CN114866431A

CN114866431A - Method and device for predicting SFC network fault based on INT and processor

Info

Publication number: CN114866431A
Application number: CN202210469536.2A
Authority: CN
Inventors: 李铮; 刘庆扬; 王康; 王祥; 毛珊珊; 彭超; 徐书明
Original assignee: Beijing Smartchip Microelectronics Technology Co Ltd; China Gridcom Co Ltd; Shenzhen Zhixin Microelectronics Technology Co Ltd
Current assignee: Beijing Smartchip Microelectronics Technology Co Ltd; China Gridcom Co Ltd; Shenzhen Zhixin Microelectronics Technology Co Ltd
Priority date: 2022-04-28
Filing date: 2022-04-28
Publication date: 2022-08-05

Abstract

The embodiment of the application provides a method, a device, a processor and a storage medium for predicting SFC network faults based on INT. The method comprises the following steps: determining a telemetry frequency and a set of telemetry parameters for the SFC; sending the telemetry frequency and the telemetry parameter group to an SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet; acquiring a parameter value of each telemetry parameter contained in each switch in an INT telemetry data packet; determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group; inputting the SFC network data into a neural network model so as to output the SFC network data of the next time point through the neural network model; and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC output by the classification model aiming at the next time point so as to quickly position the fault position and ensure the network service quality of the SFC.

Description

Method and device for predicting SFC network fault based on INT and processor

Technical Field

The present application relates to the field of network communications, and in particular, to a method, an apparatus, a storage medium, and a processor for predicting an SFC network failure based on INT.

Background

With the development and progress of society, communication networks need to carry more diversified communication services, and different communication services have different demands on network quality. Network users typically communicate with service terminals through a Service Function Chain (SFC).

The Service Function Chain (SFC) is composed of a group of ordered virtual Network Functions (NFV) to provide customized network services, which can improve the network deployment speed and resource utilization rate, and reduce the network construction cost. Software Defined Networking (SDN) provides a new architecture that enables network programmability and further provides on-demand and flexible network services for network services. 5G networks enable more flexible deployment of Service Function Chaining (SFC) based on Network Function Virtualization (NFV) and Software Defined Networking (SDN). Network failures still occur with entirely new network architectures. In the current prior art, in order to quickly locate the position of the fault occurrence and predict the type of the network fault, network data in the network can be measured by a network measurement technology.

At present, the network quality is ensured by monitoring data transmitted by the network based on a Simple Network Management Protocol (SNMP), and rapidly positioning and predicting faults by monitoring the network data. The method collects network basic data such as the number of received bytes, the number of lost packets, the number of errors and the like of the data packets, and cannot meet the fine-grained requirement of high-dynamic data in the network. Moreover, a polling mode is generally adopted for collecting network data, and the collection mode has larger delay, so that the position of the fault can not be quickly determined, and the network fault can not be solved in time.

Disclosure of Invention

The embodiment of the application aims to provide a method, a device, a storage medium and a processor for predicting SFC network faults based on INT.

In order to achieve the above object, a first aspect of the present application provides a method for predicting an SFC network failure based on INT, where the method is applied to an INT server and includes:

determining a telemetry frequency and a telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to the service type corresponding to the SFC, and the telemetry parameter group comprises a plurality of telemetry parameters;

sending the telemetry frequency and the telemetry parameter group to an SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet;

acquiring a telemetry parameter group corresponding to each switch on a network transmission path included in the INT telemetry data packet and a parameter value of each telemetry parameter included in each telemetry parameter group;

determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group;

inputting the SFC network data into a neural network model to output SFC network data for a next time point of a preset time point through the neural network model;

and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

In an embodiment of the application, the set of telemetry parameters includes a queue length, an ingress time, and an egress time of the switch, and the SFC network data includes a processing delay, a propagation delay, and a queue length of the switch; wherein the processing delay is a time difference between an ingress time of the switch and an egress time of the switch, and the propagation delay is a time difference between the ingress time of the switch and an egress time of a switch located immediately above the switch on the network transmission path.

In an embodiment of the application, sending the telemetry frequency and the set of telemetry parameters to an SDN controller to determine a network transmission path of the SFC by the SDN controller and generate the INT telemetry data packet comprises: sending the telemetry frequency and the telemetry parameter group to an SDN controller so that the SDN controller obtains a deployment sequence of each switch in the SFC and determines a network transmission path of the SFC according to the deployment sequence, wherein after the SDN controller obtains the telemetry frequency and the telemetry parameter group, the SDN controller sends a deployment sequence obtaining request to an SFC orchestrator so as to obtain the deployment sequence of each switch returned by the SFC orchestrator according to the deployment sequence obtaining request; and the INT telemetry data packet is generated by combining the telemetry parameters in the telemetry parameter group after the SDN controller generates the INT message according to the INT protocol.

In the embodiment of the application, after an INT telemetry data packet is generated by an SDN controller, the SDN controller sends the INT telemetry data packet to a first switch on a network transmission path, and the first switch transmits the INT telemetry data packet to a next switch according to the deployment sequence of the switches on the network transmission path until the INT telemetry data packet is transmitted to a last switch on the network transmission path; when each switch forwards the INT telemetry data packet, inserting a telemetry parameter value corresponding to the switch into the INT telemetry data packet to obtain a parameter value of each telemetry parameter contained in the telemetry parameter group corresponding to each switch; when the INT telemetry data packet is sent to the first switch, a plurality of telemetry parameter values corresponding to the first switch and INT head data in the INT message are inserted into the INT telemetry data packet, when the INT telemetry data packet is forwarded to the last switch, a plurality of telemetry parameter values corresponding to the last switch are inserted into INT metadata in the INT telemetry data packet, and INT head data in the INT telemetry data packet is deleted.

In an embodiment of the application, the method further comprises a training step of the neural network model, the training step comprising: acquiring historical network data of SFCs at a plurality of time points, wherein the historical network data refers to SFC network data which are arranged before a preset time point in time sequence; sequentially inputting the historical network data of each time point into a neural network model; acquiring predicted network data of a next time point of time points of input historical network data, which is output by the neural network model according to the historical network data; and adjusting the hyper-parameters of the neural network model according to the data error value between the predicted network data and the actual network data corresponding to the historical network data until the optimal hyper-parameters are determined, and determining that the training of the neural network model is finished.

In an embodiment of the application, the method further comprises: after the predicted fault type of the SFC at the next time point is determined through the classification model, the predicted fault type is sent to an SFC orchestrator, so that the deployment sequence of switches in the SFC is adjusted through the SFC orchestrator until the network quality of the SFC meets a preset service level protocol.

In embodiments of the present application, the failure types of the SFC include switch overload and/or network transmission path overload between any two adjacent switches.

A second aspect of the present application provides an apparatus for predicting an SFC network failure based on an INT, comprising:

the data determination module is used for determining the telemetry frequency and the telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to the service type corresponding to the SFC, and comprises a plurality of telemetry parameters;

the data sending module is used for sending the telemetry frequency and the telemetry parameter group to the SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet;

the data acquisition module is used for acquiring the telemetry parameter groups corresponding to the switches on the network transmission path included by the INT telemetry data packet and the parameter values of each telemetry parameter contained in each telemetry parameter group;

the network fault prediction module is used for determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group; inputting the SFC network data into the neural network model to output SFC network data for a next time point of the preset time point through the neural network model; and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

A third aspect of the present application provides a machine-readable storage medium having stored thereon instructions that, when executed by a processor, cause the processor to be configured to perform the above-described method of predicting an SFC network failure based on an INT.

A fourth aspect of the present application provides a processor configured to perform the above-mentioned method for predicting an SFC network failure based on an INT.

By the technical scheme, the telemetry parameter value of the telemetry parameter of each switch on the network transmission path can be obtained, and the SFC network data corresponding to each time point can be determined through the telemetry parameter value, so that the SFC network data can be input into the model to predict the predicted fault type of the SFC at the next time point, the position of the fault can be quickly positioned, the fault of the SFC can be adjusted in real time, and the network service quality of the SFC is ensured.

Additional features and advantages of embodiments of the present application will be described in detail in the detailed description which follows.

Drawings

The accompanying drawings, which are included to provide a further understanding of the embodiments of the disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the embodiments of the disclosure, but are not intended to limit the embodiments of the disclosure. In the drawings:

FIG. 1 is a schematic diagram of an application environment of a method for predicting SFC network failure based on INT according to an embodiment of the application;

FIG. 2 schematically illustrates a flow diagram of a method for predicting SFC network failures based on INT according to an embodiment of the present application;

FIG. 3 schematically illustrates a topology diagram of an SFC that predicts SFC network failures based on INT, in accordance with an embodiment of the present application;

FIG. 4 is a schematic diagram illustrating the structure of INT header for predicting SFC network failure based on INT according to an embodiment of the present application;

fig. 5 is a schematic diagram illustrating an INT message structure after the telemetry parameter value of the switch is inserted based on INT prediction SFC network failure according to an embodiment of the present application;

FIG. 6 is a schematic diagram of an application environment of a method for predicting SFC network failure based on INT according to another embodiment of the present application;

FIG. 7 schematically illustrates a structural diagram of an LSTM network model for SFC network fault prediction based on INT according to an embodiment of the present application;

FIG. 8 is a block diagram schematically illustrating the structure of an apparatus for predicting SFC network failure based on INT according to an embodiment of the present application;

fig. 9 schematically shows an internal structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

To make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the specific embodiments described herein are only used for illustrating and explaining the embodiments of the present application and are not used for limiting the embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The method for predicting the SFC network fault based on the INT can be applied to the application environment shown in FIG. 1. The INT server 101 is in communication with the SDN controller 102, the SFC orchestrator 103 and the switch 104 through a network, the SDN controller 102 is in communication with the INT server 101, the SFC orchestrator 103 and the switch 104 through a network, the SFC orchestrator 103 is in communication with the INT server 101 and the SFC orchestrator 103 through a network, and the switch 104 is in communication with the INT server 101 and the SDN controller 102 through a network. The switch 104 may be a switch cluster composed of a plurality of switches. Different switches may connect different target devices to forward data in the network to the target devices. The target device may be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices.

FIG. 2 is a schematic flow chart diagram illustrating a method for predicting SFC network failure based on INT according to the embodiment of the application. As shown in fig. 2, in an embodiment of the present application, a method for predicting an SFC network failure based on INT is provided, and the present embodiment is mainly exemplified by applying the method to the INT server 101 in fig. 1, and includes the following steps:

step 201, determining a telemetry frequency and a telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to a service type corresponding to the SFC, and the telemetry parameter group includes a plurality of telemetry parameters.

Step 202, sending the telemetry frequency and the telemetry parameter group to an SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet.

And step 203, acquiring the telemetry parameter group corresponding to each switch on the network transmission path included by the INT telemetry data packet and the parameter value of each telemetry parameter included in each telemetry parameter group.

And step 204, determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group.

And step 205, inputting the SFC network data into the neural network model so as to output the SFC network data for the next time point of the preset time point through the neural network model.

And step 206, inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

The SFC may include switches and network transmission paths between every two adjacent switches. Wherein a switch may be referred to as an BMv2 switch. The BMv2 switch is a software switch specifically designed to run the P4 program. The P4 technology can define the processing logic of the switch, so that the switch can realize forwarding under any network protocol, and the application range of the switch can be expanded.

As shown in fig. 3, a topological schematic of an SFC is provided. The SFC in fig. 3 may contain 3 switches, link 1 between switch 1 and switch 2, link 2 between switch 2 and switch 3, and a host connected to each switch. Wherein each link may refer to a network transmission path between every two switches. The bandwidth setting for link 1 may be 30MBps and the bandwidth setting for link 2 may be 50 MBps. Host 1 may send 10MBps of data flow information to host 3.

During the transmission of network data, if the SFC fails, the communication with the terminal is affected. To accurately predict the type of network fault in an SFC, network measurements may be made on the SFC by the INT. Wherein INT refers to in-band network telemetry. The in-band network telemetry technology can customize the type of the collected data and can ensure the accuracy of the collected data. The network failure type of the SFC may include network transmission path overload and/or switch overload. Taking the SFC shown in fig. 3 as an example, the network transmission path failure may be a link 1 overload and a link 2 overload, and the switch overload may be a switch 2 overload. For the SFC shown in FIG. 3, the possible failure types may include 2 ³ And (4) seed preparation.

The SFC orchestrator may send the service class and service priority corresponding to the SFC to the INT server. The INT server can determine the telemetry parameter group of the SFC according to the service type corresponding to the SFC. Wherein the set of telemetry parameters may include a plurality of telemetry parameters. Specifically, the telemetry parameters may be an ID of the switch, a queue depth of the switch, an entry time, an exit time, and the like. For example, the service class corresponding to the SFC may include a and B. If the required telemetry parameters for service class a are the ID, the ingress time, and the egress time of the switch, the set of telemetry parameters for SFC may include the ID, the ingress time, and the egress time of the switch. The INT server may determine the telemetry frequency of the SFC based on the service priority corresponding to the SFC. The higher the service priority SFC needs the higher the telemetry frequency. Wherein the telemetry frequency of the SFC may be 1 time per second.

In one embodiment, sending the telemetry frequency and the set of telemetry parameters to the SDN controller to determine a network transmission path of the SFC by the SDN controller and generate the INT telemetry data packet comprises: sending the telemetry frequency and the telemetry parameter group to an SDN controller so that the SDN controller obtains a deployment sequence of each switch in the SFC and determines a network transmission path of the SFC according to the deployment sequence, wherein after the SDN controller obtains the telemetry frequency and the telemetry parameter group, the SDN controller sends a deployment sequence obtaining request to an SFC orchestrator so as to obtain the deployment sequence of each switch returned by the SFC orchestrator according to the deployment sequence obtaining request; and the INT telemetry data packet is generated by combining the telemetry parameters in the telemetry parameter group after the SDN controller generates the INT message according to the INT protocol.

After determining the telemetry frequency and the set of telemetry parameters for the SFC, the INT server may send the telemetry frequency and the set of telemetry parameters to the SDN controller to determine a network transmission path for the SFC through the SDN controller and generate an INT telemetry data packet. Specifically, after sending the telemetry frequency and the set of telemetry parameters to the SDN controller, the SDN controller may send a deployment order acquisition request to the SFC orchestrator, which may send the deployment order of the SFCs to the SDN controller. The SDN controller may obtain a deployment order for each switch in the SFC. Then, the SDN controller may determine a network transmission path of the SFC according to the obtained deployment order of the switches. The network transmission path may refer to the sequence of the subsequent INT telemetry packets when accessing the switch. As shown in fig. 3, the network transmission path may be switch 1-switch 2-switch 3.

After determining the network transmission path of the SFC, the SDN controller may generate an INT message according to the INT protocol and generate an INT telemetry data packet in accordance with the telemetry parameters in the set of telemetry parameters. The INT message may include, among other things, an INT header and INT metadata. The valid bit of the instruction bitmap in the INT header may be set according to the telemetry parameters in the set of telemetry parameters for subsequent insertion of the value of the telemetry parameters of the switch into the INT telemetry packet according to the valid bit of the instruction bitmap.

As shown in table 1, INT metadata may include switch level, port level, queue level, packet level, and flow table level. Each level in the INT metadata may contain a data type corresponding thereto. The INT metadata may store telemetry parameter values corresponding to the telemetry parameters in a stack.

TABLE 1INT metadata

As shown in fig. 4, the INT header may include a flag bit (Ver ═ 2| D | E | M), Hop ML, remaining Hop count (remaininghop), and Instruction Bitmap (Instruction Bitmap). The INT header contains 4 bytes. The INT header may also include a Specific ID (domain Specific ID), DS instruction (DSInstruction), DS Flags (DS Flags), INT Metadata Stack (INT Metadata Stack). The last INT metadata (LastINTMetadata) is an INT metadata in the INT metadata stack. Here, Hop ML may refer to a metadata length that needs to be inserted every time a switch is passed. The remaining number of hops may refer to the number of times INT telemetry packets remain after each switch pass. The storage unit of the instruction bitmap may be set to 16-bits. Each bit may contain telemetry parameters to be acquired. For example, when the telemetry parameter to be collected is the switch ID, the position of bit corresponding to the switch ID may be set to be valid, so as to insert the parameter value of the corresponding telemetry parameter into INT metadata of the INT telemetry data packet.

In one embodiment, after generating the INT telemetry data packet by the SDN controller, the SDN controller sends the INT telemetry data packet to a first switch on a network transmission path, and the first switch transmits the INT telemetry data packet to a next switch according to a deployment sequence of the switches on the network transmission path until transmitting the INT telemetry data packet to a last switch on the network transmission path; when each switch forwards the INT telemetry data packet, inserting a telemetry parameter value corresponding to the switch into the INT telemetry data packet to obtain a parameter value of each telemetry parameter contained in the telemetry parameter group corresponding to each switch; when the INT telemetry data packet is sent to the first switch, a plurality of telemetry parameter values corresponding to the first switch and INT head data in the INT message are inserted into the INT telemetry data packet, when the INT telemetry data packet is forwarded to the last switch, a plurality of telemetry parameter values corresponding to the last switch are inserted into INT metadata in the INT telemetry data packet, and INT head data in the INT telemetry data packet is deleted.

After generating the INT telemetry packet by the SDN controller, the SDN controller may send the INT telemetry packet to a first switch on a network transmission path. At this time, the first switch may insert INT header data in the INT message into the INT telemetry data packet, and insert a plurality of telemetry parameter values corresponding to the first switch into INT metadata in the INT telemetry data packet.

The first switch may transmit INT telemetry packets to the next switch in the order of deployment of switches on the network transmission path until the INT telemetry packets are transmitted to the last switch on the network transmission path. At this time, a plurality of telemetry parameter values corresponding to the last switch may be inserted into INT metadata in the INT telemetry packet. After inserting the plurality of telemetry parameter values corresponding to the last switch into the INT metadata in the INT telemetry packet, the total length of the INT telemetry packet changes. Therefore, to ensure the safety of the INT telemetry packet during subsequent transmission, the last switch may delete the INT header data in the INT telemetry packet.

When each switch forwards the INT telemetry packet, the telemetry parameter value corresponding to each switch may be inserted into the INT telemetry packet to obtain the parameter value of each telemetry parameter included in the telemetry parameter value corresponding to each switch. Wherein, the corresponding telemetry parameter value of each switch can be determined according to the instruction bitmap set in the INT head. That is, each switch may insert the telemetry parameter value corresponding thereto according to the instruction bitmap set in the INT header. As shown in fig. 5, if the switch ID, the switch queue depth, the entry time, and the exit time in the instruction bitmap of the INT header are valid bits, the switch ID, the switch queue depth, the entry time, and the exit time corresponding to the ith switch may be inserted into the INT metadata in the INT telemetry packet.

For example, if the network transmission path of the SFC includes switch 1, switch 2, and switch 3 in sequence. When the switch 1 forwards the INT telemetry packet to the switch 2, the switch 2 may insert a telemetry parameter value corresponding thereto into the INT telemetry packet according to the setting of the instruction bitmap in the INT telemetry packet. Switch 2 may then forward the INT telemetry packet to switch 3, and switch 3 may insert the telemetry parameter value corresponding thereto into the INT telemetry packet. Therefore, after the INT telemetry data packet is forwarded, the INT telemetry data packet can contain a plurality of telemetry parameter values of each switch on the network transmission path.

The INT telemetry packet is sent to the first switch via User-1 (host), as shown in fig. 6. The INT telemetry packet at this point may contain a header and payload. The first switch may then insert an INT header (INTHeader) into the INT telemetry packet and insert a plurality of telemetry parameter values corresponding to the first switch into Matadata-1 of the INT telemetry packet. Wherein, Matadata-1 stores a plurality of telemetering parameter values corresponding to the first switch. The first switch may continue to forward the INT telemetry packet to the second switch and insert a plurality of telemetry parameter values corresponding to the second switch into Matadata-2 of the INT telemetry packet. The second switch may send the INT telemetry packet to the third switch and insert a plurality of telemetry parameter values corresponding to the third switch into Matadata-3 of the INT telemetry packet. If the third switch is the last switch, the third switch may extract the INT information. The INT information may refer to a plurality of telemetry parameter values, i.e., Matadata-1, Matadata-2, and Matadata-3, for each switch. The third switch may send INT information to INT Server (INT Server) and send data (Header and Payload) in INT telemetry packet after extracting INT information to User-2 (host).

The INT server may obtain a set of telemetry parameters corresponding to each switch on a network transmission path included in the INT telemetry packet, and a parameter value of each telemetry parameter included in each set of telemetry parameters. The parameter value of each telemetry parameter included in the set of telemetry parameters may refer to an identification code of the switch, a queue length, an entry time, and an exit time. After determining the telemetry parameter set corresponding to each switch, the INT server may determine SFC network data at a preset time point according to parameter values included in the telemetry parameter set. The SFC network data may include processing delay, propagation delay, and queue length of the switch, among others.

In one embodiment, the set of telemetry parameters includes a queue length, an ingress time, and an egress time of the switch, and the SFC network data includes a processing delay, a propagation delay, and a queue length of the switch; wherein the processing delay is a time difference between an ingress time of the switch and an egress time of the switch, and the propagation delay is a time difference between the ingress time of the switch and an egress time of a switch located immediately above the switch on the network transmission path.

The processing delay of the switch may be a time difference between an ingress time of the switch and an egress time of the switch. The propagation delay of a switch may be the difference in time between the ingress time of the switch and the egress time of a switch located immediately above the switch on the network transmission path. Wherein the processing delay of the switch is for the same switch. The propagation delay of a switch is for two different switches. Network transmission path overload between switches can be reflected by propagation delays of the switches. Switch overload may be reflected by the processing delay of the switch, or by the queue length of the switch.

In one embodiment, the method further comprises a training step of the neural network model, the training step comprising: acquiring historical network data of SFCs at a plurality of time points, wherein the historical network data refers to SFC network data which are arranged before a preset time point in time sequence; sequentially inputting the historical network data of each time point into a neural network model; acquiring predicted network data of a next time point of time points of input historical network data, which is output by the neural network model according to the historical network data; and adjusting the hyper-parameters of the neural network model according to the data error value between the predicted network data and the actual network data corresponding to the historical network data until the optimal hyper-parameters are determined, and determining that the training of the neural network model is finished.

After determining the SFC network data of the preset time point, the INT server may input the SFC network data to the neural network model to output the SFC network data for a next time point of the preset time point through the neural network model. When the SFC network data of the next time point is predicted by the neural network model, the neural network model may be trained first.

The neural network model may refer to an LSTM network model, among others. The model gradient extinction and gradient explosion can be prevented by predicting through the LSTM network model. Fig. 7 is a block diagram of the LSTM network model. Where c (t-1) may refer to the cell state at time t-1, h (t-1) may refer to the output value at time t-1, and x (t) may refer to the input value at time t. h (t) may refer to the output value at time t, and c (t) may refer to the cell state at time t. a (t) may refer to an input gate, f (t) may refer to a forgetting gate, and o (t) may refer to an output gate. c' (t) may refer to the instantaneous cell state at time t. W _f Weight matrix and bias term for forgetting gate, W _a As weight matrix and offset term of input gate, W _c Weight matrix and offset terms for output gates, W _o Weight matrix sum being cell stateA bias term. Sigma is sigmoid function. The LSTM network model comprises three basic structures of an input gate, a forgetting gate and an output gate, and can realize the protection and control of data. Where the input gate may determine how much new data to add to the neuron state. The forget gate can decide how much old data to discard from the previous neuron state. The output gate may determine the state of the neuron output.

In training the neural network model, first, the INT server may acquire historical network data of the SFC at a plurality of time points. The historical network data refers to SFC network data which are arranged before a preset time point in time sequence. E.g. SFC network data for preset points in time

Its corresponding historical network data may include

Where i may refer to the ith historical network data. The INT server may sequentially input the historical network data of each time point to the neural network model, and then, the INT server may acquire predicted network data of a next time point to the time point of the input historical network data, which is output by the neural network model according to the historical network data.

After acquiring the predicted network data for the next point in time, the INT server may acquire actual network data corresponding to the historical network data and determine a data error value between the predicted network data and the actual network data corresponding to the historical network data. The INT server may determine a weight gradient of the neural network based on the data error value to update the weights of the neural network model. The INT server can determine the prediction performance of the neural network model under the current hyper-parameter through a prediction performance formula. The predicted performance is formulated as

Wherein the content of the first and second substances,

is a data error value between the predicted network data and the actual network data. By the grid searching method, the optimal hyper-parameter can be determined by repeating the steps. And under the condition of determining the optimal hyperparameter, the INT server can determine that the neural network model is trained completely.

After the INT server outputs the SFC network data for the next time point of the preset time point through the trained neural network model, the INT server may input the SFC network data for the next time point into the classification model to obtain the predicted fault type of the SFC for the next time point output by the classification model. Wherein the classification model may employ different kernel functions. In particular, the classification model may refer to an SVM, and the kernel function may be a linear kernel function, a polynomial kernel function, and a gaussian kernel function. The predicted failure type of the SFC may be an overload of a switch in the SFC, an overload of a network transmission path between any two adjacent switches, an overload of a switch and a network transmission path at the same time, or the like.

The classification model may be trained prior to predicting the predicted fault type of the SFC at the next point in time by the classification model. The training steps of the classification model can be as follows: dividing historical network data of the SFC into a training set and a testing set; training the classification model through historical network data in the training set; and inputting the historical network data in the test set into the trained classification model to determine the fault type of the SFC. Wherein the classification models may use different kernel functions. For example, a linear kernel function, a polynomial kernel function, a gaussian kernel function, and the like can be employed. The classification performance of the classification models using different kernel functions can be determined by a performance index formula. The performance index formula may be the following formula (1), formula (2), formula (3), and formula (4).

Table 2 schematically shows the classification performance of classification models using different kernel functions at one performance level. As can be seen from table 2, the classification model using the gaussian kernel function has a larger weighting F1, and the corresponding prediction performance is better. Therefore, a classification model using a gaussian kernel function may be employed in predicting the fault type at the next point in time.

TABLE 2 Classification Performance of Classification models for different Kernel functions

	Gaussian kernel function	Polynomial kernel function	Linear kernel function
				Weighted accuracy	0.9902	0.9713	0.9871
Weighted precision ratio	0.9661	0.9064	0.9426
				Weighted recall	0.9653	0.8958	0.9410
Weighting F1	0.9653	0.8937	0.9405

In one embodiment, the method further comprises: after the predicted fault type of the SFC at the next time point is determined through the classification model, the predicted fault type is sent to an SFC orchestrator, so that the deployment sequence of switches in the SFC is adjusted through the SFC orchestrator until the network quality of the SFC meets a preset service level protocol.

After determining the predicted fault type for the SFC at the next time point through the classification model, the INT server may send the predicted fault type to the SFC orchestrator to adjust a deployment order of switches in the SFC through the SFC orchestrator until a network quality of the SFC meets a preset service level protocol. Wherein the preset service level agreement refers to an SLA agreement. An SLA agreement refers to a service level agreement that meets the needs of a user and is able to define the user's desired service level.

In one embodiment, the failure types of the SFC include switch overload and/or network transmission path overload between any two adjacent switches.

By the technical scheme, the telemetry parameter value of the telemetry parameter of each switch on the network transmission path can be obtained, and the SFC network data corresponding to each time point can be determined through the telemetry parameter value, so that the SFC network data can be input into the model to predict the predicted fault type of the SFC at the next time point, the position of the fault can be quickly positioned, the network fault of the SFC can be adjusted in real time, and the network service quality of the SFC can be ensured.

FIG. 2 is a flow diagram illustrating a method for predicting an SFC network failure based on an INT, according to an embodiment. It should be understood that, although the steps in the flowchart of fig. 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

In one embodiment, as shown in fig. 8, an apparatus for predicting SFC network failure based on INT is provided, which includes a data determination module 801, a data transmission module 802, a data acquisition module 803, and a network failure prediction module 804, wherein:

the data determining module 801 is configured to determine a telemetry frequency and a telemetry parameter set of the SFC, where the telemetry parameter set is determined according to a service class corresponding to the SFC, and the telemetry parameter set includes a plurality of telemetry parameters.

A data sending module 802 for sending the telemetry frequency and the set of telemetry parameters to the SDN controller to determine a network transmission path of the SFC by the SDN controller and generate an INT telemetry data packet.

And a data obtaining module 803, configured to obtain the set of telemetry parameters corresponding to each switch on the network transmission path included in the INT telemetry data packet, and the parameter value of each telemetry parameter included in each set of telemetry parameters.

The network fault prediction module 804 is configured to determine SFC network data at a preset time point according to parameter values included in the telemetry parameter set; inputting the SFC network data into a neural network model to output SFC network data for a next time point of a preset time point through the neural network model; and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

By the technical scheme, the telemetry parameter value of the telemetry parameter of each switch on the network transmission path can be acquired, and the SFC network data corresponding to each time point can be determined through the telemetry parameter value, so that the SFC network data can be input into the model to predict the predicted fault type of the SFC at the next time point, the position of the fault can be rapidly positioned, the fault of the SFC can be adjusted in real time, and the network service quality of the SFC can be ensured.

The device for predicting the SFC network fault based on the INT comprises a processor and a memory, wherein the data determining module, the data sending module, the data acquiring module, the network fault predicting module and the like are stored in the memory as program units, and the processor executes the program modules stored in the memory to realize corresponding functions.

The processor comprises a kernel, and the kernel calls the corresponding program unit from the memory. The kernel can be provided with one or more than one, and the method for predicting the SFC network fault based on the INT is realized by adjusting the kernel parameters.

The memory may include volatile memory in a computer readable medium, Random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.

The embodiment of the application provides a storage medium, wherein a program is stored on the storage medium, and the program realizes the method for predicting the SFC network fault based on the INT when being executed by a processor.

The embodiment of the application provides a processor, wherein the processor is used for running a program, and the method for predicting the SFC network fault based on the INT is executed when the program runs.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 9. The computer device includes a processor a01, a network interface a02, a memory (not shown), and a database (not shown) connected by a system bus. Wherein processor a01 of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises an internal memory a03 and a non-volatile storage medium a 04. The nonvolatile storage medium a04 stores an operating system B01, a computer program B02, and a database (not shown in the figure). The internal memory a03 provides an environment for the operation of the operating system B01 and the computer programs B02 in the non-volatile storage medium a 04. The database of the computer device is used for storing data such as telemetry parameters and parameter values of the telemetry parameters. The network interface a02 of the computer device is used for communication with an external terminal through a network connection. The computer program B02, when executed by the processor a01, implements a method of predicting SFC network failures based on INT.

Those skilled in the art will appreciate that the architecture shown in fig. 9 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

The embodiment of the application provides equipment, the equipment comprises a processor, a memory and a program which is stored on the memory and can run on the processor, and the following steps are realized when the processor executes the program: determining a telemetry frequency and a telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to the service type corresponding to the SFC, and the telemetry parameter group comprises a plurality of telemetry parameters; sending the telemetry frequency and the telemetry parameter group to an SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet; acquiring a telemetry parameter group corresponding to each switch on a network transmission path included in the INT telemetry data packet and a parameter value of each telemetry parameter included in each telemetry parameter group; determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group; inputting the SFC network data into the neural network model to output SFC network data for a next time point of the preset time point through the neural network model; and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

The present application further provides a computer program product adapted to perform a program for initializing the following method steps when executed on a data processing device: determining a telemetry frequency and a telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to the service type corresponding to the SFC, and the telemetry parameter group comprises a plurality of telemetry parameters; sending the telemetry frequency and the telemetry parameter group to an SDN controller so as to determine a network transmission path of the SFC through the SDN controller and generate an INT telemetry data packet; acquiring a telemetry parameter group corresponding to each switch on a network transmission path included in the INT telemetry data packet and a parameter value of each telemetry parameter included in each telemetry parameter group; determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group; inputting the SFC network data into a neural network model to output SFC network data for a next time point of a preset time point through the neural network model; and inputting the SFC network data of the next time point into the classification model, and acquiring the predicted fault type of the SFC of the next time point, which is output by the classification model.

In one embodiment, after generating the INT telemetry data packet by the SDN controller, the SDN controller sends the INT telemetry data packet to a first switch on a network transmission path, and the first switch transmits the INT telemetry data packet to a next switch according to a deployment sequence of the switches on the network transmission path until transmitting the INT telemetry data packet to a last switch on the network transmission path; when each switch forwards the INT telemetry data packet, inserting a telemetry parameter value corresponding to the switch into the INT telemetry data packet to obtain a parameter value of each telemetry parameter contained in the telemetry parameter group corresponding to each switch; when the INT telemetering data packet is transmitted to the first switch, a plurality of telemetering parameter values corresponding to the first switch and INT head data in an INT message are inserted into the INT telemetering data packet, when the INT telemetering data packet is forwarded to the last switch, a plurality of telemetering parameter values corresponding to the last switch are inserted into INT metadata in the INT telemetering data packet, and INT head data in the INT telemetering data packet is deleted.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

Computer-readable media, including both permanent and non-permanent, removable and non-removable media, may implement the information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in the process, method, article, or apparatus that comprises the element.

The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for predicting SFC network failures based on INT, which is applied to an INT server and comprises the following steps:

sending the telemetry frequency and the set of telemetry parameters to an SDN controller to determine a network transmission path of the SFC by the SDN controller and generate an INT telemetry data packet;

acquiring a telemetry parameter group corresponding to each switch on the network transmission path included by the INT telemetry data packet and a parameter value of each telemetry parameter included in each telemetry parameter group;

inputting the SFC network data to a neural network model to output SFC network data for a next time point of the preset time points through the neural network model;

inputting the SFC network data of the next time point into a classification model, and obtaining the predicted fault type of the SFC of the next time point output by the classification model.

2. The method of INT-based prediction of SFC network failures of claim 1, wherein said set of telemetry parameters comprises queue length, ingress time and egress time of a switch, said SFC network data comprises processing delay, propagation delay and queue length of a switch;

wherein the processing delay is a time difference between an ingress time of the switch and an egress time of the switch, and the propagation delay is a time difference between the ingress time of the switch and an egress time of a switch located immediately above the switch on the network transmission path.

3. The method of claim 1, wherein the sending the telemetry frequency and the set of telemetry parameters to an SDN controller to determine a network transmission path of an SFC by the SDN controller and generating an INT telemetry packet comprises:

sending the telemetry frequency and the telemetry parameter group to an SDN controller so that the SDN controller obtains a deployment sequence of each switch in the SFC and determines a network transmission path of the SFC according to the deployment sequence, wherein after the SDN controller obtains the telemetry frequency and the telemetry parameter group, the SDN controller sends a deployment sequence obtaining request to an SFC orchestrator so as to obtain the deployment sequence of each switch returned by the SFC orchestrator according to the deployment sequence obtaining request;

and the INT telemetry data packet is generated by combining the telemetry parameters in the telemetry parameter group after the SDN controller generates an INT message according to an INT protocol.

4. The method of claim 3, wherein after the INT telemetry data packet is generated by the SDN controller, the SDN controller sends the INT telemetry data packet to a first switch on the network transmission path, and the first switch transmits the INT telemetry data packet to a next switch according to a deployment order of the switches on the network transmission path until the INT telemetry data packet is transmitted to a last switch on the network transmission path; when each switch forwards the INT telemetry data packet, inserting a telemetry parameter value corresponding to the switch into the INT telemetry data packet to obtain a parameter value of each telemetry parameter contained in the telemetry parameter group corresponding to each switch;

when the INT telemetry data packet is sent to the first switch, a plurality of telemetry parameter values corresponding to the first switch and INT head data in the INT message are inserted into the INT telemetry data packet, when the INT telemetry data packet is forwarded to the last switch, a plurality of telemetry parameter values corresponding to the last switch are inserted into INT metadata in the INT telemetry data packet, and INT head data in the INT telemetry data packet is deleted.

5. The method of INT-based prediction SFC network failure of claim 1, further comprising a training step of said neural network model, said training step comprising:

acquiring historical network data of SFCs at a plurality of time points, wherein the historical network data refers to SFC network data which are arranged before the preset time point in time sequence;

sequentially inputting the historical network data of each time point into the neural network model;

acquiring predicted network data of a time point next to the time point of the input historical network data, which is output by the neural network model according to the historical network data;

and adjusting the hyper-parameters of the neural network model according to the data error value between the predicted network data and the actual network data corresponding to the historical network data until the optimal hyper-parameters are determined, and determining that the training of the neural network model is finished.

6. The method for SFC network failure based on INT prediction according to claim 1, characterized in that said method further comprises:

after the predicted fault type of the SFC at the next time point is determined through the classification model, the predicted fault type is sent to an SFC orchestrator so as to adjust the deployment sequence of the switches in the SFC through the SFC orchestrator until the network quality of the SFC meets a preset service level protocol.

7. The method of INT-based prediction of SFC network failures according to claim 1, wherein the failure type of SFC comprises switch overload and/or network transmission path overload between any two adjacent switches.

8. An apparatus for predicting SFC network failures based on INT, the apparatus comprising:

the data determination module is used for determining the telemetry frequency and the telemetry parameter group of the SFC, wherein the telemetry parameter group is determined according to the service type corresponding to the SFC, and the telemetry parameter group comprises a plurality of telemetry parameters;

a data sending module, configured to send the telemetry frequency and the set of telemetry parameters to an SDN controller, so as to determine a network transmission path of an SFC through the SDN controller, and generate an INT telemetry data packet;

a data acquisition module, configured to acquire a set of telemetry parameters corresponding to each switch on the network transmission path included in the INT telemetry data packet, and a parameter value of each telemetry parameter included in each set of telemetry parameters;

the network fault prediction module is used for determining SFC network data of a preset time point according to parameter values contained in the telemetry parameter group; inputting the SFC network data to a neural network model to output SFC network data for a next time point of the preset time points through the neural network model; inputting the SFC network data of the next time point into a classification model, and obtaining the predicted fault type of the SFC of the next time point output by the classification model.

9. A machine-readable storage medium having instructions stored thereon, which when executed by a processor cause the processor to be configured to perform the method of predicting SFC network failures based on INT according to any one of claims 1 to 7.

10. A processor configured to perform the method of INT-based prediction of SFC network failure according to any of claims 1 to 7.