A service system plane availability determination method and apparatus

Abstract

The application discloses a kind of service system plane availability determination method and device, the method includes: the historical data of service system is collected, historical data is analyzed and processed, obtains multiple resource index data and corresponding historical transaction data;According to multiple resource index data and historical transaction data, transaction determination model is constructed;The characteristic value of transaction data corresponding to multiple resource index data is extracted and trained to learn, obtains the plane availability determination model of multiple planes in service system;According to transaction determination model and plane availability determination model, the multiple current resource index data of service system currently obtained in real time is processed, the plane availability of multiple planes is determined, and according to the determination result, flow diversion processing is carried out.The determination result based on the plane availability of the present application, automatically triggers diversion tool, quickly solves the fault, reduces the influence on service system processing service.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a method and apparatus for determining the availability of a business system plane. Background Technology

[0002] Business operations refer to the sum of a series of processes, including production and business activities, and transaction processing, of an enterprise or organization. Business operations include the IT support system (referred to as business support system), business data, business processes, and personnel involved in the business. The business support system is the cornerstone of the business, encompassing various hardware and software IT resources that support business operations, such as network equipment, security equipment, hosts, databases, and middleware. These IT resources are organically combined to jointly undertake a set of tasks that generate specific customer value, thus forming the business support system. Currently, for the sake of canary releases and rapid fault resolution, most business support systems often deploy IT resources in parallel according to the data center plane. Normally, both planes run simultaneously to support business operations. When an anomaly is detected in one plane, all business traffic can be diverted to the normal plane, thereby reducing the impact of the fault and minimizing user perception. Figure 1 This is a schematic diagram of the business system plane usage in existing technology, such as... Figure 1 As shown, when both plane A and plane B are operating normally, the two planes operate in parallel and jointly carry out services. When plane A or plane B is abnormal, the services on the abnormal plane can be stopped by diverting traffic, and the services can be carried out by the normal plane alone.

[0003] Therefore, when a failure occurs during the operation of a business support system, determining which plane is abnormal and quickly resolving the fault is a crucial aspect of current IT system operation and maintenance. To quickly determine whether a plane is abnormal, a plane availability assessment model is needed. Currently, relevant technical solutions for system availability include the following three: Solution 1 is a system availability assessment method based on network element health and weights. Specifically, it obtains the health score of network elements (including nodes and links) based on monitored business system network element failure events according to preset rules; the weight of a node is obtained based on the number of times it appears in the business system path; the weight of a link is obtained based on the associated nodes; and the availability of the business system is assessed according to a predetermined algorithm based on the health score and weight of the network element. Solution 2 is a mobile application availability assessment method based on multi-layer dimensionality reduction. It assesses mobile application availability by assigning scores from 0 to 100 to 25 basic indicators at the indicator layer, calculating the scores of the six dimensions at the dimension layer using predetermined formulas, and then calculating the scores of the three attributes at the attribute layer using formulas. Finally, the final availability score of the target layer is calculated. Scheme 3 is a system availability assessment method based on the association and combination of minimum-dimensional availability models. It uses minimum-dimensional availability model module storage units and combines some availability model modules according to the rules of connection relationships between model modules to generate a model for evaluating system availability.

[0004] However, existing solutions for system availability assessment, whether for network element health or specific business metrics, are based on pre-defined models or calculation formulas. They are only applicable to availability assessment in specific scenarios and have certain limitations. Furthermore, they cannot simultaneously obtain the status of the current business system's data center plane. Therefore, during a fault, the plane availability determination cannot be used to quickly redirect traffic to resolve the fault, failing to meet the principle of timely detection and handling of business system faults. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a business system plane availability determination method and apparatus that overcomes or at least partially solves the above problems.

[0006] According to one aspect of the present invention, a method for determining the availability of a business system plane is provided, comprising:

[0007] Collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data;

[0008] An anomaly determination model is constructed based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data;

[0009] Extract the feature values ​​of the abnormal data corresponding to various resource indicator data, and train and learn based on the feature values ​​of the abnormal data corresponding to various resource indicator data to obtain the planar availability determination model of multiple planes in the business system.

[0010] Based on the anomaly determination model and the plane availability determination model of the multiple planes, the current resource indicator data of the business system obtained in real time are processed to determine the plane availability of the multiple planes in the business system, and traffic diversion processing is performed based on the determination results.

[0011] According to another aspect of the present invention, a business system plane availability determination device is provided, comprising:

[0012] The data acquisition module is used to collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data;

[0013] The model building module is used to build an anomaly determination model based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data; extract the feature values ​​of the anomaly data corresponding to the various resource indicator data; train and learn based on the feature values ​​of the anomaly data corresponding to the various resource indicator data; and obtain the planar availability determination model of multiple planes in the business system.

[0014] The processing module is used to process various current resource indicator data of the business system obtained in real time based on the anomaly determination model and the plane availability determination model of the multiple planes, determine the plane availability of the multiple planes in the business system, and perform traffic diversion processing based on the determination results.

[0015] According to another aspect of the present invention, a computing device is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0016] The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the above-described business system plane availability determination method.

[0017] According to another aspect of the present invention, a computer storage medium is provided, wherein the storage medium stores at least one executable instruction, the executable instruction causing a processor to perform an operation corresponding to the above-described business system plane availability determination method.

[0018] According to a method and apparatus for determining the availability of business system planes based on the present invention, historical data of the business system is collected, analyzed, and processed to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data. An anomaly determination model is constructed based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data. Feature values ​​of the anomaly data corresponding to the various resource indicator data are extracted, and training is performed based on these feature values ​​to obtain a plane availability determination model for multiple planes in the business system. Based on the anomaly determination model and the plane availability determination models for multiple planes, the current resource indicator data of the business system obtained in real time is processed to determine the plane availability of multiple planes in the business system, and traffic redirection is performed based on the determination results. This invention automatically determines anomalies based on various resource indicator data in the business system according to the constructed anomaly determination model, and then automatically determines the plane availability of multiple planes during the current fault period through the trained plane availability determination model. Based on the plane availability determination results, traffic redirection tools are automatically triggered to quickly resolve the fault and reduce the impact on the business system's processing operations.

[0019] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0021] Figure 1 This diagram illustrates the use of the business system plane in the prior art.

[0022] Figure 2 A flowchart of a business system plan availability determination method provided by an embodiment of the present invention is shown;

[0023] Figure 3 A schematic diagram of the sigmoid function provided in an embodiment of the present invention is shown;

[0024] Figure 4 This diagram illustrates a logistic regression curve obtained through machine training according to an embodiment of the present invention.

[0025] Figure 5 This diagram illustrates the anomaly determination model provided in an embodiment of the present invention.

[0026] Figure 6 This diagram illustrates the process of obtaining the centroid point using the k-means algorithm according to an embodiment of the present invention.

[0027] Figure 7a This illustration shows a schematic diagram of training clustering for a planar availability model when taking 4 cluster centroids, as provided in an embodiment of the present invention.

[0028] Figure 7b This diagram illustrates the distribution of one of the test points in historical anomalies provided in an embodiment of the present invention.

[0029] Figure 8 This invention provides a schematic diagram of the structure of a business system planar availability determination device according to an embodiment of the present invention.

[0030] Figure 9 A schematic diagram of the structure of a computing device provided in an embodiment of the present invention is shown. Detailed Implementation

[0031] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0032] Figure 2 A flowchart of an embodiment of a business system plane availability determination method according to the present invention is shown, as follows: Figure 2 As shown, the method includes the following steps:

[0033] Step S210: Collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data.

[0034] In this step, historical data from the business system is collected using data acquisition tools. This historical data can include various hardware and software resources and business data. Hardware and software resources include components such as hosts, instances, application proxies (using HAProxy as an example in this embodiment), and databases (using Redis as an example in this embodiment). Business data includes log volume, domain traffic, and business volume. The collected hardware and software resources and business data are organized according to the data center layout to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data. This data is then compiled into a table (Table 1) and saved to the database. Table 1 shows the various resource indicator data and historical anomaly data corresponding to each resource indicator data saved according to a preset time dimension, where the preset time dimension can be a minute dimension.

[0035]

[0036]

[0037] Table 1 Historical Data Saving Style

[0038] Step S220: Construct an anomaly determination model based on multiple resource indicator data and the historical anomaly data corresponding to each resource indicator data.

[0039] In an optional approach, step S220 further includes: for each plane, analyzing and processing multiple resource indicator data and historical anomaly data corresponding to each resource indicator data to obtain anomaly generation threshold conditions for each resource indicator data; and training and learning based on multiple resource indicator data and anomaly generation threshold conditions for each resource indicator data to obtain anomaly judgment model for each resource indicator data.

[0040] In this step, for each plane, based on the historical anomaly data corresponding to each type of resource indicator data, the machine learning algorithm of logistic regression is used to train and learn each plane as a dimension to obtain an anomaly judgment model for each type of resource indicator data. The anomaly judgment model is used to automatically determine whether the current resource indicator data of the business system obtained in real time can generate anomalies.

[0041] In this embodiment, the anomaly detection model is trained using a logistic regression algorithm. The logistic regression algorithm uses known independent variables to predict the value of a discrete dependent variable. The specific algorithm is described in detail below:

[0042] Suppose the dataset has n independent features, x1 to x2. n Let n be the n features of the sample, where [x1…x n [c0, c2, ..., c] is the input vector. The training process involves determining [c0, c2, ..., c] n-1The value of ] makes the expression have the highest accuracy for multiple input vectors; in order to minimize the error between the predicted value and the true value, a polynomial function is fitted as the objective function of the regression algorithm as shown in equation (1):

[0043] f(x) = c0 + c1x i +…+c n-1 x n (1)

[0044] To ensure that f(x) has good logical judgment properties, it is best to directly express the probability that a sample with feature x is classified into a certain class. This can be achieved by setting a threshold range for f(x). For example, f(x) > 0.5 indicates that x is classified into the positive class, and f(x) < 0.5 indicates that x is classified into the negative class. Furthermore, f(x) is always between [0,1]. In addition, the sigmoid function of the neural network is introduced as the activation function, where the sigmoid function is expressed as follows (2):

[0045]

[0046] in, Figure 3 A schematic diagram of the sigmoid function curve, such as... Figure 3 As shown, the domain of the sigmoid function is all real numbers, the range is between [0,1], and the value at point 0 is 0.5.

[0047] Furthermore, f(x) needs to be transformed into a sigmoid function. p(x) = 1 is defined as the probability that a sample with feature x is assigned to class 1. Then p(x) / [1-p(x)] is defined as the odds ratio, and the logarithm is introduced as follows (3):

[0048]

[0049] Solving for p(x) using equations (1)-(3) yields equation (4):

[0050]

[0051] After obtaining the sigmoid function, the n parameters c in equation (4) are fitted by linear regression for machine training.

[0052] Furthermore, machine learning was performed using a logistic regression algorithm on various resource indicator data and the historical anomaly data corresponding to each resource indicator data. The independent variables were set as: a resource indicator data and the anomaly generation threshold conditions for generating anomalies of that resource indicator data; the dependent variable was set as: whether the resource indicator data generated anomalies.

[0053] Based on the aforementioned independent and dependent variables, a model for determining anomalies in this resource indicator data is obtained through training, as follows: Figure 4 As shown, Figure 4 The trained logistic regression curve is shown. The circular nodes below the curve represent resource indicator data that do not change, while the square nodes above the curve represent resource indicator data that do change.

[0054] Based on the aforementioned anomaly detection model for resource indicator data, this model can automatically determine whether an anomaly can be generated for this type of resource indicator data obtained in real time from the business system. Figure 5 A schematic diagram for using the anomaly detection model, such as... Figure 5 As shown, if the current real-time data of this type of resource indicator from the business system falls below the logistic regression curve, then the current data will not generate an anomaly; if it falls above the logistic regression curve, then it is determined that an anomaly has been generated.

[0055] Furthermore, each type of resource indicator data collected is used to determine whether an anomaly has occurred using its respective anomaly detection model. The generated anomalies are categorized and saved in a database table according to their associated data center location, resource type, time, and anomaly details. The data saving format can be found in Table 2, which shows the anomaly data for various resource indicator data saved by time dimension.

[0056] Business System Abnormal type Abnormal data resource Belonging plane Details of the change time System x host Host CPU utilization IP1 Plane A 99.0% outside the range 2020-10-20 17:15 System x haproxy Backend instance response time IP1_PORT Plane A 1500 is out of range 2020-10-20 17:15 System x Log class log volume Module 1 Plane A 3000 is out of range 2020-10-20 17:15 System x Log class log volume Module 2 Plane B 500 is out of range 2020-10-20 17:15 System x Log class Log keyword 1 Module 1 Plane A 320 is out of range 2020-10-20 17:15 System x Service Gateway Success rate of call System x Plane A 95.6% exceeded the range 2020-10-20 17:15 System x Service Gateway Average call time System x Plane A 3000 is out of range 2020-10-20 17:15 … … … … … … …

[0057] Table 2. Changes in various resource indicator data.

[0058] Step S230: Extract the feature values ​​of the abnormal data corresponding to various resource indicator data, and train and learn based on the feature values ​​of the abnormal data corresponding to various resource indicator data to obtain the planar availability determination model of multiple planes in the business system.

[0059] In an optional approach, step S230 further includes: for each plane, extracting the anomaly data corresponding to the various resource index data of that plane, obtaining the plane availability data when the plane undergoes anomaly based on the anomaly data; extracting the feature values ​​of the anomaly data; training and learning the feature values ​​of the anomaly data and the plane availability data when the plane undergoes anomaly through a clustering algorithm to obtain the plane availability determination model of that plane.

[0060] In this step, based on Table 2, abnormal data corresponding to various resource indicator data generated by each plane are extracted. Based on the abnormal data, the availability of each plane during the historical fault occurrence period is obtained. Feature values ​​of the abnormal data are extracted, and then machine training is performed using the k-means clustering algorithm to obtain the plane availability determination model for each plane. Through the plane availability determination model of each plane, the availability level of the plane corresponding to the abnormality generated when the current fault occurs can be automatically determined, thereby deciding whether it is necessary to resolve the fault by diverting traffic.

[0061] In one alternative approach, the characteristic values ​​of the anomaly data include: anomaly type, level category characteristic value, order of magnitude characteristic value, and anomaly duration; wherein, the anomaly type includes: log-related anomalies, host-related anomalies, instance-related anomalies, application agent-related anomalies, and service gateway-related anomalies.

[0062] Specifically, firstly, the abnormal data corresponding to various resource indicators of the plane is extracted, and the feature values ​​of the abnormal data are extracted, including: abnormal type, resource level to which the abnormality belongs, abnormality level, abnormality quantity classification, abnormality impact on business level, and abnormality duration, etc. Among them, the abnormality type includes log-related abnormalities, host-related abnormalities, instance-related abnormalities, application proxy-related abnormalities (such as HAProxy-related abnormalities), service gateway-related abnormalities, etc.; the resource level to which the abnormality belongs includes core, important, minor, and general, etc.; the abnormality level includes core, important, minor, and general, etc.; the abnormality quantity classification includes less than 1, 1-5, 6-10, 11-30, and more than 30, etc.; the abnormality impact on business level includes core, important, minor, and general, etc.; the abnormality duration is: less than 5 minutes, 5-10 minutes, 10-30 minutes, 30-60 minutes, and more than 60 minutes, etc. Among them, the resource level to which the change belongs, the change level, and the business level affected by the change are level-based feature values; the number of changes is classified as order-of-magnitude feature values.

[0063] In an optional approach, step S230 further includes: performing clustering operations based on the feature values ​​of the aberration data and the plane availability data when the plane undergoes aberration, to obtain multiple cluster centroids; statistically analyzing the hierarchical and order-of-magnitude feature values ​​per unit time, and drawing an aberration distribution map based on the hierarchical and order-of-magnitude feature values ​​per unit time and the multiple cluster centroids; for each aberration data, calculating the distance between the feature value of the aberration data and the multiple cluster centroids, and determining the plane availability determination model of the plane based on the distance to determine the cluster to which the aberration data belongs.

[0064] Specifically, based on the feature values ​​of the abnormal data and the availability data of the plane when the plane is abnormal, the availability of the plane is trained by machine using the k-means algorithm. The basic idea of ​​the k-means algorithm is to iteratively find a partitioning scheme of k clusters such that the overall error obtained when the mean of these k clusters is used to represent the corresponding samples is minimized. That is, the basis of the k-means algorithm is the minimum sum of squared error criterion. The cost function of the k-means algorithm is as follows (5):

[0065]

[0066] In formula (5) Let represent the mean of the i-th cluster.

[0067] To minimize the value of the aforementioned cost function, intuitively speaking, the more similar the samples within each class are, the smaller the squared error between them and the mean of that class. Therefore, by summing the squared errors obtained for all classes, we can verify whether each cluster is optimal when divided into k classes.

[0068] Since the cost function in equation (5) cannot be minimized analytically, the samples can be clustered into k clusters using an iterative method, where k is a pre-set constant. The algorithm first randomly selects k cluster centroids, and then repeats steps 1-2 below until convergence is achieved.

[0069] Step 1: For each sample i, calculate the class it should belong to using equation (6):

[0070] c (i) ∶=argmin j ||x (i) -μ j || 2 (6)

[0071] Step 2: For each class j, repeatedly calculate the centroid of that class using equation (7):

[0072]

[0073] Figure 6 This is a schematic diagram illustrating the process of obtaining the centroid point using the k-means algorithm, as shown below. Figure 6 As shown, three cluster centroids are first randomly selected, and then... Figure 6 The convergence process of step 1-2-3-4 is repeated, and the centroid of this class is repeatedly calculated. Finally, three clusters are obtained, with each cluster corresponding to a centroid.

[0074] Furthermore, historical anomaly data for a single plane within a unit of time (e.g., every minute or every five minutes) is statistically analyzed. The summative values ​​for the corresponding level-based characteristic values ​​and order-of-magnitude characteristic values ​​are calculated, and a distribution map of historical anomalies is plotted. Multiple cluster centroids are selected, representing multiple plane availability levels (high, medium, low, very low, etc.). The distances between all historical anomaly points and multiple centroids are calculated, and the planes are divided into multiple classes according to the distance from largest to smallest. The centroids for each class are then recalculated. This process is repeated until all historical anomaly points are classified into the distribution map of historical anomalies. This yields the availability determination model for the plane. Figure 7a This diagram illustrates the training of a planar availability model for clustering when four cluster centroids are selected. Figure 7a As shown, calculate the distances between all historical anomaly points and the four centroids, divide them into four categories according to the distance from largest to smallest, recalculate the centroids of each category, and repeat the above process until all historical anomaly points are classified into the distribution map of historical anomalies, and then the usability determination model of the plane can be obtained.

[0075] For the anomaly data of each plane generated by the anomaly determination model during the fault, statistical rank-based feature values ​​and order-of-magnitude feature values ​​are calculated and displayed on the distribution map of historical anomalies. By calculating the distance between the plane and different centroids, the planes are sorted from largest to smallest based on this distance. The plane with the smallest distance is the availability level of the current plane. Figure 7b The diagram shows a schematic representation of the distribution of one of the test points in the historical anomalies, as shown below. Figure 7b As shown, sorting by distance, it can be determined that the test point is closest to the one with the lowest usability level. Therefore, based on the plane usability determination model, the current usability of the plane is determined to be extremely low.

[0076] Step S240: Based on the anomaly determination model and the plane availability determination model of multiple planes, process the current resource indicator data of the business system obtained in real time, determine the plane availability of multiple planes in the business system, and perform traffic diversion processing based on the determination results.

[0077] In an optional approach, step S240 further includes: collecting current data from the business system, processing the current data to obtain various current resource indicator data; inputting the various current resource indicator data into the anomaly determination model for determination, and obtaining the anomaly determination result for each resource; for resources whose anomaly determination result indicates anomaly, extracting the feature values ​​of the anomaly data corresponding to the current resource indicator data of the resource, inputting the feature values ​​into the plane availability determination model of the plane to which the resource belongs for processing, obtaining the plane availability determination result of the plane, and performing traffic diversion processing based on the determination result.

[0078] The plane availability determination model in step S230 above can quickly determine the plane availability level at the time of the current failure. In this step, rules can be set in advance, such as automatically triggering traffic diversion tools when the plane availability level is low or extremely low, thereby diverting all traffic of the business system from the plane with low or extremely low availability to the plane with medium or high availability, thereby quickly resolving the failure, reducing the failure handling time and the perception of internal and external users.

[0079] The method in this embodiment collects historical data from the business system, analyzes and processes this data to obtain various resource indicator data and corresponding historical anomaly data for each resource indicator data. An anomaly determination model is constructed based on these data. Feature values ​​of the anomaly data corresponding to the various resource indicator data are extracted and trained to obtain a plane availability determination model for multiple planes within the business system. Based on this model and the plane availability determination model, the current resource indicator data of the business system is processed to determine the plane availability of multiple planes within the business system, and traffic redirection is performed based on the determination results. This invention automatically determines anomalies based on various resource indicator data in the business system according to the constructed anomaly determination model, and then automatically determines the plane availability of multiple planes during the current fault period using the trained plane availability determination model. Based on the plane availability determination results, traffic redirection tools are automatically triggered to quickly resolve the fault and reduce the impact on the business system's processing operations. Meanwhile, this invention uses the collection of various basic layer hardware and software resources and various business layer indicator data in the business system as the basis for historical data, resulting in a wider data collection scope, a more complete training model, and more accurate results. Through machine learning algorithms, it generates anomaly judgment models and planar availability judgment models, which can update the model status in real time and conform to the current status of the business system.

[0080] Figure 8 A schematic diagram of an embodiment of a business system plane availability determination device according to the present invention is shown. Figure 8 As shown, the device includes: a data acquisition module 810, a model building module 820, and a processing module 830.

[0081] The data acquisition module 810 is used to collect historical data from the business system, analyze and process the historical data, and obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data.

[0082] The model building module 820 is used to build an anomaly determination model based on multiple resource indicator data and the historical anomaly data corresponding to each resource indicator data; extract the feature values ​​of the anomaly data corresponding to multiple resource indicator data, and train and learn based on the feature values ​​of the anomaly data corresponding to multiple resource indicator data to obtain the planar availability determination model of multiple planes in the business system.

[0083] In an alternative approach, the model building module 820 is further configured to: for each plane, analyze and process multiple resource indicator data and historical anomaly data corresponding to each resource indicator data to obtain anomaly generation threshold conditions for each resource indicator data; and train and learn based on multiple resource indicator data and anomaly generation threshold conditions for each resource indicator data to obtain anomaly judgment model for each resource indicator data.

[0084] In an optional manner, the model building module 820 is further configured to: for each plane, extract the anomaly data corresponding to the various resource indicators of that plane, obtain the plane availability data when the plane undergoes anomaly based on the anomaly data; extract the feature values ​​of the anomaly data; and train and learn the feature values ​​of the anomaly data and the plane availability data when the plane undergoes anomaly through a clustering algorithm to obtain the plane availability determination model of that plane.

[0085] In one alternative approach, the characteristic values ​​of the anomaly data include: anomaly type, level category characteristic value, order of magnitude characteristic value, and anomaly duration; wherein, the anomaly type includes: log-related anomalies, host-related anomalies, instance-related anomalies, application agent-related anomalies, and service gateway-related anomalies.

[0086] In an optional manner, the model building module 820 is further configured to: perform clustering operations based on the feature values ​​of the aberration data and the availability data of the plane when the aberration occurs, to obtain multiple cluster centroids; statistically analyze the hierarchical and order-of-magnitude feature values ​​per unit time, and draw an aberration distribution map based on the hierarchical and order-of-magnitude feature values ​​and multiple cluster centroids per unit time; for each aberration data, calculate the distance between the feature value of the aberration data and multiple cluster centroids, and determine the cluster to which the aberration data belongs based on the distance to determine the plane availability determination model of the plane.

[0087] The processing module 830 is used to process various current resource indicator data of the business system obtained in real time based on the anomaly judgment model and the plane availability judgment model of multiple planes, to determine the plane availability of multiple planes in the business system, and to perform traffic diversion processing based on the judgment results.

[0088] In an optional manner, the processing module 830 is further configured to: collect current data from the business system, process the current data to obtain various current resource indicator data; input the various current resource indicator data into the anomaly determination model for determination, and obtain the anomaly determination result for each resource; for resources whose anomaly determination result indicates anomaly, extract the feature value of the anomaly data corresponding to the current resource indicator data of the resource, input the feature value into the plane availability determination model of the plane to which the resource belongs for processing, obtain the plane availability determination result of the plane, and perform traffic diversion processing based on the determination result.

[0089] The apparatus of this embodiment collects historical data from the business system, analyzes and processes the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data; constructs an anomaly determination model based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data; extracts feature values ​​of the anomaly data corresponding to the various resource indicator data, and trains and learns based on the feature values ​​of the anomaly data corresponding to the anomaly data of the various resource indicator data to obtain a plane availability determination model for multiple planes in the business system; processes the current resource indicator data of the business system obtained in real time based on the anomaly determination model and determines the plane availability of multiple planes in the business system, and performs traffic diversion processing based on the determination results. This invention automatically determines anomalies based on various resource indicator data in the business system according to the constructed anomaly determination model, and then automatically determines the plane availability of multiple planes during the current fault occurrence period through the trained plane availability determination model. Based on the plane availability determination results, it automatically triggers traffic diversion tools to quickly resolve the fault and reduce the impact on the business system's processing operations.

[0090] This invention provides a non-volatile computer storage medium storing at least one executable instruction that can execute a business system plane availability determination method in any of the above method embodiments.

[0091] Executable instructions can specifically be used to cause the processor to perform the following operations:

[0092] Collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data;

[0093] An anomaly detection model is constructed based on multiple resource indicator data and historical anomaly data corresponding to each resource indicator data.

[0094] Extract the feature values ​​of the abnormal data corresponding to various resource indicator data, and train and learn based on the feature values ​​of the abnormal data corresponding to various resource indicator data to obtain the planar availability determination model of multiple planes in the business system.

[0095] Based on the anomaly detection model and the multi-plane availability detection model, the current resource indicator data of the business system obtained in real time are processed to determine the availability of multiple planes in the business system, and traffic diversion is performed based on the detection results.

[0096] Figure 9 The diagram shows a structural schematic of an embodiment of the computing device of the present invention. The specific embodiments of the present invention do not limit the specific implementation of the computing device.

[0097] like Figure 9 As shown, the computing device may include:

[0098] Processor, Communications Interface, Memory, and Communications Bus.

[0099] The processor, communication interface, and memory communicate with each other via a communication bus. The communication interface is used to communicate with other network elements, such as clients or other servers. The processor executes programs, specifically the relevant steps in one embodiment of the business system plane availability determination method described above.

[0100] Specifically, the program may include program code, which includes computer operation instructions.

[0101] The processor may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The server may include one or more processors of the same type, such as one or more CPUs; or it may include processors of different types, such as one or more CPUs and one or more ASICs.

[0102] Memory is used to store programs. Memory may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive.

[0103] Specifically, the program can be used to cause the processor to perform the following operations:

[0104] Collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data;

[0105] An anomaly detection model is constructed based on multiple resource indicator data and historical anomaly data corresponding to each resource indicator data.

[0106] Extract the feature values ​​of the abnormal data corresponding to various resource indicator data, and train and learn based on the feature values ​​of the abnormal data corresponding to various resource indicator data to obtain the planar availability determination model of multiple planes in the business system.

[0107] Based on the anomaly detection model and the multi-plane availability detection model, the current resource indicator data of the business system obtained in real time are processed to determine the availability of multiple planes in the business system, and traffic diversion is performed based on the detection results.

[0108] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system, or other device. Various general-purpose systems can also be used in conjunction with the teachings herein. The required structure for constructing such systems is apparent from the above description. Furthermore, the embodiments of the present invention are not directed to any particular programming language. It should be understood that the content of the invention described herein can be implemented using various programming languages, and the above description of specific languages ​​is for the purpose of disclosing the best mode of implementation of the invention.

[0109] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0110] Similarly, it should be understood that, in order to simplify the invention and aid in understanding one or more of the various inventive aspects, features of the embodiments of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the above description of exemplary embodiments of the invention. However, this disclosure should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the following claims, inventive aspects lie in fewer than all features of a single foregoing disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into this detailed description, wherein each claim itself is a separate embodiment of the invention.

[0111] Those skilled in the art will understand that modules in the device of the embodiments can be adaptively changed and placed in one or more devices different from that embodiment. Modules, units, or components in the embodiments can be combined into a single module, unit, or component, and further, they can be divided into multiple sub-modules, sub-units, or sub-components. Except where at least some of such features and / or processes or units are mutually exclusive, any combination can be used to combine all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or device so disclosed. Unless expressly stated otherwise, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0112] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the following claims, any of the claimed embodiments can be used in any combination.

[0113] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components according to the embodiments of the present invention. The present invention can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0114] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A method for determining the availability of a business system plane, characterized in that, include: Collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data; Based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data, an anomaly judgment model is constructed by training and learning through a machine learning algorithm of logistic regression, with each plane as the dimension. The independent variables are set as: a resource indicator data and the anomaly generation threshold condition for the resource indicator data to generate anomalies; the dependent variable is set as: whether the resource indicator data generates anomalies. Extract the feature values ​​of the abnormal data corresponding to various resource indicator data, and train and learn based on the feature values ​​of the abnormal data corresponding to various resource indicator data to obtain the planar availability determination model of multiple planes in the business system. Based on the anomaly determination model and the plane availability determination model of the multiple planes, the current resource indicator data of the business system obtained in real time are processed to determine the plane availability of the multiple planes in the business system, and traffic diversion processing is performed based on the determination results. The step of extracting feature values ​​of abnormal data corresponding to multiple resource indicator data, and training and learning based on the feature values ​​of abnormal data corresponding to multiple resource indicator data to obtain a plane availability determination model for multiple planes in the business system, further includes: For each plane, extract the anomaly data corresponding to various resource indicators for that plane, and obtain the plane availability data when the anomaly occurs based on the anomaly data; Extract the feature values ​​of the abnormal data; The feature values ​​of the abnormal data and the plane availability data when the plane changes are trained and learned through a clustering algorithm to obtain the plane availability determination model.

2. The method according to claim 1, characterized in that, The step of constructing an anomaly determination model based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data, using each plane as a dimension, and training and learning through a machine learning algorithm of logistic regression, further includes: For each plane, the various resource indicator data and the historical anomaly data corresponding to each resource indicator data are analyzed and processed to obtain the anomaly generation threshold conditions for each resource indicator data. The model for judging the anomalies of each resource indicator is obtained by training and learning based on the various resource indicator data and the anomaly generation threshold conditions of each resource indicator data.

3. The method according to claim 1, characterized in that, The feature values ​​of the abnormal data include: abnormal type, level category feature value, order of magnitude feature value, and duration of abnormality; The types of anomalies include: log-related anomalies, host-related anomalies, instance-related anomalies, application proxy-related anomalies, and service gateway-related anomalies.

4. The method according to claim 3, characterized in that, The step of training and learning the plane availability determination model by using a clustering algorithm to obtain the feature values ​​of the abnormal data and the plane availability data when the plane is abnormal further includes: Based on the feature values ​​of the abnormal data and the plane availability data when the plane undergoes abnormality, clustering operations are performed to obtain multiple cluster centroids. The graded feature values ​​and the order-of-magnitude feature values ​​are statistically analyzed per unit time, and an anomaly distribution map is drawn based on the graded feature values, the order-of-magnitude feature values, and multiple cluster centroids per unit time. For each abnormal data point, calculate the distance between the feature value of the abnormal data point and multiple cluster centroids. Based on the distance, determine the cluster to which the abnormal data point belongs and determine the plane availability determination model for that plane.

5. The method according to any one of claims 1-4, characterized in that, The step of processing various current resource indicator data of the business system obtained in real time based on the anomaly determination model and the plane availability determination model of the multiple planes, determining the plane availability of the multiple planes in the business system, and performing traffic diversion processing based on the determination results further includes: Collect current data from the business system, process the current data, and obtain various current resource indicator data. Multiple current resource indicator data are input into the anomaly determination model for determination, and anomaly determination results are obtained for each type of resource. For resources whose anomaly determination result indicates an anomaly has occurred, the feature values ​​of the anomaly data corresponding to the current resource indicator data of the resource are extracted. The feature values ​​are then input into the plane availability determination model of the plane to which the resource belongs for processing to obtain the plane availability determination result of the plane. Traffic diversion processing is then performed based on the determination result.

6. A business system planar availability determination device, characterized in that, include: The data acquisition module is used to collect historical data from the business system, analyze and process the historical data to obtain various resource indicator data and historical anomaly data corresponding to each resource indicator data; The model building module is used to construct an anomaly determination model based on the various resource indicator data and the historical anomaly data corresponding to each resource indicator data, taking each plane as a dimension, and training and learning through a machine learning algorithm of logistic regression. The independent variables are: a resource indicator data and the anomaly generation threshold condition for that resource indicator data; the dependent variable is: whether the resource indicator data generates anomalies. Feature values ​​of the anomaly data corresponding to the various resource indicator data are extracted, and the model is trained and learned based on these feature values ​​to obtain a plane availability determination model for multiple planes in the business system. The processing module is used to process various current resource indicator data of the business system obtained in real time based on the anomaly determination model and the plane availability determination model of the multiple planes, determine the plane availability of the multiple planes in the business system, and perform traffic diversion processing based on the determination results. The model building module is further configured to: for each plane, extract the abnormal data corresponding to various resource indicators of that plane, obtain the plane availability data when the plane undergoes an abnormality based on the abnormal data; extract the feature values ​​of the abnormal data; and train and learn the feature values ​​of the abnormal data and the plane availability data when the plane undergoes an abnormality through a clustering algorithm to obtain the plane availability determination model of that plane.

7. The apparatus according to claim 6, characterized in that, The model building module is further used for: For each plane, the various resource indicator data and the historical anomaly data corresponding to each resource indicator data are analyzed and processed to obtain the anomaly generation threshold conditions for each resource indicator data. The model for judging the anomalies of each resource indicator is obtained by training and learning based on the various resource indicator data and the anomaly generation threshold conditions of each resource indicator data.

8. A computing device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction, which causes the processor to perform an operation corresponding to the business system plane availability determination method as described in any one of claims 1-5.

9. A computer storage medium, characterized in that, The storage medium stores at least one executable instruction, which causes the processor to perform the operation corresponding to the business system plane availability determination method as described in any one of claims 1-5.

Images