Network slice monitoring method, system and storage medium based on in-band network telemetry

Abstract

The present invention discloses a network slice monitoring method, system and storage medium based on in-band network telemetry, the method comprising: obtaining underlying physical network information and resource requests of multiple network slices; determining the deployment order of each network slice according to the resource request, and deploying multiple network slices on the underlying physical network according to the deployment order according to the underlying physical network information; planning a detection path for each successfully deployed network slice and generating an in-band network telemetry INT probe; sending the INT probe to the underlying physical network according to the detection path and forwarding it on the underlying physical network, so as to monitor the network slice using the INT probe and obtain INT telemetry information. This method can obtain fine-grained performance information of network slices after deployment, and then can timely adjust the deployment of network slices and perform rapid fault location according to the collected network slice performance information, so as to better manage the network and improve the utility of the network.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular to a network slice monitoring method, system and storage medium based on in-band network telemetry. Background Art

[0002] NS (Network Slicing) is regarded as a key technology in 5G mobile networks. It supports multiple different vertical services on a shared physical infrastructure and ensures the flexibility and efficiency of the network. The network slice consists of Figure 1 As shown in the figure, VNF stands for virtualized network function. In actual networks, multiple network slices are often deployed on a shared common physical underlying architecture and provide services at the same time. Network managers need to collect status information of these deployed network slices to ensure business needs and quickly locate faults to improve network utility. However, performance monitoring after network slice deployment is a difficult problem. Traditional monitoring technologies often have problems with granularity and real-time performance that do not meet requirements. Summary of the invention

[0003] The present invention aims to solve at least one of the technical problems in the related art to a certain extent. To this end, one object of the present invention is to propose a network slice monitoring method, system and storage medium based on in-band network telemetry to achieve refined monitoring of network slices, thereby assisting the management of network slices and performing rapid fault location.

[0004] In the first aspect, an embodiment of the present invention proposes a network slice monitoring method based on in-band network telemetry, the method comprising: obtaining underlying physical network information and resource requests of multiple network slices; determining the deployment order of each of the network slices according to the resource requests, and deploying multiple network slices on the underlying physical network according to the deployment order according to the underlying physical network information; planning a detection path for each successfully deployed network slice and generating an in-band network telemetry INT probe; sending the INT probe to the underlying physical network according to the detection path and forwarding it in the underlying physical network, so as to use the INT probe to monitor the network slice and obtain INT telemetry information.

[0005] In the second aspect, an embodiment of the present invention proposes a network slice monitoring system based on in-band network telemetry, the system comprising: an underlying physical network; a slice orchestration management controller, used to obtain the underlying physical network information and resource requests of multiple network slices, and determine the deployment order of each of the network slices according to the resource requests, and deploy multiple network slices on the underlying physical network according to the deployment order according to the underlying physical network information; an in-band network telemetry INT framework, used to plan a detection path for each successfully deployed network slice and generate an in-band network telemetry INT probe, and send the INT probe to the underlying physical network and forward it on the underlying physical network according to the detection path, so as to use the INT probe to monitor the network slice and obtain INT telemetry information.

[0006] In a third aspect, an embodiment of the present invention proposes a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the network slice monitoring method based on in-band network telemetry described in the first aspect is implemented.

[0007] The network slice monitoring method, system and storage medium based on in-band network telemetry of the embodiment of the present invention use in-band network telemetry technology to realize performance monitoring after the network slice is deployed on the underlying physical network. First, the information of the underlying physical network is obtained, and the network slices are arranged according to the comprehensive resources of the slice request, the deployment order of the network slices is determined, and then the deployment of the network slices is realized. After the network slices are deployed on the underlying physical network, a dedicated detection path is planned and a dedicated INT probe is generated for each successfully deployed network slice, and forwarded to the underlying physical network. The INT probe performs in-band network telemetry in the corresponding telemetry domain and reports the telemetry information, thereby completing the performance monitoring of the network slice. In this way, the performance information of the fine-grained network slice after deployment can be obtained, and then the deployment of the network slice can be adjusted in time and the fault location can be quickly located according to the collected network slice performance information, so as to better manage the network and improve the utility of the network.

[0008] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram of the structure of a network slice according to an embodiment of the present invention;

[0010] Figure 2 is a flow chart of a network slice monitoring method based on in-band network telemetry according to an embodiment of the present invention;

[0011] Figure 3is a schematic diagram of the SR label stack structure of an embodiment of the present invention;

[0012] Figure 4 is a structural block diagram of a network slice monitoring system based on in-band network telemetry according to an embodiment of the present invention;

[0013] Figure 5 It is a structural block diagram of a network slice monitoring system based on in-band network telemetry according to a specific embodiment of the present invention. DETAILED DESCRIPTION

[0014] With the use of new mature paradigms, such as NFV (Network Function Virtualization) and SDN (Software Defined Network), network slicing technology enables the basic InP (Infrastructure Provider) to form multiple separate E2E (End-to-End) logical networks on a shared common physical infrastructure, and deliver different types of services according to the multiple requests and business needs of each vertical business.

[0015] 5G network slicing spans multiple parts of the 5G network, including 5G RAN (Radio Access Network), 5G TN (Transport Network), and 5G CN (Core Network). 5G network slicing can be roughly divided into two types: 1) RAN slicing, including dynamic RAN construction and slice-oriented wireless resource virtualization and scheduling mechanism; 2) E2E slicing (i.e. core network slicing), which embeds vertical VN (Virtual Network) into a shared physical infrastructure.

[0016] The core problem of E2E slicing implementation in 5G networks is to embed the virtual network composed of VNFs to support the vertical service system into the physical infrastructure. Some works use SFC (Service Function Chain) with a chain structure to represent network slices. The chain structure can only support a limited number of applications. Therefore, a more general situation is to use virtual networks to represent the end-to-end structure of customized network slices, so that more types of applications can be supported. Virtual networks are generally represented by graph data structures. Implementing a network slice instance mainly includes two steps: 1) embedding virtual networks by locating virtual nodes and virtual links; 2) establishing traffic steering rules by the SDN controller to configure switches to route traffic accordingly. Among them, the first step of embedding virtual networks into physical infrastructure is the widely studied virtual network embedding (VNE) problem, and there are many algorithms. There are two types of typical VNE algorithms: 1) The first type of algorithm divides the embedding process into two stages, first completing the mapping of virtual nodes and then completing the mapping of virtual links; 2) The second type of algorithm completes the mapping of virtual nodes and virtual links at the same time.

[0017] After the network slice is embedded, it is necessary to monitor the performance of the network slice after deployment. The parameters that need to be monitored include: available bandwidth, end-to-end delay, packet arrival rate, throughput, buffer size, etc. Due to the influence of the deployment strategy of the virtual network, the use of underlying physical network resources may be unbalanced, resulting in the rejection of network slice deployment, thereby reducing the utility of the network. Therefore, it is necessary to collect fine-grained underlying network status information in real time to locate faults and dynamically adjust the network slice deployment strategy. Traditional network measurement methods, such as passive detection and active detection methods, have the problems of coarse granularity and poor real-time performance. INT (In-band Network Telemetry) is an idealized network monitoring technology that overcomes the shortcomings of traditional telemetry systems and can collect network status information when user traffic data packets traverse network physical components, thereby greatly improving the performance of the telemetry system. Therefore, the solution of using in-band network telemetry for network slice performance monitoring is an effective strategy.

[0018] In INT technology, the selection of INT monitoring flow sets often affects the value of the collected information. Real-time and fine-grained network status information needs to be collected to have higher value. There are currently some INT monitoring flow generation schemes, such as: 1) Monitoring flow generation schemes based on network topology, such as strategies based on Depth First Search (DFS) and Euler-trail; 2) Monitoring flow generation schemes based on minimizing telemetry load, such as strategies that minimize the number of generated monitoring flows and strategies that minimize the difference in the number of telemetry items of each generated monitoring flow. However, these monitoring flow generation schemes are not oriented to specific businesses. Therefore, a fine-grained, real-time INT monitoring solution that can be oriented to each network slice is urgently needed.

[0019] To this end, the present invention proposes a network slice monitoring method, system and storage medium based on in-band network telemetry.

[0020] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0021] The following describes a network slice monitoring method, system, and storage medium based on in-band network telemetry according to an embodiment of the present invention with reference to the accompanying drawings.

[0022] Figure 2 A network slice monitoring method based on in-band network telemetry according to an embodiment of the present invention is characterized in that the method includes:

[0023] S1, obtains the underlying physical network information and resource requests for multiple network slices.

[0024] Specifically, the underlying physical network information and resource requests of multiple network slices can be obtained through the slice orchestration management controller. Among them, the underlying physical network information may include the topology of the underlying physical network, the CPU computing capacity of the node, the bandwidth capacity of the link, the latency of the link, etc.; the resource request of the network slice may include the resource request of the network slice virtual node and the resource request of the virtual link, which may specifically include: the CPU computing resource request of the virtual node, the bandwidth resource request of the virtual link, the latency request of the virtual link, etc.

[0025] S2, determines the deployment order of each network slice according to the resource request, and deploys multiple network slices on the underlying physical network in the deployment order according to the underlying physical network information.

[0026] Specifically, the slice orchestration management controller can determine the deployment order of each network slice according to the resource request, and deploy multiple network slices on the underlying physical network according to the deployment order based on the underlying physical network information.

[0027] Specifically, the slice orchestration management controller can calculate the comprehensive resources requested by each network slice according to the resource request, and then determine the deployment order according to the amount of comprehensive resources requested; optionally, network slices with more requested resources can be deployed first. When deploying network slices, the slice orchestration management controller can deploy network slices on the underlying physical network in sequence according to the determined network slice deployment order, facing the deployment requirements of network slices with higher service security levels.

[0028] In this embodiment, due to the high security requirements of the business, the deployment strategy of the network slice is restricted, and VNFs (virtualized network functions) from different network slices are not allowed to be deployed on the physical nodes of the same underlying physical network. In this scenario, any physical node can only deploy one or more VNFs from a certain network slice, and the remaining unoccupied physical resources are wasted; at the same time, due to the different importance of nodes in the network topology, nodes with higher importance often have a greater impact on the network. Therefore, the deployment strategy of the network slice should comprehensively consider minimizing the waste of physical resources and minimizing the importance of topological resources of the allocated physical nodes. The sum of the weights of the two factors in the strategy is one, and can be adjusted dynamically to target a variety of application scenarios.

[0029] S3 plans detection paths for each successfully deployed network slice and generates in-band network telemetry INT probes.

[0030] Specifically, the INT framework can be used to plan detection paths and generate in-band network telemetry INT probes for each successfully deployed network slice. As an example, the INT framework uses a Hamiltonian path-based path planning algorithm to plan a dedicated detection path for each successfully deployed network slice, and the detection path can cover all physical nodes that need to be detected.

[0031] Specifically, since the VNF of a network slice is often deployed on multiple physical nodes of the underlying physical network, the detection path needs to cover these physical nodes to ensure the integrity of the telemetry information; at the same time, in order to ensure the reliability of the telemetry process, the overall risk of the detection path should be the lowest. To this end, the INT framework can use a Hamiltonian path-based path planning algorithm to plan the detection path. The specific algorithm process is as follows: 1) Determine the set of physical nodes that need to be detected based on the VNF deployment of the network slice; 2) Generate a fully connected graph on the detection node set, and the weight of the edge in the fully connected graph is the shortest path length between the corresponding nodes of the original underlying physical network; 3) Find a Hamiltonian path on the fully connected graph; 4) Replace each edge in the Hamiltonian path with the shortest path between the corresponding nodes of the original underlying physical network to obtain the detection path. By using a Hamiltonian path-based path planning algorithm, both coverage requirements and reliability requirements can be guaranteed.

[0032] In this embodiment, INT technology can provide detailed and real-time network status information to the control plane (including the above-mentioned slice orchestration management controller and INT framework) by adding network status metadata to the INT probe at line rate on the data plane, where the bottom layer is the underlying physical network composed of numerous hosts and programmable switches. INT technology can be used to monitor SDN networks, optical networks (Optical Network) and industrial wireless networks (Instructive Wireless Network). In the INT system architecture, network managers can define the processing flow of data packets inside the device by configuring programmable switching devices and encapsulate the information of the underlying physical network into the INT probe. In addition, INT supports each data flow and real-time traffic monitoring. This feature makes INT a suitable tool to measure the performance of real-time network slicing deployments.

[0033] S4, according to the detection path, sends the INT probe to the underlying physical network and forwards it in the underlying physical network, so as to use the INT probe to monitor the network slice and obtain INT telemetry information.

[0034] In one embodiment of the present invention, telemetry items and telemetry frequencies of each network slice can also be obtained, and a telemetry identifier of each network slice can be generated; wherein, when the network slice is monitored using an INT probe, an INT header is added to the corresponding INT probe according to the telemetry items and telemetry frequencies through the INT source node, wherein the INT source node is the starting point of the corresponding detection path.

[0035] Among them, telemetry items may include: available bandwidth, end-to-end delay, packet arrival rate, throughput, buffer size, etc. Telemetry frequencies may include 1Hz, 2Hz, 5Hz, etc. The telemetry identifier facilitates the distinction of telemetry requirements for different network slices.

[0036] Specifically, the INT framework contains three types of INT switches: INT source node, INT transit node, and INT sink node. INT probes can be INT telemetry packets, which can be generated by a network controller and use the SR (Source Routing) function to predefine the forwarding path. INT probes do not perform three-layer IP forwarding in the underlying physical network, but forward according to the output port defined by the SR label stack, so that INT probes can be forwarded according to the predefined path, which provides an effective method for controlling the path traversed by the INT probe. The structure of the SR label stack is as follows: Figure 3 shown.

[0037] In this embodiment, the programmable underlying physical network on which the present invention is based requires a custom encapsulation format for the INT probe. After the INT probe is generated, an SR label stack is added, and the SR label stack is filled with the output ports of each switching device on the forwarding path of the INT probe. After the INT probe is forwarded to the underlying physical network, the INT source node adds the INT header and INTmetadata stack to form a complete INT probe, and each switching device on the detection path adds the information of the underlying physical network of the node to the INT probe.

[0038] Different INT probes are distinguished by the VLAN ID in the Tag field in the Ethernet header. This field can have 12 bits, so it can support the distinction of up to 4096 different probes. The SR label stack is between the IP header and the UDP header, and the length of the stack is fixed at 512 bits. For each output port, a length of 4 bits is allocated, so each switch can support up to 16 output ports.

[0039] Compared with traditional IP packets, INT probes have two more fields: INT header and INT data. INT header contains all fixed INT parameters and information, with a fixed length of 12 bytes; INT data contains information about each hop of each switch traversed, stored in the INT metadata stack. Among them, the parts that need to be customized include four fields: INTPort, Original Dest Port, O, and INT len. Instruction Bitmap indicates the telemetry information that the device needs to collect. INT metadata stack is a stack of variable length, and each INT telemetry data occupies a size of 4 bytes. It should be noted that the size of the INT probe still needs to follow the limit of the maximum transmission unit in the network, so each INT probe cannot carry too much telemetry information.

[0040] Among them, INT Port: For the UDP protocol, INT requires a special port: INT Port to distinguish it from other protocols;

[0041] Original Dest Port: Since INT Port is needed to represent the data packet to execute the INT protocol, this position replaces the original destination port field in UDP, so a subsequent Original Dest Port is needed to represent the original UDP destination port;

[0042] O: This one-bit field is used to indicate whether the INT packet is sent to the network controller for analysis. A value of 1 in the O field indicates that the packet needs to be sent to the network controller, and a value of 0 in the O field indicates that the packet is just an INT packet that needs to be forwarded.

[0043] INT len: This field represents the total data length of INT header and INT data;

[0044] Instruction Bitmap: Instruction Bitmap has 16 bits in total, of which the last 8 bits are reserved. Each bit in the first 8 bits represents a type of data that needs to be collected, as shown in Table 1:

[0045] Table 1

[0046]

[0047]

[0048] Furthermore, the INT probe adds the INT header and INT metadata stack at the INT source node to form a complete INT probe, adds the information of the underlying physical network of the node to the INT probe at the INT transmit node, removes these two parts at the INT sink node, and restores the original data packet, thereby ensuring that the telemetry information is only forwarded within the telemetry domain. In other words, the INT probe traverses the INT source node, INT transit node, and INT sink node in sequence. Afterwards, the INT sink node forwards the data packet with the INT information removed to the destination corresponding to the original IP address, and reports the telemetry information to the INT framework.

[0049] Specifically, the INT source node first adds an INT header to the corresponding INT probe according to the telemetry item, telemetry frequency, and telemetry identifier. Then, the INT metadata stack is added and the data of the underlying physical network of the node is recorded in the INT metadata stack. Finally, the output port is determined according to the SR label stack, and the probe data packet is forwarded to the corresponding output port. When the INT transit node receives the probe data packet, it collects the corresponding information according to the INT instruction and fills it into the INT metadata stack. After that, the output port is still determined according to the SR label stack, and the data packet is sent to the output port for forwarding. When the INT sink node receives the probe data packet, it first fills the underlying physical network information that needs to be collected into the INT metadata stack according to the INT instruction. After that, all telemetry information is reported to the INT framework. Finally, the INT header and INT metadata stack are deleted, the original data packet is obtained and forwarded to the destination corresponding to the original IP address.

[0050] In this embodiment, the underlying physical network can use the P4 programming language to configure the programmable switch to customize the parsing and processing flow of data packets within the device. The processing of data packets follows the "match-action" processing logic of "key-value". The control plane (including the above-mentioned slice orchestration management controller and in-band network telemetry INT framework) manages the flow table in the processing pipeline through P4Runtime, such as adding or deleting flow table entries. Among them, the P4 language is a target-independent, protocol-independent and reconfigurable SDN underlying physical network programmable language. Network managers can customize the processing flow of switching equipment, thereby greatly expanding the ability of the underlying physical network to process data packets.

[0051] The network slice monitoring method based on in-band network telemetry proposed in the present invention uses in-band network telemetry technology to realize performance monitoring after the network slice is deployed on the underlying physical network. First, the information of the underlying physical network is obtained, and the network slices are arranged according to the comprehensive resources of the slice request, and the deployment order of the network slices is determined, and then the deployment of the network slices is realized according to the goal of maximizing resource utilization efficiency. After the network slices are deployed on the underlying physical network, a dedicated detection path is planned and a dedicated INT probe is generated for each successfully deployed network slice, and forwarded to the underlying physical network. The INT probe performs in-band network telemetry in the corresponding telemetry domain and reports the telemetry information, thereby completing the performance monitoring of the network slice. In this way, the performance information of the fine-grained network slice after deployment can be obtained, and then the deployment of the network slice can be adjusted in time and the fault location can be quickly located according to the collected network slice performance information, so as to better manage the network and improve the utility of the network.

[0052] In one embodiment of the present invention, the deployment order of each network slice is determined according to the resource request, including: determining the total amount of physical resources occupied by each network slice on the underlying physical network and the importance of physical nodes according to the virtual node resource request and the virtual link resource request; obtaining the objective function of each network slice according to the total amount of physical resources and the importance of physical nodes, and obtaining the deployment order according to the objective function.

[0053] As an implementation method, the objective function can be expressed as follows:

[0054]

[0055] Where T(NSR) represents the objective function of network slicing, Represents the total amount of physical resources of the network slice, p u Indicates the importance of the physical node of the network slice. Both α and β are greater than 0, indicating adjustable parameters, and α+β=1; A virtual node representing a network slice Whether it is mapped to a physical node If yes, The value of is 1, otherwise it is 0; u Represents a physical node Is VNF deployed on the is not less than 1, then z u The value is 1, otherwise it is 0.

[0056] Furthermore, the importance of the physical nodes of the network slice can be obtained by the following formula:

[0057] p u =d u '+b u'+a' u ,

[0058]

[0059]

[0060] a' u =a u *(N-1),

[0061] Among them, d u Represents the physical node of the network slice The graph model G of the underlying physical network I In the above equation, N represents the number of physical nodes in the network slice. represents the spacing centrality, σ st (u) indicates passing through a physical node The total number of shortest paths, σ st represents the total number of shortest paths, represents the closeness to the center, Represents a physical node The sum of distances to all physical nodes.

[0062] In this implementation, obtaining the deployment order according to the objective function may include: sorting the objective functions from large to small, and using the sorting result as the deployment order.

[0063] Specifically, the resource request of a network slice is represented by a virtual network and implemented using a graph data structure. A network slice request NSR is G S (N S ,E S ), where N S Represents the set of virtual nodes of the virtual network, E S Represents the collection of virtual links connecting these virtual nodes. A virtual network node is represented by Indicates that each virtual network node represents a VNF instance, such as RUU (Remote Radio Unit), BBU (Base Band Unit), MEC (Mobile Edge Computing) server, Encoder, and Gateway. Each virtual network node There is a parameter Represents a physical resource request; each virtual link is passed To characterize, Indicates its link bandwidth request.

[0064] After determining the resource request of each network slice, the network slices need to be arranged according to the resource request, so as to arrange the deployment order of the slices. The slices with more requested resources should be deployed first. The comprehensive resources requested by a network slice can be calculated according to the following formula:

[0065]

[0066] The underlying physical network is modeled by a graph G I (N I ,E I ), where N I Represents the set of physical nodes of the underlying physical network, E I Represents a collection of physical links on the underlying physical network. For a physical node on the underlying physical network Represents a physical node physical resources (such as computing resources); for a physical link on the underlying physical network Represents the bandwidth capacity of the physical link. The deployment of network slicing is to connect the network slice G S (N S ,E S ) is mapped to the underlying physical network G I (N I ,E I ) is a typical VNE problem.

[0067] When calculating the utilization of physical resources, it is necessary to understand the deployment of VNFs on physical nodes. In order to illustrate the deployment of VNFs on physical nodes of the underlying physical network, a binary variable 1 indicates a virtual node Mapped to physical nodes At the same time, in order to better calculate the efficiency of physical resource utilization, the binary variable z is introduced u , indicating a physical node Whether VNF is deployed on the Not less than 1, z u The value of is 1; otherwise, z u The value of is 0. For the underlying physical network as a whole, the total amount of physical resources occupied is The total amount of resources requested by the network slice is Since the remaining resources of each physical node in the underlying physical network can no longer be used, some physical resources are wasted. The amount of wasted physical resources is Since the total resources requested by the network slice are fixed values, minimizing the wasted physical resources is to minimize

[0068] From the perspective of topological information, the security requirement of a node in the network topology is positively correlated with its importance in the topology. Generally speaking, the more important a node is, the higher its security requirement is. In this embodiment, the node importance index includes: node degree, spacing centrality and proximity centrality. Each index can be calculated as follows:

[0069] 1) Node degree: d u Represents a physical node In G I The higher the degree of a node, the higher its importance. Its normalization can be expressed as N is the number of physical nodes.

[0070] 2) Spacing center degree: σ st (u) indicates passing through a physical node The total number of shortest paths, σ st represents the total number of shortest paths. Its normalization can be expressed as

[0071] 3) Closeness to centrality: Represents a physical node The sum of the distances to all physical nodes. Its normalization can be expressed as a' u =a u *(N-1).

[0072] Therefore, considering the above three indicators, the importance index of the physical node is p u =d u '+b u '+a' u .

[0073] Taking into account the importance of physical nodes and resource utilization, the optimization goal (i.e., objective function) of the network slicing deployment algorithm is as follows:

[0074]

[0075] This objective function, on the one hand, minimizes the waste of physical resources, and on the other hand, minimizes the importance of topological resources of the allocated physical nodes. In practical applications, different parameters can be set to target different application scenarios.

[0076] In one embodiment of the present invention, planning a detection path for a successfully deployed network slice may include: determining the physical nodes that need to be detected based on the deployment status of the successfully deployed network slice to obtain a set of detection nodes; generating a fully connected graph based on the set of detection nodes, wherein the weight of the edge in the fully connected graph is the shortest path length between the corresponding physical nodes on the underlying physical network; finding a Hamiltonian path on the fully connected graph; and replacing each edge in the Hamiltonian path with the shortest path between the corresponding physical nodes on the underlying physical network to obtain a detection path.

[0077] Specifically, finding a Hamiltonian path on the fully connected graph may include: finding a Hamiltonian circuit on the fully connected graph; and deleting an edge with the largest weight in the Hamiltonian circuit to obtain the Hamiltonian path.

[0078] Specifically, the INT framework can be used to obtain relevant information about the underlying physical network and the status of network slices after deployment, and to clarify the deployment status of each VNF of each network slice on the underlying physical network. After that, the INT framework plans a dedicated detection path for each successfully deployed network slice, and the detection path should cover all physical nodes on the underlying network where the network slice VNF is deployed. And the overall risk weight of the path should be the lowest, so as to ensure the smooth completion of telemetry.

[0079] In order to achieve the goal of covering all physical nodes where the network slice VNF is deployed, the planned detection path needs to include all physical nodes that need to be detected. Similarly, in order to ensure that the overall risk weight of the planned detection path is the lowest, the risk weight between any two detection nodes needs to be minimized. Therefore, in order to achieve the above two goals, a path planning algorithm based on Hamiltonian path can be used.

[0080] Assuming that a network slice has been deployed on the underlying physical network and the distribution of its VNFs deployed on each physical node is known, then these physical nodes where VNFs are deployed can be regarded as the set of physical nodes that need to be detected N T Assume that the detection path generated for the network slice is p, and the set of physical nodes on the path is N p , the set of links on the path is E p , then there are N T ∈N p In order to maximize the reliability of the acquired detection path, the corresponding objective function is:

[0081]

[0082] in, Indicates the reliability of the physical link

[0083] The path planning algorithm based on the Hamiltonian path can obtain a path that covers all physical nodes that need to be detected and has the smallest overall risk weight, thereby providing an ideal INT detection path. The Hamiltonian path is a path that goes from a specified starting point to a specified end point, passing through all other nodes only once. With the help of the Hamiltonian path, the nodes that need to be telemetered can be traversed, ensuring the integrity of the telemetry information. The following is a detailed description of the process of the path planning algorithm based on the Hamiltonian path:

[0084] 1) Obtain the deployment status of a network slice and obtain the set of physical nodes N that need to be detected T ;

[0085] 2) On the underlying physical network G I (N I ,E I ), calculate the set N T Any two points The shortest path and its length;

[0086] 3) In the set N T Construct a fully connected graph G on F , the weight of the edge between nodes is the underlying physical network G I (N I ,E I )The length of the shortest path between corresponding nodes;

[0087] 4) In the fully connected graph G F Find a Hamiltonian circuit l on H , due to the full connection Figure 1 There must be a Hamiltonian circuit, so we can definitely find an l H ;

[0088] 5) In the Hamiltonian circuit l H Delete the edge with the largest weight from the H ;

[0089] 6) The Hamiltonian path p H Each edge in the network uses the underlying physical network G I (N I ,E I ) is replaced by the shortest path between the corresponding nodes to obtain the final detection path p.

[0090] Furthermore, the detection path traversed by each successfully deployed network slice on the underlying physical network is obtained, and the forwarding ports of each switching device on the detection path are recorded. After that, a corresponding INT probe is generated for each successfully deployed network slice, an SR label stack is inserted into the INT probe, and the collected routing information is filled into the stack. During the INT probe forwarding process, the INT probe is directed to which port through continuous stack popping operations.

[0091] The P4 language is used to program the switch on the underlying physical network to customize the processing flow of the switching device for data packets. The programmable switching device performs monitoring on the corresponding detection path, inserts the underlying physical network status information (i.e., telemetry information) that needs to be collected into the INT probe, and reports the telemetry information to the INT database and stores it. At the INTsource node, the INT probe can be modified to insert the INT header and INT metadata stack, and the underlying physical network information of the node is inserted into the INT metadata stack. The location of the INT header is between the SR stack and the UDP header. At the INT transit node, according to the instruction information in the Instruction Bitmap, the corresponding INTmetadata is inserted into the INT metadata stack. The INT sink node receives the probe data packet with INT telemetry data and uploads all telemetry information to the INT server. After that, the INT header and INT metadatastack in the INT probe are deleted, and the original data packet is forwarded to the destination based on the original IP address.

[0092] In summary, the network slice monitoring method based on in-band network telemetry in the embodiment of the present invention first obtains the information of the underlying physical network, and then orchestrates the arrived network slices and deploys them in sequence using corresponding methods. Afterwards, the corresponding INT probe is generated by the INT framework according to the obtained network slice deployment situation, and the corresponding detection path is filled into the SR label stack of the probe, and the probe is forwarded to the underlying physical network. In the corresponding telemetry domain, the underlying physical network uses the INT probe to collect the information that needs to be collected on the path, and generates a telemetry report to the INT framework. In this way, the expansion potential of the programmable underlying physical network is fully utilized, and the performance information of the fine-grained network slice after deployment can be obtained. The control plane can adjust the deployment of the network slice in time and perform rapid fault location based on the collected network slice performance information, so as to better manage the network and improve the effectiveness of the network.

[0093] Figure 4It is a structural block diagram of a network slice monitoring system based on in-band network telemetry according to an embodiment of the present invention.

[0094] like Figure 4 As shown, it is characterized in that the network slice monitoring system 100 based on in-band network telemetry includes: an underlying physical network 110, a slice orchestration management controller 120 and an in-band network telemetry INT framework 130.

[0095] Among them, the slice orchestration management controller 120 is used to obtain the underlying physical network information and the resource requests of multiple network slices, and determine the deployment order of each network slice according to the resource request, and deploy multiple network slices on the underlying physical network 110 according to the deployment order according to the underlying physical network information. The in-band network telemetry INT framework 130 is used to plan the detection path for each successfully deployed network slice and generate the in-band network telemetry INT probe, and send the INT probe to the underlying physical network 110 according to the detection path and forward it on the underlying physical network 110, so as to use the INT probe to monitor the network slice and obtain INT telemetry information.

[0096] In one embodiment of the present invention, the INT framework 130 can also be used to: obtain telemetry items and telemetry frequencies of each network slice, and generate a telemetry identifier for each network slice, so that when the network slice is monitored using an INT probe, an INT header is added to the corresponding INT probe according to the telemetry items and telemetry frequencies through the INT source node, wherein the INT source node is the starting point of the corresponding detection path.

[0097] In one embodiment of the present invention, the resource request includes a virtual node resource request and a virtual link resource request of the virtual network, and the slice orchestration management controller 120 is specifically used to: determine the total amount of physical resources occupied by each network slice on the underlying physical network 110 and the importance of physical nodes based on the virtual node resource request and the virtual link resource request; obtain the objective function of each network slice based on the total amount of physical resources and the importance of physical nodes, and obtain the deployment order based on the objective function.

[0098] Specifically, the objective function is expressed as follows:

[0099]

[0100] Where T(NSR) represents the objective function of network slicing, Represents the total amount of physical resources of the network slice, p u Indicates the importance of the physical node of the network slice. Both α and β are greater than 0, indicating adjustable parameters, and α+β=1; A virtual node representing a network slice Is it mapped to a physical node? If yes, The value of is 1, otherwise it is 0; u Represents a physical node Is VNF deployed on the is not less than 1, then z u The value is 1, otherwise it is 0.

[0101] Furthermore, the importance of the physical nodes of the network slice is obtained by the following formula:

[0102] p u =d u '+b u '+a' u ,

[0103]

[0104]

[0105] a' u =a u *(N-1),

[0106] Among them, d u Represents the physical node of the network slice The graph model G of the underlying physical network I In the above equation, N represents the number of physical nodes in the network slice. represents the spacing centrality, σ st (u) indicates passing through a physical node The total number of shortest paths, σ st represents the total number of shortest paths, represents the closeness to the center, Represents a physical node The sum of distances to all physical nodes.

[0107] In this embodiment, the slice orchestration management controller 120 can be specifically used to: sort the objective functions from large to small, and use the sorting results as the deployment order.

[0108] In a specific embodiment of the present invention, Figure 5 As shown, the slice orchestration management controller 120 includes an information collection module 121, a network slice management module 122 and a network slice deployment module 123.

[0109] In this embodiment, the information collection module 121 is mainly used to periodically collect information about the underlying physical network, including topological information, node information, and link information of the underlying physical network, so as to obtain geometric information, computing resources, and communication resources. Before the network slice is deployed, it is necessary to collect information such as the topological structure of the underlying physical network, the node CPU computing capacity, the link bandwidth capacity, and the link delay; after the network slice is deployed, in addition to the above information, the information collection module 121 also needs to collect the deployment results of the network slice, including whether each slice is mapped successfully, the mapping results of the virtual nodes of each successfully mapped slice, and the mapping results of each virtual link.

[0110] The network slice management module 122 is used to determine the resource request of the network slice and perform the orchestration between network slices. The network slice resource request is represented by a virtual network and implemented using a graph data structure. A network slice request NSR is G S (N S ,E S ), where N S Represents the set of virtual nodes of the virtual network, E S Represents the set of virtual links connecting these virtual nodes. Indicates that each virtual network node represents a virtual VNF instance, such as a virtual remote radio unit, a virtual baseband unit, a virtual mobile edge computing server, a virtual encoder, and a virtual gateway. Each virtual network node There is a parameter Represents a physical resource request; each virtual link is passed To characterize, Indicates its link bandwidth request.

[0111] After determining the request for each network slice, the network slices need to be orchestrated according to the resource request, so as to arrange the deployment order of the network slices. The network slices with more requested resources should be deployed first. The comprehensive resources requested by a network slice can be calculated according to the following formula:

[0112]

[0113] The network slice deployment module 123 is responsible for embedding the orchestrated network slices into the underlying physical network 110. The underlying physical network 110 is modeled by a graph G I (N I ,E I ), where N I represents a set of physical nodes of the underlying physical network 110, E I Represents a collection of physical links of the underlying physical network 110. For a node on the underlying physical network 110 Represents the physical resources of this node (such as computing resources); for a physical link on the underlying physical network Represents the bandwidth capacity of the physical link. The deployment of network slicing is to connect the network slice G S (N S ,E S ) is mapped to the underlying physical network G I (N I ,E I ) is a typical VNE problem. In order to solve the VNE problem, the present invention deploys network slices according to the objective function.

[0114] In the process of deploying network slices, for scenarios with high security requirements, a physical node can only be deployed with one or more VNFs from a certain network slice, and VNFs from different network slices are not allowed to be deployed on a physical node. For any physical node, since only VNFs from a certain network slice can be deployed, the remaining physical resources are wasted. At the same time, due to the different locations of different physical nodes in the network, the topological importance of different physical nodes is different. Therefore, when considering the deployment strategy of network slices, it is necessary to comprehensively consider the utilization rate of physical resources and the topological importance.

[0115] When calculating the utilization of physical resources, it is necessary to understand the deployment of VNFs on the physical node. In order to illustrate the deployment of VNFs on physical nodes, a binary variable is introduced. 1 indicates a virtual node Mapped to physical nodes At the same time, in order to better calculate the efficiency of physical resource utilization, the binary variable z is introduced u , indicating a physical node Whether VNF is deployed on the Not less than 1, z u The value of is 1; otherwise, z u The value of is 0. For the underlying physical network as a whole, the total amount of physical resources occupied is The total amount of resources requested by the network slice is Since the remaining resources of each underlying physical node can no longer be used, some physical resources are wasted. The amount of wasted physical resources is Since the total resources requested by the network slice are fixed values, minimizing the wasted physical resources is to minimize

[0116] From the perspective of topological information, the security requirement of a node in the network topology is positively correlated with its importance in the topology. Generally speaking, the more important a node is, the higher its security requirement is. Node importance indicators include: node degree, spacing centrality, and proximity centrality. Each indicator can be calculated using the following formula:

[0117] 1) Node degree: d u Represents a physical node In G I The higher the degree of a node, the higher its importance. Its normalization can be expressed as N is the number of physical nodes.

[0118] 2) Spacing center degree: σ st (u) indicates passing through a physical node The total number of shortest paths, σ st represents the total number of shortest paths. Its normalization can be expressed as

[0119] 3) Closeness to centrality: It represents the sum of the distances from u to all nodes. Its normalization can be expressed as a' u =a u *(N-1).

[0120] Therefore, considering the above three indicators, the importance index of the physical node is p u =d u '+b u '+a' u .

[0121] Taking into account the importance of physical network nodes and resource utilization, the optimal objective function of the network slicing deployment algorithm is as follows:

[0122]

[0123] The objective function minimizes the waste of physical resources on the one hand, and minimizes the importance of topological resources of the allocated physical nodes on the other hand. In practical applications, different parameters can be set to target different application scenarios.

[0124] In one embodiment of the present invention, the INT framework 130 is specifically used to: determine the physical nodes that need to be detected according to the VNF deployment status of the successfully deployed network slice, and obtain a set of detection nodes; generate a fully connected graph based on the set of detection nodes, wherein the weight of the edge in the fully connected graph is the shortest path length between the corresponding physical nodes on the underlying physical network; find a Hamiltonian path on the fully connected graph; replace each edge in the Hamiltonian path with the shortest path between the corresponding physical nodes on the underlying physical network to obtain a detection path.

[0125] Specifically, when the INT framework 130 finds a Hamiltonian path on the fully connected graph, it is specifically used to: find a Hamiltonian circuit on the fully connected graph; delete the edge with the largest weight in the Hamiltonian circuit to obtain the Hamiltonian path.

[0126] Specifically, see Figure 5 The INT framework 130 may include an INT path planning module 131 , an INT probe generating module 132 , an INT management module 133 and an INT database 134 .

[0127] In this embodiment, the INT path planning module 131 plans an exclusive detection path for each successfully deployed network slice. The INT path planning module 131 first communicates with the information collection module 121 and the network slice deployment module 123 to obtain the underlying physical network related information and the situation after the network slice is deployed, and to clarify the deployment of each VNF of each network slice on the underlying physical network 110. After that, the INT path planning module 131 plans an exclusive detection path for each successfully deployed network slice, and the detection path should cover all nodes on the underlying physical network 110 where the network slice VNF is deployed. And the overall risk weight of the path should be the lowest, so as to ensure the smooth completion of telemetry.

[0128] In order to achieve the goal of covering all physical nodes where the network slice VNF is deployed, the planned detection path needs to include all physical nodes that need to be detected. Similarly, in order to ensure that the overall risk weight of the planned detection path is the lowest, the risk weight between any two detection nodes needs to be minimized. Therefore, in order to achieve the above two goals, a path planning algorithm based on Hamiltonian path can be used.

[0129] Assuming that a network slice has been deployed on the underlying physical layer, and the distribution of its VNFs deployed on each physical node is known, then these physical nodes where VNFs are deployed can be regarded as the set of detection nodes N that need to be detected. T Assume that the exclusive detection path generated by the INT path planning module 131 for the slice is p, and the set of nodes on the path is N p , the set of links on the path is E p, then there are N T ∈N p In order to maximize the reliability of the acquired detection path, the corresponding objective function is:

[0130]

[0131] The path planning algorithm based on the Hamiltonian path can obtain a path that covers all physical nodes that need to be detected and has the smallest overall risk weight of the path, thereby providing an ideal detection path. The Hamiltonian path is a path that goes from a specified starting point to a specified end point, passing through all other nodes only once. With the help of the Hamiltonian path, the physical nodes that need to be telemetered can be traversed, ensuring the integrity of the telemetry information. The following is a detailed description of the process of the path planning algorithm based on the Hamiltonian path:

[0132] 1) Obtain the deployment status of a network slice and obtain the set of physical nodes N that need to be detected T ;

[0133] 2) On the underlying physical network G I (N I ,E I ), calculate the set N T Any two points The shortest path and its length;

[0134] 3) In the set N T Construct a fully connected graph G on F , the weight of the edge between nodes is the underlying physical network G I (N I ,E I )The length of the shortest path between corresponding nodes;

[0135] 4) In the fully connected graph G F Find a Hamiltonian circuit l on H , due to the full connection Figure 1 There must be a Hamiltonian circuit, so we can definitely find an l H ;

[0136] 5) In the Hamiltonian circuit l H Delete the edge with the largest weight from the H ;

[0137] 6) The Hamiltonian path p H Each edge in the network uses the underlying physical network G I (N I ,E I ) is replaced by the shortest path between the corresponding nodes to obtain the final detection path p.

[0138] The INT probe generation module 132 is responsible for generating a corresponding INT probe for each successfully deployed network slice and forwarding it to the underlying physical network 110. The INT probe generation module 132 first communicates with the INT path planning module 131 to obtain the detection path traversed by each successfully deployed network slice on the underlying physical network 110, and records the forwarding ports of each switching device on the detection path. Afterwards, the INT probe generation module 132 generates a corresponding INT probe for each successfully deployed network slice, inserts an SR label stack into the INT probe, and fills the collected routing information into the stack. During the INT probe forwarding process, the INT probe is directed to which port through continuous stack popping operations.

[0139] In this embodiment, the programmable switching devices with INT function enabled in the underlying physical network 110 can be divided into three categories: INT source node, INT transit node and INT sink node. The P4 language is used to program the switch on the underlying physical network 110 to configure the switch and customize the processing flow of the switching device for the INT probe. The programmable switching device performs monitoring on the corresponding detection path, inserts the underlying physical network status information (i.e., telemetry information) to be collected into the INT probe, and reports the telemetry information message to the INT database 134. At the INT source node, the INT probe is modified, the INT header and INT metadata stack are inserted, and the underlying physical network information of the node is inserted into the INTmetadata stack. The location of the INT header is between the SR stack and the UDP header. At the INT transit node, according to the instruction information in the Instruction Bitmap, the corresponding INT metadata is inserted into the INT metadatastack. The INT sink node receives the probe data packet with INT telemetry data and uploads all telemetry information to the INTserver. Afterwards, the INT header and INT metadata stack in the INT probe are deleted, and the original data packet is forwarded to the destination according to the original IP address.

[0140] The INT management module 133 is responsible for communicating with the information collection module 121 to obtain the status of each network slice after deployment, and determine the detection path that each network slice should telemeter. In the P4 programming language, the control plane (including the above-mentioned slice orchestration management controller 120 and the INT framework 130) and the underlying physical network 110 use the P4 runtime to communicate. P4Runtime is a set of protocols based on Protobuf and gRPC frameworks. Through the P4 runtime, the control plane can control devices that support p4. P4 Runtime is also called the southbound protocol, which is used to issue flow tables, rules, etc., such as adding and deleting flow table entries in the flow table. After obtaining the telemetry information reported by the underlying physical network 110, the INT management module 133 stores the telemetry information in the INT database 134, thereby assisting in the deployment of network slices and rapid fault location.

[0141] It should be noted that for other specific implementations of the network slice monitoring system based on in-band network telemetry in an embodiment of the present invention, reference can be made to the network slice monitoring method based on in-band network telemetry in the above-mentioned embodiment of the present invention.

[0142] The network slice monitoring system based on in-band network telemetry of the embodiment of the present invention first obtains the information of the underlying physical network, and then orchestrates the arrived network slices and deploys them in sequence using corresponding methods. Afterwards, the INT framework generates a corresponding INT probe based on the obtained network slice deployment situation, fills the corresponding detection path into the SR label stack of the INT probe, and forwards the INT probe to the underlying physical network. In the corresponding telemetry domain, the underlying physical network uses the INT probe to collect the information that needs to be collected on the path, and generates telemetry information to the INT framework. Therefore, through this monitoring system, the expansion potential of the programmable underlying physical network is fully utilized, and fine-grained performance information of the network slice after deployment can be obtained. Therefore, the network controller can adjust the deployment of the network slice in time and perform rapid fault location based on the collected network slice performance information, so as to better manage the network and improve the effectiveness of the network.

[0143] The present invention also provides a computer-readable storage medium.

[0144] In this embodiment, a computer program is stored on a computer-readable storage medium. When the computer program is executed by a processor, the network slice monitoring method based on in-band network telemetry of the above embodiment is implemented.

[0145] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0146] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0147] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0148] In the description of the present invention, it is to be understood that the terms “center”, “longitudinal”, “lateral”, “length”, “width”, “thickness”, “up”, “down”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inside”, “outside”, “clockwise”, “counterclockwise”, “axial”, “radial”, “circumferential”, etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0149] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0150] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0151] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0152] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A network slice monitoring method based on in-band network telemetry, characterized in that: The method comprises: Obtaining underlying physical network information and resource requests for multiple network slices; Determine a deployment order of each of the network slices according to the resource request, and deploy multiple network slices on the underlying physical network according to the deployment order according to the underlying physical network information; Plan detection paths and generate in-band network telemetry INT probes for each successfully deployed network slice; Sending the INT probe to the underlying physical network according to the detection path and forwarding it in the underlying physical network, so as to monitor the network slice using the INT probe and obtain INT telemetry information; Planning a detection path for a successful network slice deployment includes: Determine the physical nodes that need to be detected according to the VNF deployment status of the successfully deployed network slice, and obtain a detection node set; Generate a fully connected graph based on the detection node set, wherein the weight of the edge in the fully connected graph is the shortest path length between corresponding physical nodes on the underlying physical network; Finding a Hamiltonian path on the fully connected graph; Each edge in the Hamiltonian path is replaced by the shortest path between corresponding physical nodes of the underlying physical network to obtain the detection path.

2. The network slice monitoring method based on in-band network telemetry according to claim 1 is characterized in that: The resource request includes a virtual node resource request and a virtual link resource request of the virtual network, and determining the deployment order of each of the network slices according to the resource request includes: Determine the total amount of physical resources occupied by each of the network slices on the underlying physical network and the importance of the physical nodes according to the virtual node resource request and the virtual link resource request; The objective function of each network slice is obtained according to the total amount of physical resources and the importance of the physical nodes, and the deployment order is obtained according to the objective function.

3. The network slice monitoring method based on in-band network telemetry according to claim 2 is characterized in that: The objective function is expressed by the following formula: ， in, represents the objective function of network slicing, represents the total amount of physical resources of the network slice, Indicates the importance of the physical nodes of the network slice, and are all greater than 0, indicating adjustable parameters, and ; A virtual node representing the network slice Is it mapped to a physical node? If yes, The value of is 1, otherwise it is 0; Represents a physical node Is VNF deployed on the Not less than 1, then The value is 1, otherwise it is 0.

4. The network slice monitoring method based on in-band network telemetry according to claim 3 is characterized in that: The importance of the physical nodes of the network slice is obtained by the following formula: ， ， ， ， in, Represents the physical node of the network slice A graph model of the underlying physical network In the degree, N represents the number of physical nodes of the network slice, , represents the spacing centrality, Indicates passing through a physical node The total number of shortest paths, represents the total number of shortest paths, , represents the closeness to the centrality, Represents a physical node The sum of distances to all physical nodes.

5. The network slice monitoring method based on in-band network telemetry according to claim 3 is characterized in that: The obtaining the deployment order according to the objective function includes: The objective functions are sorted from large to small, and the sorting result is used as the deployment order.

6. The network slice monitoring method based on in-band network telemetry according to claim 1 is characterized in that: The step of finding a Hamiltonian path on the fully connected graph comprises: Finding a Hamiltonian circuit on the fully connected graph; The edge with the largest weight in the Hamiltonian circuit is deleted to obtain the Hamiltonian path.

7. The network slice monitoring method based on in-band network telemetry according to claim 1 is characterized in that: The method further comprises: Obtain telemetry items and telemetry frequencies for each network slice, and generate telemetry identifiers for each network slice; When the network slice is monitored using the INT probe, an INT header is added to the corresponding INT probe according to the telemetry item and the telemetry frequency through the INT source node, wherein the INT source node is the starting point of the corresponding detection path.

8. A network slice monitoring system based on in-band network telemetry, characterized in that: The system comprises: The underlying physical network; A slice orchestration management controller, configured to obtain the underlying physical network information and resource requests of multiple network slices, determine a deployment order of each of the network slices according to the resource requests, and deploy the multiple network slices on the underlying physical network according to the deployment order according to the underlying physical network information; An in-band network telemetry INT framework is used to plan a detection path for each successfully deployed network slice and generate an in-band network telemetry INT probe, and send the INT probe to the underlying physical network according to the detection path and forward it on the underlying physical network, so as to monitor the network slice using the INT probe and obtain INT telemetry information; The INT framework is specifically used for: Determine the physical nodes that need to be detected according to the VNF deployment status of the successfully deployed network slice, and obtain a detection node set; Generate a fully connected graph based on the detection node set, wherein the weight of the edge in the fully connected graph is the shortest path length between corresponding physical nodes on the underlying physical network; Finding a Hamiltonian path on the fully connected graph; Each edge in the Hamiltonian path is replaced by the shortest path between corresponding physical nodes of the underlying physical network to obtain the detection path.

9. The network slice monitoring system based on in-band network telemetry according to claim 8, characterized in that: The resource request includes a virtual node resource request and a virtual link resource request of the virtual network, and the slice orchestration management controller is specifically used to: Determine the total amount of physical resources occupied by each of the network slices on the underlying physical network and the importance of the physical nodes according to the virtual node resource request and the virtual link resource request; The objective function of each network slice is obtained according to the total amount of physical resources and the importance of the physical nodes, and the deployment order is obtained according to the objective function.

10. The network slice monitoring system based on in-band network telemetry according to claim 9, characterized in that: The objective function is expressed by the following formula: ， in, represents the objective function of network slicing, represents the total amount of physical resources of the network slice, Indicates the importance of the physical nodes of the network slice, and are all greater than 0, indicating adjustable parameters, and ; A virtual node representing the network slice Is it mapped to a physical node? If yes, The value of is 1, otherwise it is 0; Represents a physical node Is VNF deployed on the Not less than 1, then The value is 1, otherwise it is 0.

11. The network slice monitoring system based on in-band network telemetry according to claim 10, characterized in that: The importance of the physical nodes of the network slice is obtained by the following formula: ， ， ， ， in, Represents the physical node of the network slice A graph model of the underlying physical network In the degree, N represents the number of physical nodes of the network slice, , represents the spacing centrality, Indicates passing through a physical node The total number of shortest paths, represents the total number of shortest paths, , represents the closeness to the centrality, Represents a physical node The sum of distances to all physical nodes.

12. The network slice monitoring system based on in-band network telemetry according to claim 11, characterized in that: The slice arrangement management controller is specifically used for: The objective functions are sorted from large to small, and the sorting result is used as the deployment order.

13. The network slice monitoring system based on in-band network telemetry according to claim 8, characterized in that: When the INT framework finds a Hamiltonian path on the fully connected graph, it is specifically used to: Finding a Hamiltonian circuit on the fully connected graph; The edge with the largest weight in the Hamiltonian circuit is deleted to obtain the Hamiltonian path.

14. The network slice monitoring system based on in-band network telemetry according to claim 8, characterized in that: The INT framework is also used to: The telemetry items and telemetry frequencies of each network slice are obtained, and a telemetry identifier of each network slice is generated, so that when the network slice is monitored by the INT probe, an INT header is added to the corresponding INT probe according to the telemetry items and the telemetry frequency through the INT source node, wherein the INT source node is the starting point of the corresponding detection path.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the network slice monitoring method based on in-band network telemetry as described in any one of claims 1 to 7 is implemented.

Images