Customer service work order management system based on network employer service

Abstract

The invention discloses a customer service work order management system based on network employer service, which relates to the technical field of work order management and comprises a multifunctional service module, a distributed management module, an elastic module, an operation and maintenance monitoring module and a real-time notification module. The multifunctional service module comprises a user service unit, a work order service unit, a communication service unit, a knowledge base service unit and a data analysis unit, and is used for serving a network employer; the distributed management module is used for verifying user validity, determining user legality and creating a work order based on a legal user; and the elastic module is used when the customer service work order service response is overtime. According to the invention, omnibearing real-time monitoring and deep analysis are carried out on the system, discovered problems and work order state change conditions are pushed to related personnel in time, timely transmission of information is ensured, failure retry is carried out based on an exponential backoff algorithm, the reliability of notification is ensured, and notification loss caused by network fluctuation is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of work order management, and specifically relates to a customer service work order management system based on network employer services. Background Art

[0002] In the era of the booming development of the current network economy, it has become normal for employers to seek various services through network platforms. However, there are many problems in the existing customer service work order management systems when dealing with matters related to network employer services. In terms of verifying the validity of users, the existing systems have simple methods and poor security. They mostly use static password verification, which is vulnerable to brute force attacks and password leakage risks, and cannot accurately determine the legitimacy of users. When there is a problem of response timeout in the customer service work order service, the existing systems cannot react in time. The call chain between services is not set with a circuit breaker protection. Once a service fails, a large number of requests will pile up, consuming system resources, and ultimately causing the performance of the entire system to decline or even collapse, seriously affecting the continuity and stability of the service. The monitoring of system performance indicators is not comprehensive, only some key indicators are concerned, and there is a lack of real-time nature. In response to this, we propose a customer service work order management system based on network employer services. Summary of the Invention

[0003] To solve the above technical problems, a customer service work order management system based on network employer services is provided. This technical solution solves the problems of weak user legitimacy verification mechanism, lack of automatic circuit breaker mechanism, and imperfect performance indicator monitoring.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is: a customer service work order management system based on network employer services, including: a multi-functional service module, a distributed management module, an elastic module, an operation and maintenance monitoring module, and a real-time notification module;

[0005] The multi-functional service module includes a user service unit, a work order service unit, a communication service unit, a knowledge base service unit, and a data analysis unit, and is used to provide services for network employers;

[0006] The distributed management module is used to verify the validity of users, determine the legitimacy of users, and create work orders based on legitimate users;

[0007] The elastic module automatically melts when the customer service work order service response times out, and dynamically expands microservice instances based on CPU / memory metrics to ensure the stable operation of the system under high load;

[0008] The operation and maintenance monitoring module is used to collect the log data of microservices, store and analyze it, and locate problems; conduct metric monitoring based on Prometheus, collect various performance metrics of the system, including CPU usage rate, memory usage rate, network traffic, monitor and analyze the metrics in real time, discover potential problems in the system, and handle them;

[0009] The real-time notification module pushes the problems found in the operation and maintenance monitoring process and the changes in the work order status, performs failure retries based on the exponential backoff algorithm, and after the retries reach a certain number of times, sends the message to the dead letter queue for manual processing to ensure reliable notifications.

[0010] Preferably, the user service unit is responsible for user identity management, including the identity management of different roles such as customers, customer service representatives, and administrators; performs permission and role assignment based on the role-based access control model, provides a user information query interface, and supports two communication methods: gRPC / REST;

[0011] The work order service unit is used to implement the full life cycle management of work orders, including creation, assignment, escalation, resolution, and closure operations; automatically escalates work orders that have not been processed for a long time through SLA monitoring, provides label and classification management functions, and classifies and retrieves work orders.

[0012] Preferably, the communication service unit supports multi-channel access, including email, SMS, WebSocket, and third-party IM, and communicates with customer service representatives in multiple ways; adopts an intelligent routing mechanism to allocate sessions based on the load of customer service representatives to improve service efficiency;

[0013] The knowledge base service unit implements intelligent search through Elasticsearch, combines IK word segmentation technology for search, and uses NLP technology to process work order content;

[0014] The data analysis unit provides a real-time dashboard to display the customer service response time and resolution rate; predicts the trend of work order volume through predictive analysis for resource allocation; automatically generates reports, generates the reports in PDF format and sends them via Email.

[0015] Preferably, the distributed management module is managed using the Saga pattern. The management steps are as follows: The work order service initiates a transaction to generate a pre-work order; synchronously calls the user service to verify the user status; if the verification passes, the work order is submitted, and if the verification fails, the pre-work order is compensatorily deleted, records transaction events based on Kafka, and ensures eventual consistency based on the event sourcing mechanism.

[0016] Preferably, the management steps within the distributed management module are specifically represented as follows: Symbols are predefined, T represents the work order transaction, P represents the pre-work order, V represents the user status verification result, V = true represents successful verification, V = false represents failed verification, S represents the work order status, S = 0 represents the pre-work order status, S = 1 represents the submitted work order status, E represents the set of transaction events recorded by Kafka, and A represents the sequence of transaction operations under the event sourcing mechanism;

[0017] Work order service initiation event: When the work order service initiates transaction T, a pre-work order P is generated, and at the same time, the work order status S is set to the pre-work order status 0, which is represented by the following formula:

[0018]

[0019] where T start represents the start of transaction T, and P generate represents the generation of the pre-work order;

[0020] Synchronous call to the user service to verify the user status: The work order service synchronously calls the user service to verify the user status and obtains the verification result V, which is represented as:

[0021]

[0022] where user_verify( ) is a function that represents the user service performing the verification operation and returning the verification result;

[0023] Verification steps: If the verification is passed, the work order status S is changed from the pre-work order status 0 to the submitted work order status 1, and the formula is as follows:

[0024]

[0025] If the verification fails, a compensation operation is performed to delete the pre-work order P, and the work order status S is set to the initial status. The formula is as follows:

[0026]

[0027] where P delete represents the deletion of the pre-work order;

[0028] Record transaction events based on Kafka: During the entire transaction processing process, each key step generates a transaction event and is recorded in the event set E of Kafka, which is represented by the formula:

[0029]

[0030] where V ture 、V false 、S submit represent the events of successful verification, failed verification, and work order submission respectively.

[0031] Preferably, the elastic module is fused based on the Hystrix fuse. The specific steps for dynamically expanding microservice instances based on CPU / memory metrics are as follows: First, build monitoring systems, containerization technologies, and container orchestration platform components. The expansion method specifically includes metric collection and monitoring, and analyzing historical data to determine the threshold range; use the Kubernetes HPA automatic scaling policy configuration to perform scaling operations based on the threshold, and finally perform exception handling and rollback mechanisms.

[0032] Preferably, the operation and maintenance monitoring module collects logs based on Fluentd + Elastic Stack. Fluentd is used to collect the log data of each microservice and send the log data to Elastic Stack for storage and analysis. Based on the search and analysis of Elastic Stack, retrieve logs according to the dimensions of time, service name, and log level to locate problems; Fluentd sends the collected log data generated by microservices to Elastic Stack for storage and analysis. When it is necessary to find problems, perform search and analysis based on Elastic Stack according to the dimensions of time, service name, and log level.

[0033] Preferably, the specific steps for the operation and maintenance monitoring module to monitor and analyze are as follows:

[0034] Determine the monitoring metrics and frequencies. Based on requirements, determine the frequencies of the monitoring metrics and monitor based on Prometheus;

[0035] Collect data. For CPU utilization rate, the monitoring tool obtains the ratio of the working time to the total time of the CPU within a unit time; the memory utilization rate is obtained by calculating the ratio of the used memory to the total memory; the network traffic data is obtained from the network interface, including the number of bytes and packets received and sent;

[0036] Store the collected performance metric data in the database. Based on the visualization tool, display the collected performance metric data in the form of charts. Create a line chart to show the changing trends of CPU utilization rate and memory utilization rate over time, and use a bar chart to show the network traffic conditions in different time periods;

[0037] Based on the normal operating range of the system and business requirements, set thresholds for performance metrics. When the monitored metrics exceed the set thresholds, transmit the information to the real-time notification module to alert the management staff.

[0038] Preferably, the method for determining the threshold is to collect the data of various performance metrics of the system within the unit time of normal operation, calculate the mean and standard deviation of the collected data, and set the historical data as x1, x2, …… x n , and the mean calculation formula is:

[0039]

[0040] The calculation formula for the standard deviation s’ is as follows:

[0041]

[0042] The threshold is set by adding a multiple of the standard deviation to the mean;

[0043] The comparison method is to directly compare the collected data with the obtained threshold. If the performance index collected in real time is greater than the obtained threshold, it is determined that the index exceeds the threshold, and information is sent to the real-time notification module.

[0044] Preferably, the backoff algorithm is used in the real-time notification module for failure retry to handle the situation of failed notification sending. The expression is:

[0045] T’ = T base ×(1 + r×random(0,1))×2 n’

[0046] Where T’ represents the current retry interval time, that is, the duration to wait for the next retry after this failure. T base is the initial retry interval time, set to a small value, which is the starting benchmark for the retry interval. r is the randomization factor, a small positive number, which introduces randomness in each retry interval time to avoid network congestion caused by failed requests being retried simultaneously. Due to the existence of the squeezing randomization factor, their retry times are different, reducing the impact on system resources during simultaneous retries. random(0,1) is a random function that returns a value between 0 and 1, including 0 but not including 1. Each time this function is called, a different random number is generated to increase the randomness of the retry interval. n’ represents the number of retries, starting from 0. As the number of retries increases, the n’ term causes the retry interval time to increase exponentially. A maximum number of retries N max and a maximum retry interval time T max are also set to avoid infinite retries and overly long waiting times. That is, when n’ ≥ N max the retries are stopped; when T’ > T max the retry interval time is set to T max .

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0048] The present invention determines the legitimacy of users by verifying user validity, ensuring the security and standardization of system services from the source. Only legitimate users can create work orders, which effectively avoids interference from illegal operations to the system, reduces the generation of invalid work orders, and improves the quality and efficiency of work order processing. The automatic fusing mechanism can quickly cut off faulty services, prevent the spread of faults, and avoid paralysis of the entire system caused by a single service failure. The system conducts all-round real-time monitoring and in-depth analysis, and promptly pushes the discovered problems and changes in work order status to relevant personnel to ensure the timely transmission of information. Based on the exponential backoff algorithm for failure retry, the reliability of notifications is guaranteed, and notification loss caused by network fluctuations is avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a framework diagram of the work order management system of the present invention;

[0050] Figure 2 It is a framework diagram of the internal structure of the multi-functional service module of the present invention;

[0051] Figure 3 It is a flowchart of the working steps of the elastic module of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0053] Referring to Figures 1 to 3 As shown, a customer service work order management system based on network employer services includes: a multi-functional service module, a distributed management module, an elastic module, an operation and maintenance monitoring module, and a real-time notification module;

[0054] The multi-functional service module includes a user service unit, a work order service unit, a communication service unit, a knowledge base service unit, and a data analysis unit, and is used to provide services for network employers;

[0055] The distributed management module is used to verify user validity, determine user legitimacy, and create work orders based on legitimate users;

[0056] When the customer service work order service response times out, the elastic module intervenes, monitors the system operation status, triggers the automatic fusing mechanism when detecting service response timeout. In the work order system, the automatic fusing cuts off the call to the timeout service. The elastic module pays attention to CPU and memory metrics, and expands microservice instances according to the dynamically changing metrics to ensure the stable operation of the system under high load;

[0057] The operation and maintenance monitoring module is used to collect the log data of microservices, store and analyze it, and locate problems; perform metric monitoring based on Prometheus, collect various performance metrics of the system, including CPU usage, memory usage, and network traffic, and perform real-time monitoring and analysis on the metrics to discover potential problems in the system and handle them;

[0058] The real-time notification module pushes the problems found during the operation and maintenance monitoring process and the changes in the work order status, performs failure retries based on the exponential backoff algorithm, and after a certain number of retries, sends the message to the dead letter queue for manual processing to ensure reliable notifications.

[0059] The distributed management module of the present invention determines the legitimacy of users by verifying user validity, ensuring the security and standardization of system services from the source. Only legitimate users can create work orders, which effectively avoids interference to the system caused by illegal operations, reduces the generation of invalid work orders, improves the quality and efficiency of work order processing. In the network employer service scenario, it ensures that the identity of the employer is real and legal, enables customer service resources to be accurately allocated to users with real needs, enhances the service experience of employers, and at the same time reduces the costs incurred by enterprises due to handling false work orders;

[0060] The design of the elasticity module provides strong guarantee for the system in a complex and changeable load environment. When the customer service work order service response times out, the automatic circuit breaker mechanism can quickly cut off the faulty service, prevent the spread of faults, and avoid the paralysis of the entire system caused by a single service failure. At the same time, the function of dynamically expanding microservice instances based on CPU / memory metrics enables the system to intelligently adjust resource configuration according to the actual load situation. During the business peak period, the system can automatically increase the number of microservice instances to ensure sufficient resources to handle a large number of work orders and guarantee the system response speed and stability; while during the off-peak period, it automatically reduces the number of instances to reduce resource consumption, achieving efficient utilization of resources and greatly enhancing the availability and reliability of the system;

[0061] The operation and maintenance monitoring module uses Fluentd and Elastic Stack to collect the log data of microservices, stores and analyzes it, and can quickly and accurately locate problems occurring in the system. Through Prometheus for metric monitoring, it collects key performance metrics such as CPU usage, memory usage, and network traffic in real time, and performs all-round real-time monitoring and in-depth analysis on the system. This not only helps to timely discover potential problems in the system, but also can predict possible problems through the analysis of historical data, and take measures in advance for optimization and prevention. Operation and maintenance personnel can deeply understand the running state of the system based on these monitoring data, quickly troubleshoot the cause of faults, and perform targeted processing, effectively reducing the system failure time and improving the overall performance and stability of the system;

[0062] The real-time notification module plays a crucial role in the operation and maintenance monitoring and work order management processes. It can promptly push the discovered problems and changes in work order status to relevant personnel, ensuring the timely transmission of information. It performs failure retries based on the exponential backoff algorithm to guarantee the reliability of notifications and avoid notification loss due to network fluctuations. When the number of retries reaches a certain limit, the message is sent to the dead letter queue for manual processing, further ensuring that important information is not overlooked. In customer service work order management, this enables employers, customer service staff, and managers to promptly understand the progress of work order processing, improve communication efficiency, facilitate timely adjustment of service strategies, and enhance customer satisfaction.

[0063] In summary, through the collaborative work of each module, the customer service work order management system of the present invention realizes the efficient management of the entire process from user verification, work order processing, system operation and maintenance to information notification. This greatly improves the quality and efficiency of network employer services, enhances the competitiveness of enterprises in the market. The stability and reliability of the system enable employers to obtain a better quality and more efficient service experience, help enterprises establish a good brand image, attract more customers, and promote the sustainable growth of business.

[0064] The user service unit is responsible for user identity management, including the identity management of different roles such as customers, customer service, and administrators; performs permission and role assignment based on the role-based access control model, provides a user information query interface, and supports two communication methods: gRPC / REST.

[0065] The work order service unit is used to realize the full life cycle management of work orders, including operations such as creation, assignment, escalation, resolution, and closure; automatically escalates work orders that have not been processed for a long time through SLA monitoring, provides label and classification management functions, and classifies and retrieves work orders.

[0066] The user service unit and the work order service unit are important components of the system. The user service unit is mainly responsible for user identity management, permission and role assignment, and providing a query interface. In terms of identity management, it establishes user, role, and association tables through a database, uses encryption and multi-factor authentication to ensure security, abstracts permissions and stores them based on the RBAC model, performs permission verification and dynamic adjustment accordingly, and at the same time, designs an interface that supports multiple conditional queries and is compatible with gRPC and REST communication methods;

[0067] The work order service unit realizes the full life cycle management and label classification retrieval functions of work orders. When creating a work order, it automatically numbers and assigns initial attributes, intelligently assigns orders through an assignment algorithm, escalates overdue work orders according to SLA monitoring, and audits and closes them after processing. In terms of label classification management, it provides an interface for administrators to operate. Users can add labels and select classifications for work orders, and use full-text retrieval technology to achieve efficient combined retrieval, facilitating users to classify, manage, and query work orders, and improving work order processing efficiency.

[0068] The communication service unit supports multi-channel access, including email, SMS, WebSocket and third-party IM, and communicates with customer service in various ways; it adopts intelligent routing mechanism to distribute sessions based on customer service load and improve service efficiency;

[0069] The knowledge base service unit uses Elasticsearch to implement intelligent search and combines IK word segmentation technology to perform searches; it uses NLP technology to process work order content;

[0070] The data analysis unit provides a real-time dashboard that displays customer service response time and resolution rate. It uses predictive analysis to forecast trends in work order volume and allocate resources. It also automatically generates reports in PDF format and sends them via email.

[0071] The communication service unit, knowledge base service unit and data analysis unit of this application each play a key role in the system and work together to improve the system performance. The communication service unit supports multi-channel access. Email access uses the mail server protocol to send and receive emails and convert formats; SMS access relies on the SMS gateway interface to interact with the service provider; WebSocket achieves real-time communication by building services; third-party IM achieves message synchronization based on the interfaces of each platform. At the same time, the unit adopts an intelligent routing mechanism to monitor the customer service load and use a dynamic weighted polling algorithm to give priority to assigning sessions to customers with low loads to improve service efficiency; the knowledge base service unit uses Elasticsearch ch and IK word segmentation technologies build indexes to achieve intelligent search and return results by relevance. With the help of NLP technology, text classification algorithms are used to classify work orders, and matching solutions are found through semantic understanding to assist customer service work. The data analysis unit collects customer service response time and resolution rate data from each microservice, and uses visualization tools to display them, making it easy to intuitively view changes in work efficiency. Time series analysis algorithms are used to predict work order volume trends, and deployment strategies are formulated based on the results and system resources. The document generation library can also be used to convert the analysis results into PDF reports, which are automatically sent through the email sending library to provide data support for decision-making, help the system optimize resource allocation, and improve overall service quality.

[0072] The distributed management module adopts Saga mode management. The management steps are as follows: the work order service initiates a transaction and generates a pre-work order; the user service is synchronously called to verify the user status; if the verification passes, the work order is submitted; if the verification fails, the pre-work order is deleted as compensation. Transaction events are recorded based on Kafka, and event tracing mechanism is used to ensure eventual consistency.

[0073] In a distributed system, the interactions between microservices are complex. The Saga pattern splits complex distributed transactions into multiple local transactions. When the work order service initiates a transaction to generate a pre-work order, even if there are problems in the subsequent user service verification step, the pre-work order can be deleted through compensation operations, preventing the system from experiencing overall data chaos due to local failures, ensuring that transactions can be properly handled in various abnormal situations, and maintaining the stable operation of the system. By using Kafka to record transaction events, a complete event log is provided for the entire transaction processing process. No matter which step the transaction has problems in, it can be traced back and analyzed based on these event records. Based on the event sourcing mechanism, the system can reconstruct the transaction state according to these records, ensuring that the data between microservices is ultimately consistent, avoiding data inconsistency caused by network fluctuations and service failures, and maintaining the accuracy and integrity of the data.

[0074] The management steps within the distributed management module are specifically represented as follows: predefined symbols, where T represents the work order transaction, P represents the pre-work order, V represents the user status verification result, V = true represents verification passed, V = false represents verification failed, S represents the work order status, S = 0 represents the pre-work order status, S = 1 represents the submitted work order status, E represents the set of transaction events recorded by Kafka, and A represents the transaction operation sequence under the event sourcing mechanism.

[0075] The work order service initiates a transaction: When the work order service initiates transaction T, it generates a pre-work order P and simultaneously sets the work order status S to the pre-work order status 0, which is represented by the following formula:

[0076]

[0077] where T start represents the start of transaction T, and P generate represents the generation of a pre-work order;

[0078] Synchronously call the user service to verify the user status: The work order service synchronously calls the user service to verify the user status and obtains the verification result V, which is represented as:

[0079]

[0080] where user_verify() is a function representing the user service performing the verification operation and returning the verification result;

[0081] Verification step: If the verification passes, the work order status S is changed from the pre-work order status 0 to the submitted work order status 1, and the formula is as follows:

[0082]

[0083] If the verification fails, perform a compensation operation to delete the pre-work order P and set the work order status S to the initial state. The formula is as follows:

[0084]

[0085] Where P delete represents deleting the pre-work order;

[0086] Record transaction events based on Kafka: During the entire transaction processing, each key step generates a transaction event and records it in the event set E of Kafka. The formula is expressed as:

[0087]

[0088] Where V true 、V false 、S submit represent the events of verification passed, verification failed, and work order submission respectively.

[0089] This application uses well-defined symbols and formulas to represent management steps, presenting complex distributed transaction management logic in a concise and intuitive way. Whether it is developers, testers, or operation and maintenance personnel, they can quickly understand the entire transaction processing flow and key links. Through clear formulas, they can clearly know how the work order status changes under different verification results and how transaction events are recorded, reducing human understanding and communication errors and improving the efficiency of team collaboration; In the system design stage, these formulas provide precise guidance for architecture design and module development. Developers can accurately write code according to the formulas to implement the interaction between various functional modules and transaction processing logic. When implementing the interaction between the work order service and the user service, developers can clearly determine when to call the user service for verification and how to process the work order status according to the verification result based on the formulas, thus improving the accuracy and efficiency of development, reducing development costs and error rates;

[0090] When the system fails, based on these symbols and formulas, technicians can quickly locate the problem. By analyzing the transaction event set E recorded in Kafka and combining the transaction operation sequence A, they can accurately trace the execution of each step in the transaction processing. If the work order status is abnormal, by checking the relevant events recorded in the event set E and the corresponding formulas, they can determine whether there is a problem in the verification step or a failure in the work order submission link, and then conduct targeted debugging and repair, shortening the troubleshooting time and improving the reliability and stability of the system.

[0091] Such as Figure 3As shown, the elastic module is fused based on the Hystrix fuse. The specific steps for dynamically expanding microservice instances based on CPU / memory metrics are as follows: First, build the monitoring system, containerization technology, and container orchestration platform components. The expansion method specifically includes metric collection and monitoring, and analyzing historical data to determine the threshold range; use the Kubernetes HPA automatic scaling policy configuration to perform scaling operations based on the threshold, and finally perform exception handling and rollback mechanisms.

[0092] The specific step process is as follows:

[0093] Metric collection and monitoring

[0094] Metric exposure: Integrate the metric exposure interface in the microservice code. Taking Prometheus as an example, the corresponding client library can be used to expose CPU and memory usage metrics;

[0095] Data collection: Prometheus regularly pulls data from the metric exposure interface of the microservice through the configured task and stores it in the time series database;

[0096] Visual monitoring: Use the Grafana tool to connect to the Prometheus database and create a dashboard to visually display the CPU and memory usage of the microservice;

[0097] Threshold setting

[0098] Analyze historical data: Based on the historical operation data of the system, analyze the peak and average values of CPU and memory usage of the microservice in different time periods and business scenarios;

[0099] Determine the threshold range: Combine business requirements and system performance requirements to set reasonable CPU and memory usage rate thresholds. When the CPU usage rate continuously exceeds 70% or the memory usage rate exceeds 80%, trigger expansion;

[0100] Automatic scaling policy configuration

[0101] Configure resource metrics: Create an HPA resource object in Kubernetes, specify the microservice Deployment or ReplicaSet to be monitored, as well as the monitored metrics and target thresholds;

[0102] Automatically adjust the number of replicas: HPA will automatically adjust the number of replicas of the microservice according to the monitored metrics. When the metrics exceed the threshold, increase the replicas; when they are below the threshold, reduce the replicas;

[0103] Custom scaling controller

[0104] Develop the controller: Develop a custom scaling controller with the help of a programming language and the Kubernetes API client library;

[0105] Implementation Logic: Implement the logic of metric monitoring, threshold judgment, and replica number adjustment in the controller. When it is detected that the CPU usage rate exceeds the threshold for 5 consecutive minutes, increase 2 replicas;

[0106] Expansion Process

[0107] Scaling-up Operation

[0108] Resource Evaluation: When the monitoring metrics reach the scaling-up threshold, the automatic scaling mechanism evaluates the cluster resource status to ensure that there are sufficient resources to create new instances;

[0109] Create New Instances: Kubernetes creates new microservice Pod instances according to the Deployment configuration;

[0110] Load Balancing: After the new instances are started, the load balancer distributes the traffic to the new instances;

[0111] Scaling-down Operation

[0112] Continuous Observation of Metrics: When the monitoring metrics are lower than the scaling-down threshold and last for a period of time, trigger the scaling-down operation;

[0113] Instance Selection and Deletion: Kubernetes selects appropriate Pod instances for deletion while ensuring that the normal operation of the system is not affected;

[0114] Exception Handling and Rollback Mechanism

[0115] Exception Detection: During the scaling process, the monitoring system monitors in real time whether exceptions occur, such as insufficient resources and failure to start new instances;

[0116] Rollback Operation: If an exception occurs, automatically trigger the rollback mechanism to restore to the state before the scaling.

[0117] The operation and maintenance monitoring module collects logs based on Fluentd + Elastic Stack. Fluentd is used to collect the log data of each microservice and send the log data to Elastic Stack for storage and analysis. Based on the search and analysis of ElasticStack, retrieve the logs according to the dimensions of time, service name, and log level to locate problems; Fluentd sends the collected log data generated by the microservice to Elastic Stack for storage and analysis. When it is necessary to find problems, conduct search and analysis based on Elastic Stack according to the dimensions of time, service name, and log level.

[0118] The operation and maintenance monitoring module of this application adopts the solution of Fluentd combined with Elastic Stack to process logs, bringing many advantages to system operation and maintenance. Fluentd is responsible for collecting logs from each microservice. With its lightweight and high performance, it can adapt to various log formats and sources, and can also preprocess logs to achieve efficient integration and send them to Elastic Stack. Elasticsearch in Elastic Stack stores a large amount of logs and has high availability. At the same time, it provides rich analysis functions, can analyze log data in real time, and gain insights into potential system trends. In terms of problem location, the method of retrieving logs according to time, service name, and log level is extremely efficient. When the system fails, operation and maintenance personnel can quickly lock down the problem with its help. Moreover, the in-depth analysis of log data can assist in system optimization, discover performance bottlenecks, and provide a basis for architecture adjustment and resource allocation. This system enhances the maintainability and reliability of the system, facilitates operation and maintenance personnel to control the system status, and ensures the stable operation of the system.

[0119] The specific steps for monitoring and analysis of the operation and maintenance monitoring module are as follows:

[0120] Determine the monitoring metrics and frequencies. Based on requirements, determine the frequencies of the monitoring metrics and perform monitoring based on Prometheus;

[0121] Collect data. For CPU usage, the monitoring tool obtains the ratio of the working time to the total time of the CPU within a unit time; the memory usage is obtained by calculating the ratio of the used memory to the total memory; the network traffic data is obtained from the network interface, including the number of bytes and packets received and sent.

[0122] Store the collected performance metric data in the database. Based on the visualization tool, display the collected performance metric data in the form of charts. Create a line chart to show the changing trends of CPU usage and memory usage over time, and use a bar chart to show the network traffic conditions in different time periods;

[0123] Based on the normal operating range of the system and business requirements, set thresholds for the performance metrics. When the monitored metrics exceed the set thresholds, transmit the information to the real-time notification module to alert the management personnel.

[0124] This application determines monitoring metrics and frequencies based on actual needs to ensure that system operation and maintenance focus on key points. For systems sensitive to traffic changes during peak business periods, the network traffic monitoring frequency can be increased; for compute-intensive services, the CPU usage metric is emphasized. This enables precise investment of operation and maintenance resources, conforms to the actual business situation, avoids ineffective monitoring, and improves monitoring efficiency; scientific methods are used to collect data on CPU usage, memory usage, and network traffic to ensure the accuracy and integrity of the data. These key metrics comprehensively reflect the system's operating status, providing a reliable basis for subsequent analysis. Through continuous and precise data collection, subtle changes in the system can be detected in a timely manner, laying a solid foundation for problem troubleshooting and performance optimization; visualization tools are used to present the data in the form of charts, making complex data intuitive. Line charts clearly show the fluctuations in CPU and memory usage over time, enabling operation and maintenance personnel to quickly grasp the resource usage trends; bar charts intuitively display network traffic at different times, helping to identify traffic peaks and valleys. This visualization method reduces the difficulty of data understanding and allows operation and maintenance personnel to quickly gain insights into the overall operating status of the system.

[0125] Among them, the method for determining the threshold is to collect data on various performance metrics of the system within a unit time of normal operation, calculate the mean and standard deviation of the collected data, and set the historical data as x1, x2, …… x n , and the mean calculation formula is:

[0126]

[0127] The calculation formula for the standard deviation s’ is:

[0128]

[0129] The threshold is set by adding a multiple of the standard deviation to the mean;

[0130] The comparison method is to directly compare the collected data with the obtained threshold. If the real-time collected performance metric is greater than the obtained threshold, it is determined that the metric exceeds the threshold, and information is sent to the real-time notification module.

[0131] The backoff algorithm is used in the real-time notification module for failure retry to handle the situation of failed notification sending, and the expression is:

[0132] T’ = T base ×(1 + r × random(0,1)) × 2 m

[0133] where T’ represents the current retry interval time, that is, the duration to wait for the next retry after this failure, and T baseis the initial retry interval, set to a small value, which is the starting benchmark of the retry interval. r is the randomization factor, which is a small positive number that introduces randomness into each retry interval to avoid network congestion caused by simultaneous retries of failed requests. Due to the existence of the squashing randomization factor, their retry times are different, reducing the impact of simultaneous retries on system resources. random(0,1) is a random function that returns a value between 0 and 1, inclusive of 0 but exclusive of 1. Each call to this function generates a different random number to increase the randomness of the retry interval. n’ represents the number of retries, starting from 0. As the number of retries increases, 2 n’ This item makes the retry interval increase exponentially. A maximum number of retries N will also be set during the retry process max and a maximum retry interval time T max to avoid infinite retries and overly long waiting times. That is, when n’≥N max the retries are stopped; when T’>T max the retry interval is set to T max .

[0134] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, various changes and improvements will occur to the present invention, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A customer service work order management system based on network employer services, characterized in that It includes: A multi-functional service module, a distributed management module, an elastic module, an operation and maintenance monitoring module, and a real-time notification module; The multi-functional service module includes a user service unit, a work order service unit, a communication service unit, a knowledge base service unit, and a data analysis unit, and is used to provide services for network employers; The distributed management module is used to verify the validity of users, determine the legality of users, and create work orders based on legitimate users; When the customer service work order service response times out, the elastic module automatically fuses and dynamically expands microservice instances based on CPU / memory metrics to ensure the stable operation of the system under high load; The operation and maintenance monitoring module is used to collect the log data of microservices, store and analyze it, and locate problems; Based on Prometheus for metric monitoring, collect various performance metrics of the system, including CPU usage, memory usage, network traffic, monitor and analyze the metrics in real time, discover potential problems in the system, and handle them; The real-time notification module pushes the problems found during the operation and maintenance monitoring process and the changes in the work order status, performs failure retries based on the exponential backoff algorithm, and after the retries reach a certain number of times, sends the message to the dead letter queue for manual processing to ensure the reliability of notifications.

2. The customer service work order management system based on network employer services according to claim 1, characterized in that, The user service unit is responsible for user identity management, including the identity management of different roles such as customers, customer service representatives, and administrators; performs permission and role assignment based on the role-based access control model, provides a user information query interface, and supports two communication methods: gRPC / REST; The work order service unit is used to implement the full life cycle management of work orders, including creation, assignment, escalation, resolution, and closure operations; automatically escalates work orders that have not been processed for a long time through SLA monitoring, provides label and classification management functions, and classifies and retrieves work orders.

3. The customer service work order management system based on network employer services according to claim 1, characterized in that, The communication service unit supports multi-channel access, including email, SMS, WebSocket, and third-party IM, and communicates with customer service representatives in various ways; adopts an intelligent routing mechanism to allocate sessions based on the load of customer service representatives to improve service efficiency; The knowledge base service unit realizes intelligent search through Elasticsearch, combines IK word segmentation technology for search, and uses NLP technology to process work order content; The data analysis unit provides a real-time dashboard to display the customer service response time and resolution rate; Through predictive analysis, predict the trend of work order volume and perform resource allocation; Automatically generate reports, generate the reports in PDF format and send them via Email.

4. A customer service work order management system based on network employer services according to claim 1, characterized in that, The distributed management module is managed using the Saga pattern. The management steps are as follows: The work order service initiates a transaction to generate a pre-work order; Synchronously call the user service to verify the user status; If the verification is passed, submit the work order. If the verification fails, compensate and delete the pre-work order, record transaction events based on Kafka, and ensure eventual consistency based on the event sourcing mechanism.

5. The customer service work order management system based on network employer services according to claim 4, characterized in that, The management steps within the distributed management module are specifically represented as follows: predefined symbols are used, where T represents the work order transaction, P represents the pre-work order, V represents the user status verification result, V = true indicates successful verification, V = false indicates failed verification, S represents the work order status, S = 0 represents the pre-work order status, S = 1 represents the submitted work order status, E represents the set of transaction events recorded by Kafka, and A represents the sequence of transaction operations under the event tracing mechanism; The work order service initiates a transaction: When the work order service initiates transaction T, a pre-work order P is generated, and simultaneously the work order status S is set to the pre-work order status 0, which is represented by the following formula: Among which T start indicates the start of transaction T, and P generate indicates the generation of a preliminary work order; Synchronously call the user service to verify the user status: The work order service synchronously calls the user service to verify the user status and obtains the verification result V, which is expressed as: where user_verify() is a function representing that the user service performs the verification operation and returns the verification result; Verification step: If the verification is successful, the work order status S is changed from the pre-work order status 0 to the submitted work order status 1, and the formula is as follows: If the verification fails, a compensation operation is performed to delete the pre-work order P, and the work order status S is set to the initial state. The formula is as follows: Among which P delete indicates deleting the preliminary work order; Based on Kafka to record transaction events: During the entire transaction processing process, a transaction event is generated at each key step and recorded in the Kafka event set E. The formula is expressed as: Among them, V true , V false , S submit respectively represent the events of verification passed, verification failed, and work order submitted.

6. The customer service work order management system based on network employer services according to claim 1, wherein The elastic module performs fusing based on the Hystrix fuse, and the specific steps for dynamically expanding microservice instances based on CPU / memory metrics are as follows: First, build a monitoring system, containerization technology, and container orchestration platform components. The expansion methods specifically include metric collection and monitoring, and analyze historical data to determine the threshold range; Use the Kubernetes HPA automatic scaling policy configuration to perform scaling operations based on the threshold, and finally perform exception handling and rollback mechanisms.

7. A customer service work order management system based on network employer services according to claim 1, characterized in that The operation and maintenance monitoring module collects logs based on Fluentd + Elastic Stack. Fluentd is used to collect the log data of each microservice and send the log data to Elastic Stack for storage and analysis. Based on the search and analysis of Elastic Stack, retrieve the logs according to the time, service name, and log level dimensions to locate problems; Fluentd sends the log data generated by the collected microservices to Elastic Stack for storage and analysis. When it is necessary to find problems, perform search and analysis based on ElasticStack according to the time, service name, and log level dimensions.

8. A customer service work order management system based on network employer services according to claim 1, characterized in that, The specific steps for the operation and maintenance monitoring module to monitor and analyze are as follows: Determine the monitoring metrics and frequencies. Based on requirements, determine the frequencies of the monitoring metrics and perform monitoring based on Prometheus; Collect data. For the CPU usage rate, the monitoring tool obtains the ratio of the working time to the total time of the CPU within a unit time; The memory usage rate is obtained by calculating the ratio of the used memory to the total memory; The network traffic data is obtained from the network interface, including the number of bytes and packets received and sent; Store the collected performance metric data in a database. Based on a visualization tool, display the collected performance metric data in the form of charts. Create a line chart to show the changing trends of CPU usage and memory usage over time, and use a bar chart to show the network traffic conditions in different time periods; Based on the normal operating range of the system and business requirements, set thresholds for performance metrics. When the monitored metrics exceed the set thresholds, transfer the information to the real-time notification module to alert the management staff.

9. The customer service work order management system based on network employer services according to claim 8, characterized in that Among them, the method for determining the threshold collects the data of various performance indicators of the system within a unit time of normal operation, calculates the mean and standard deviation of the collected data, and sets the historical data as x1, x2, …… x n , and the mean calculation formula is: The calculation formula for the standard deviation s’ is: The threshold is set by adding a multiple of the standard deviation to the mean; The comparison method is to directly compare the collected data with the obtained threshold. If the real-time collected performance metric is greater than the obtained threshold, it is determined that the metric exceeds the threshold and information is sent to the real-time notification module.

10. A customer service work order management system based on network employer services according to claim 1, characterized in that, The backoff algorithm is used in the real-time notification module for failure retry to handle the situation of failed notification sending. The expression is: T’ = T base × (1 + r × random(0, 1)) × 2 n’ Where T’ represents the current retry interval, that is, the duration to wait for the next retry after this failure, and T base is the initial retry interval, set to a small value, which is the starting benchmark of the retry interval. r is a randomization factor, a small positive number, which introduces randomness into each retry interval to avoid network congestion caused by simultaneous retries of failed requests. Due to the existence of the squashing randomization factor, their retry times are different, reducing the impact of simultaneous retries on system resources. random(0,1) is a random function that returns a value between 0 and 1, including 0 but not including 1. Each time this function is called, a different random number is generated to increase the randomness of the retry interval. n’ represents the number of retries, starting from 0. As the number of retries increases, 2 n’ this term makes the retry interval increase exponentially. A maximum number of retries N max and a maximum retry interval T max are also set during the retry process to avoid infinite retries and overly long waiting times. That is, when n’≥N max the retries are stopped; when T’>T max the retry interval is set to T max .

Images