehensive construction scheme to complete the backup based on instance sharing, which improves the resource utility of the backup. Therefore, χ rises significantly.

[0293] Optionally, after comprehensive consideration of the objective function value, acceptance rate, and reliability x of the service chain unit cost improvement, the above simulation experiment can be optimized as l =0.3, w d =0.7 as the weight coefficient.

[0294] By setting different weight coefficients, the trade-off between load balancing performance and network quality performance during service chain deployment can be flexibly adjusted. Therefore, in actual applications, the weight coefficient can be adjusted according to actual needs to meet different performance requirements.

[0295] like Fig.10 As shown, different methods have different performances in terms of network quality and deployment cost balance; Method 1 proposed in this implementation performs best when latency and resource consumption are comprehensively considered. Method 3 does not perform parallel processing to shorten the overall processing latency, so it generates more latency overhead; and as the number of service chain deployments increases, this method can effectively reduce the additional resource consumption caused by improving reliability, and the objective function value grows slowly. Method 3 performed better in the early days because the number of service chains was small, but as the number of service chains increased, the objective function value increased more because the number of backup virtual nodes increased. In summary, the deployment method that only backs up the service chain has a large latency, and backup will increase resource costs.

[0296] like Fig.11 As shown, the end-to-end delay of different methods when deploying delay-sensitive services is shown; Method 3 and Method 4 have a larger delay because the service chain is not parallelized. The end-to-end delay of Method 1 and Method 2, which are parallelized, is lower. Although both are parallelized algorithms, there are still subtle differences between Method 1 and Method 2. This is because the deployment of non-delay-sensitive services will affect the deployment of delay-sensitive services. When processing non-delay-sensitive services, Method 2 will not combine the service chains, resulting in the optimal path being occupied early, which will cause greater transmission delays when delay-sensitive services are deployed. Taking into account the performance of multiple service chains, Method 1 in this embodiment has a good load balancing effect.

[0297] like Fig.12As shown in the figure, the end-to-end delay of different methods when deploying non-delay-sensitive services is shown; Method 1 and Method 2, which process service chains in parallel, have better performance in terms of delay. Since there are more service chains to be deployed corresponding to non-delay-sensitive services, the structure of the combined comprehensive construction plan is complex. Based on load balancing considerations, the number of virtual links of a small number of service chains to be deployed with high resource consumption may increase, resulting in an increase in the number of forwarding hops and a slight increase in delay. However, its network performance on all service chains to be deployed in the entire request message is the best.

[0298] like Fig.13 As shown, the reliability x of the service chain per unit cost improvement of different methods is shown; according to the formula for updating the reliability index, it can be seen that the improvement in reliability is determined by the number of backup NVF instances; usually, the final reliability index after the redundant backup is completed by different methods is not much different. Therefore, the objective function value of the denominator is mainly evaluated. For the service chain processed in parallel, additional virtual links will be added, resulting in additional resource consumption. After the comprehensive construction scheme is formed by merging, the deployment cost is saved by reducing the number of virtual nodes. Therefore, the reliability x of the unit cost improvement of method 1 and method 4 after parallelization and merging is not much different, because the resources consumed by large-scale parallelization of method 1 are greater than the resources saved by the backup of the comprehensive construction scheme. Although some performance in method 1 is slightly worse, method 1 has the best comprehensive network performance on all service chains to be deployed in the entire request message.

[0299] In summary, although there are many studies that provide solutions to the construction and mapping of service chains from different angles, most of them are limited to the construction of a single service chain, ignoring the differences in requirements between different services and VNFs, and the optimization objectives are limited to one of the aspects of energy consumption of deployment, reliability of service chains, and end-to-end delay of service chains, making it difficult to achieve balanced optimization of multiple objectives in service chain deployment. However, this embodiment integrates multiple service chains for parallel processing and merging to form a comprehensive construction plan, and builds a service chain deployment optimization model based on network performance, service quality, and service delay time to reduce end-to-end delay and resource costs during deployment, while ensuring the basic requirements of IoT terminals for service reliability, and taking into account the interests of network service providers and terminals, so that the deployment method in this embodiment is superior to other methods in terms of improving service quality, reducing deployment costs, and meeting the comprehensive performance of different requirements of IoT services.

[0300] The service chain deployment device for the cloud-edge collaborative Internet of Things provided by the present invention is described below. The service chain deployment device for the cloud-edge collaborative Internet of Things described below and the service chain deployment method for the cloud-edge collaborative Internet of Things described above can be referenced to each other.

[0301] like Fig.14 As shown, the service chain deployment device for cloud-edge collaborative Internet of Things provided in this embodiment includes a first construction module 1401, a second construction module 1402, a third construction module 1403 and a deployment module 1404;

[0302] The first construction module 1401 is used to obtain a first construction plan for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each virtual network function VNF in each service chain to be deployed;

[0303] The second construction module 1402 is used to determine the VNFs with parallel function requirements in each service chain to be deployed according to the service type of each VNF, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain the second construction scheme of each service chain to be deployed;

[0304] The third construction module 1403 is used to merge the at least two virtual nodes in the second construction schemes of different service chains to be deployed to obtain a comprehensive construction scheme if at least two virtual nodes in the second construction schemes of different service chains to be deployed both carry VNFs of the same network function type and have an instance sharing function;

[0305] The deployment module 1404 is used to establish a service chain deployment optimization model for the comprehensive construction plan, obtain the deployment plan for each virtual node and each virtual link in the comprehensive construction plan according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on the physical node of the cloud-edge collaborative network and deploy each virtual link on the cloud-edge collaborative network according to the deployment plan.

[0306] This embodiment divides each service chain to be deployed into multiple NVFs on the one hand, and arranges and parallelizes the VNFs in each service chain to be deployed according to the preset arrangement order, preset dependency and business type of each VNF in each service chain to be deployed, so as to form a second construction plan for each service chain to be deployed, and merges the virtual nodes that carry VNFs of the same network function type and have instance sharing functions in the second construction plans of multiple service chains to be deployed, so as to merge heterogeneous service chains, so that the construction plan of the service chain to be deployed is more reasonable, and the virtual nodes can be effectively reduced, and the end-to-end delay and resource consumption cost in the mapping process of the service chain to be deployed can be reduced; on the other hand, a service chain deployment optimization model is constructed with the goal of minimizing the total load rate of the cloud-edge collaborative network and minimizing the delay time of all service chains to be deployed in the comprehensive construction plan, and according to the optimal solution of the service chain deployment optimization model, each virtual node and each virtual link in the comprehensive construction plan are reasonably deployed, which can effectively reduce the service delay of the tasks to be processed and improve the reliability and stability of the cloud-edge collaborative network service.

[0307] Fig.15An example of a physical structure diagram of an electronic device is shown in FIG. Fig.15 As shown, the electronic device may include: a processor (processor) 1501, a communication interface (Communications Interface) 1502, a memory (memory) 1503 and a communication bus 1504, wherein the processor 1501, the communication interface 1502, and the memory 1503 communicate with each other through the communication bus 1504. The processor 1501 can call the logic instructions in the memory 1503 to execute a service chain deployment method for a cloud-edge collaborative Internet of Things, the method comprising: obtaining a first construction plan for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each VNF in each service chain to be deployed; determining the VNF with parallel function requirements in each service chain to be deployed according to the business type of each VNF, and parallelizing the VNF with parallel function requirements in the first construction plan of each service chain to be deployed to obtain a second construction plan for each service chain to be deployed; if there are at least two virtual nodes in the second construction plans of different service chains to be deployed that both carry VNFs of the same network function type and have an instance sharing function, then at least two virtual nodes in the second construction plans of the different service chains to be deployed are merged to obtain a comprehensive construction plan; establishing a service chain deployment optimization model for the comprehensive construction plan, obtaining a deployment plan for each virtual node and each virtual link in the comprehensive construction plan according to the optimal solution of the service chain deployment optimization model, and deploying each virtual node on a physical node of the cloud-edge collaborative network and each virtual link on the cloud-edge collaborative network according to the deployment plan.

[0308] In addition, the logic instructions in the above-mentioned memory 1503 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk.

[0309] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the service chain deployment method for cloud-edge collaborative Internet of Things provided by the above methods. The method includes: obtaining a first construction plan for each service chain to be deployed according to a preset arrangement order and preset dependency relationship of each VNF in each service chain to be deployed; determining a VNF with parallel function requirements in each service chain to be deployed according to a business type of each VNF, and parallelizing the VNF with parallel function requirements in the first construction plan of each service chain to be deployed Processing to obtain a second construction plan for each service chain to be deployed; if there are at least two virtual nodes in the second construction plans of different service chains to be deployed, both of which carry VNFs of the same network function type and have instance sharing functions, then merge at least two virtual nodes in the second construction plans of the different service chains to be deployed to obtain a comprehensive construction plan; establish a service chain deployment optimization model for the comprehensive construction plan, and according to the optimal solution of the service chain deployment optimization model, obtain a deployment plan for each virtual node and each virtual link in the comprehensive construction plan, and deploy each virtual node on a physical node of the cloud-edge collaborative network and deploy each virtual link on the cloud-edge collaborative network according to the deployment plan.

[0310] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the computer program is executed by a processor to execute the service chain deployment method for cloud-edge collaborative Internet of Things provided by the above methods, the method comprising: obtaining a first construction plan for each service chain to be deployed according to a preset arrangement order and preset dependency relationship of each VNF in each service chain to be deployed; determining a VNF with parallel function requirements in each service chain to be deployed according to a business type of each VNF, and parallelizing the VNF with parallel function requirements in the first construction plan of each service chain to be deployed, so as to obtain each service chain to be deployed. a second construction plan for the service chains to be deployed; if there are at least two virtual nodes in the second construction plans of different service chains to be deployed, both of which carry VNFs of the same network function type and have instance sharing functions, then merge at least two virtual nodes in the second construction plans of the different service chains to be deployed to obtain a comprehensive construction plan; establish a service chain deployment optimization model for the comprehensive construction plan, and according to the optimal solution of the service chain deployment optimization model, obtain the deployment plan for each virtual node and each virtual link in the comprehensive construction plan, and deploy each virtual node on the physical node of the cloud-edge collaborative network and deploy each virtual link on the cloud-edge collaborative network according to the deployment plan.

[0311] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without creative effort.

[0312] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0313] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A service chain deployment method for cloud-edge collaborative Internet of Things, characterized in that: include: Obtain a first construction plan for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each virtual network function VNF in each service chain to be deployed; According to the service type of each VNF, determine the VNFs with parallel function requirements in each service chain to be deployed, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain the second construction scheme of each service chain to be deployed; If there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, then at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged to obtain a comprehensive construction scheme; Establish a service chain deployment optimization model for the comprehensive construction solution, obtain a deployment solution for each virtual node and each virtual link in the comprehensive construction solution according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on a physical node of the cloud-edge collaborative network and each virtual link on a physical link of the cloud-edge collaborative network according to the deployment solution; The service chain deployment optimization model for establishing the comprehensive construction solution includes: According to the attribute information of each physical node and each physical link in the cloud-edge collaborative network and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme are obtained; Constructing the service chain deployment optimization model according to the total load rate and the total delay time; The attribute information of the physical node includes the total CPU resources, the remaining CPU resources, the processed data packet size and the processing capacity; the attribute information of the physical link includes the total bandwidth resources, the remaining bandwidth resources, the propagation delay time and the processed data packet size; the attribute information of the virtual node includes the CPU resources required for deployment; the attribute information of the virtual link includes the bandwidth resources required for deployment; The acquiring, according to the attribute information of each physical node and each physical link in the cloud-edge collaborative network and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme includes: Obtaining a first load rate of each physical node according to the total CPU resources and the remaining CPU resources of each physical node of the cloud-edge collaborative network; Obtaining a second load rate of each physical node according to the total CPU resources of each physical node and the CPU resources required for deploying the virtual nodes carried on each physical node; Obtaining a first load rate of each physical link according to the total bandwidth resources and the remaining bandwidth resources of each physical link of the cloud-edge collaborative network; Obtaining a second load rate of each physical link according to the total bandwidth resources of each physical link and the bandwidth resources required for the deployment of the virtual links carried on each physical link; Obtaining the total load rate according to the first load rate and the second load rate of all physical nodes in the cloud-edge collaborative network and the first load rate and the second load rate of all physical links in the cloud-edge collaborative network; According to the propagation delay time of each physical link, the propagation delay time of each service chain to be deployed is obtained; According to the data packet size processed by each physical link and the bandwidth resources required for the deployment of each virtual link, the transmission delay time of each service chain to be deployed is obtained; According to the data packet size and processing capacity processed by each physical node, as well as the CPU resources required for the deployment of each virtual node, the processing delay time of each service chain to be deployed is obtained; The total delay time is obtained according to the propagation delay time, the transmission delay time, the processing delay time and the queuing delay time of all the service chains to be deployed in the comprehensive construction solution.

2. The service chain deployment method for cloud-edge collaborative Internet of Things according to claim 1 is characterized in that: The method of determining, according to the service type of each VNF, the VNFs with parallel function requirements in each service chain to be deployed, and performing parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain the second construction scheme of each service chain to be deployed includes: For any VNF, if the service type of the VNF and the service type of the next VNF ​​adjacent to the VNF are both monitor types, it is determined that the VNF has a parallel function requirement; If there is a dependency relationship between any two VNFs with parallel functional requirements in each service chain to be deployed, the any two VNFs are serially connected according to the dependency relationship, and the two serially connected VNFs are used as new VNFs with parallel functional requirements; The new VNFs with parallel function requirements in the first construction plan of each service chain to be deployed are processed in parallel to obtain a second construction plan of each service chain to be deployed.

3. The service chain deployment method for cloud-edge collaborative Internet of Things according to claim 1 is characterized in that: If there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, at least two virtual nodes in the second construction schemes of the different service chains to be deployed are merged to obtain a comprehensive construction scheme, including: If there are at least two virtual nodes in the second construction scheme of different service chains to be deployed, both of which carry the same network function type VNF and have an instance sharing function, then the number of VNFs carried by the at least two virtual nodes is calculated; If the number of VNFs carried by the at least two virtual nodes is less than or equal to a first preset value, the at least two virtual nodes are merged.

4. The service chain deployment method for cloud-edge collaborative Internet of Things according to any one of claims 1 to 3, characterized in that: If there are at least two virtual nodes in the second construction schemes of different service chains to be deployed, both of which carry VNFs of the same network function type and have an instance sharing function, merging at least two virtual nodes in the second construction schemes of the different service chains to be deployed includes: Determine the service type of each service chain to be deployed according to the maximum tolerable delay index of each service chain to be deployed; wherein the service type of the service chain to be deployed includes delay-sensitive services and non-delay-sensitive services; The service chain to be deployed, whose service type is non-delay-sensitive service, is used as the target service chain; If there are at least two virtual nodes in the second construction scheme of different target service chains, both of which carry the same network function type VNF and have instance sharing function, the at least two virtual nodes are merged.

5. The service chain deployment method for cloud-edge collaborative Internet of Things according to any one of claims 1 to 3, characterized in that: The service chain deployment optimization model for establishing the comprehensive construction solution includes: Calculate the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed in the comprehensive construction solution, and use the product as the reliability index of each service chain to be deployed; Backing up the VNF with the smallest reliability index in the service chain to be deployed whose reliability index is less than or equal to the second preset value; After updating the reliability index of the virtual node corresponding to the backed-up VNF, recalculating the product of the reliability indexes of the virtual nodes corresponding to all VNFs in each service chain to be deployed until the reliability index of each service chain to be deployed is greater than the second preset value; According to the comprehensive construction plan after the last backup, the service chain deployment optimization model is established.

6. A service chain deployment device for cloud-edge collaborative Internet of Things, characterized in that: include: The first construction module is used to obtain a first construction scheme for each service chain to be deployed according to a preset arrangement order and a preset dependency relationship of each virtual network function VNF in each service chain to be deployed; The second construction module is used to determine the VNFs with parallel function requirements in each service chain to be deployed according to the service type of each VNF, and perform parallel processing on the VNFs with parallel function requirements in the first construction scheme of each service chain to be deployed to obtain a second construction scheme of each service chain to be deployed; The third construction module is configured to merge the at least two virtual nodes in the second construction schemes of different service chains to be deployed to obtain a comprehensive construction scheme if at least two virtual nodes in the second construction schemes of different service chains to be deployed both carry VNFs of the same network function type and have an instance sharing function. A deployment module, used to establish a service chain deployment optimization model of the comprehensive construction solution, obtain a deployment plan for each virtual node and each virtual link in the comprehensive construction solution according to the optimal solution of the service chain deployment optimization model, and deploy each virtual node on a physical node of the cloud-edge collaborative network and each virtual link on a physical link of the cloud-edge collaborative network according to the deployment plan; The deployment module is specifically used for: According to the attribute information of each physical node and each physical link in the cloud-edge collaborative network and the attribute information of each virtual node and each virtual link in the comprehensive construction scheme, the total load rate of the cloud-edge collaborative network and the total delay time of all service chains to be deployed in the comprehensive construction scheme are obtained; Constructing the service chain deployment optimization model according to the total load rate and the total delay time; The attribute information of the physical node includes the total CPU resources, the remaining CPU resources, the processed data packet size and the processing capacity; the attribute information of the physical link includes the total bandwidth resources, the remaining bandwidth resources, the propagation delay time and the processed data packet size; the attribute information of the virtual node includes the CPU resources required for deployment; the attribute information of the virtual link includes the bandwidth resources required for deployment; The deployment module is further used to: Obtaining a first load rate of each physical node according to the total CPU resources and the remaining CPU resources of each physical node of the cloud-edge collaborative network; Obtaining a second load rate of each physical node according to the total CPU resources of each physical node and the CPU resources required for deploying the virtual nodes carried on each physical node; Obtaining a first load rate of each physical link according to the total bandwidth resources and the remaining bandwidth resources of each physical link of the cloud-edge collaborative network; Obtaining a second load rate of each physical link according to the total bandwidth resources of each physical link and the bandwidth resources required for the deployment of the virtual links carried on each physical link; Obtaining the total load rate according to the first load rate and the second load rate of all physical nodes in the cloud-edge collaborative network and the first load rate and the second load rate of all physical links in the cloud-edge collaborative network; According to the propagation delay time of each physical link, the propagation delay time of each service chain to be deployed is obtained; According to the data packet size processed by each physical link and the bandwidth resources required for the deployment of each virtual link, the transmission delay time of each service chain to be deployed is obtained; According to the data packet size and processing capacity processed by each physical node, as well as the CPU resources required for the deployment of each virtual node, the processing delay time of each service chain to be deployed is obtained; The total delay time is obtained according to the propagation delay time, the transmission delay time, the processing delay time and the queuing delay time of all the service chains to be deployed in the comprehensive construction solution.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the steps of the service chain deployment method for cloud-edge collaborative Internet of Things as described in any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the service chain deployment method for cloud-edge collaborative Internet of Things as described in any one of claims 1 to 5 are implemented.

Images