Low-code platform application deployment method and system based on AI intelligent scheduling

Abstract

The invention discloses a low-code platform application deployment method and system based on AI intelligent scheduling, and relates to the technical field of software development and deployment. The method comprises the following steps of executing quality detection judgment, load scheduling configuration deployment and resource scheduling configuration deployment. The execution quality detection result is obtained by tracking the execution efficiency of the application code generated by the low-code platform, it is ensured that the quality of the generated application code meets the expectation, potential performance bottlenecks are found in time, and the running efficiency problem caused by improper configuration is avoided. On the basis, whether the requirement for application deployment parameter initialization setting exists or not is judged, and resource waste and excessive reservation are effectively avoided through deployment parameter initialization and load fluctuation monitoring. Furthermore, by monitoring the resource utilization rate in real time, it is ensured that resources in the application deployment process are effectively utilized, idle or excessive consumption of the resources is avoided, and the deployment efficiency, stability and resource utilization efficiency of the platform are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of software development and deployment technology, and in particular to a method and system for deploying low-code platform applications based on AI intelligent scheduling. Background Technology

[0002] With the acceleration of digital transformation, low-code platforms, with their visual, component-based, and configuration-driven features, have become a crucial support for enterprises to quickly build business applications. However, as low-code platforms handle more and more business in real-world production environments, support larger user scales, and ensure more stringent service levels, their deployment, operation, and resource management gradually become key bottlenecks in system stability, cost efficiency, and scalability. Against this backdrop, introducing AI (Artificial Intelligence) intelligent scheduling capabilities into the application deployment phase of low-code platforms is not only a trend upgrade but also a strategic requirement.

[0003] In the existing low-code platform application deployment process, the user first uploads a JSON template file through the platform interface. This file contains the front-end layout, back-end business logic, database table structure, and user-customized configurations. The platform performs a hash operation on this JSON file to generate a unique identifier and create a Docker image. After generating the Docker image, a low-code platform application container is created based on the configuration information of the server to be deployed. During the container startup phase, the user service client obtains the user center address and key of the target server. The Spring Boot service running inside the container reads the JSON template embedded in the image and initializes the data table structure, interface registration, business routing, etc.; for different servers or environments, differentiated configurations are also populated. Throughout the application's runtime, front-end requests are forwarded to the static resource directory in the image, while back-end data requests are handled by the Spring Boot service. Furthermore, after successful container deployment, if an application update is required, the version numbers of the old and new JSON template files are compared to determine if an update is necessary. If the new template file has a higher version number, a template compatibility check is performed. If the check passes, the platform updates or redeploys the container; if the check fails, the update is abandoned, and the current version is retained. Before updating, the platform will back up the current JSON template, server running status, database snapshot, etc. If a failure occurs during the update process, the backup data can be used to implement a rollback mechanism.

[0004] For example, Chinese patent application CN120406986A discloses a method and device for deploying low-code platform applications, including: obtaining a JSON template file of the low-code platform application, generating a hash identifier for the JSON template file, and generating a Docker image definition file; matching the hash identifier of the JSON template with the hash identifiers of historical Docker images in a preset image library; if the matching fails, generating a Docker image of the JSON template file based on the Docker image definition file; and generating a low-code platform application container based on the configuration information of the server to be deployed and the Docker image.

[0005] The above-mentioned technology has at least the following technical problems: In the deployment process of low-code platform applications, the current resource configuration (such as memory, network bandwidth, etc.) is relatively static, usually initially allocated based on estimates or server specifications, rather than based on real-time operational status monitoring or dynamic load adjustment. As a result, the application often has resources idle or underutilized most of the time, while they may be insufficient during sudden load surges or business peaks, leading to low resource utilization efficiency.

[0006] Another point to consider is that different applications have different requirements for computing and storage resources. However, if the resource scheduling system cannot identify and adjust the ratio of computing and storage resources, it may lead to an overabundance of computing resources and an underabundance of storage resources, or vice versa, further resulting in low resource utilization during the deployment of low-code platform applications. Summary of the Invention

[0007] To address the low resource utilization problem in the deployment of low-code platform applications in existing technologies, this invention provides a method and system for deploying low-code platform applications based on AI intelligent scheduling. The technical solution is as follows: On the one hand, a method for deploying low-code platform applications based on AI intelligent scheduling is provided. This method includes: tracking the execution efficiency of application code generated by the low-code platform to obtain execution quality detection results; determining whether there is a need for initializing application deployment parameters for configuring the database connection environment based on the obtained execution quality detection results; if there is a need, determining the load request fluctuations during deployment after initializing the application deployment parameters, and determining whether to trigger dynamic load scheduling configuration based on the fluctuation determination results, and obtaining resource utilization detection results; if there is no need, sending an instruction to determine the resource utilization during deployment after obtaining the fluctuation determination results, and determining whether to trigger resource scheduling configuration to improve resource utilization by adjusting deployment strategy parameters based on the obtained resource utilization detection results.

[0008] On the other hand, an AI-based intelligent scheduling low-code platform application deployment system is provided, including: an execution quality detection and judgment module, a load scheduling configuration and deployment module, and a resource scheduling configuration and deployment module. The execution quality detection and judgment module is used to track the execution efficiency of the application code generated by the low-code platform to obtain execution quality detection results. Based on the obtained execution quality detection results, it determines whether there is a need for initializing application deployment parameters for configuring the database connection environment. If there is a need, the load scheduling configuration and deployment module is used to determine the load request fluctuations during deployment after initializing application deployment parameters, and then determine whether to trigger dynamic load scheduling configuration based on the fluctuation determination results, and obtain resource utilization detection results. If there is no need, the resource scheduling configuration and deployment module is used to send an instruction to determine the resource utilization during deployment after obtaining the fluctuation determination results, and determine whether to trigger resource scheduling configuration to improve resource utilization by adjusting deployment strategy parameters based on the obtained resource utilization detection results.

[0009] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. Low-code platforms offer significant advantages in application development, but challenges remain in actual application deployment, particularly in handling application performance, resource scheduling, and system stability. Therefore, introducing AI-based intelligent scheduling for application deployment is crucial. This invention significantly improves the efficiency and quality of application deployment by introducing automated and intelligent scheduling. Specifically, firstly, by tracking and quality-checking the execution efficiency of application code generated by the low-code platform, it can promptly identify critical configurations such as the need to initialize the database connection environment. This process ensures that all necessary configurations are in place before application deployment, avoiding performance bottlenecks or system failures caused by improper environment configuration. Furthermore, it enhances the application's adaptability before deployment, enabling efficient operation in different deployment environments. Secondly, this invention analyzes load request fluctuations during deployment to determine whether dynamic load scheduling configuration needs to be triggered. Real-time monitoring of load fluctuations allows the system to flexibly respond to traffic peaks or sudden load changes, preventing crashes or performance degradation due to excessive load. This results in a smoother operation of the low-code platform during application deployment and an improved user experience. Furthermore, this invention also relates to a resource utilization detection mechanism. By monitoring resources in real time, it determines whether deployment strategy parameters need to be adjusted to optimize resource configuration. Through intelligent scheduling, resource waste or overload can be effectively avoided, improving resource utilization efficiency during application deployment. Compared with existing technologies, the deployment of traditional low-code platform applications often relies on manual settings and experience-based judgment, which is prone to human error and lacks flexible load scheduling and resource optimization mechanisms. In contrast, this method uses AI intelligent scheduling to automatically perform real-time monitoring and decision-making, greatly improving the automation and intelligence of the deployment process, reducing manual intervention, and lowering the probability of operational errors. In summary, this invention not only ensures the rapid deployment of low-code platform applications but also optimizes the performance, resource utilization, and deployment stability after deployment. Through intelligent scheduling and resource optimization, while improving deployment quality, it also helps reduce operation and maintenance costs and enhances the adaptability and sustainability of low-code platform applications under different load environments, demonstrating significant technical advantages.

[0010] 2. Before deploying application code generated on a low-code platform, real-time monitoring of the error rate and response latency helps to promptly identify potential problems in the application code, ensuring that the code meets qualification standards before deployment and preventing errors from propagating to the deployment environment. Compared to traditional static inspection mechanisms, this method provides a more flexible and real-time error detection and configuration adjustment method by dynamically tracking the execution error rate and combining it with deviation control thresholds, significantly improving the controllability and flexibility of execution quality. Traditional methods typically perform uniform configuration before deployment, lacking the ability to dynamically adjust, while this method can adaptively adjust based on real-time feedback, improving the intelligence and automation level of deployment. On this basis, through comprehensive analysis of error rate and response latency, deployment strategies are dynamically adjusted to ensure that the application can run stably under different environments and loads, improving the quality assurance of application deployment. Furthermore, by combining error rate deviation with response latency deviation to adjust the tracking time interval and connection pool parameters (such as idle connection timeout time and connection recycling time), deployment parameters can be made more closely aligned with actual operating conditions, thereby improving the efficiency of connection pool resource utilization and reducing problems caused by connection timeout or recycling latency. Furthermore, fine-grained judgment of request rate fluctuations during deployment, combined with concurrency correction, failure rate, and response time standard deviation to determine whether fluctuations are abnormal, helps to respond more agilely to load changes and avoid peak or abnormal loads impacting service performance. When load fluctuations are abnormal, dynamic load scheduling is triggered, adjusting scaling ratios, rates, and cooldown times to enable resources to adapt to load changes more quickly and reasonably, exhibiting greater elasticity. Finally, resource utilization detection results determine whether to trigger resource scheduling configuration, ensuring not only performance stability but also improving resource utilization, reducing idle waste, and optimizing costs. Overall, this invention achieves closed-loop intelligent control before deployment, during deployment, and during runtime, thereby improving deployment efficiency, system stability, resource utilization, and the level of operational automation. Compared with existing technologies, this invention avoids the problem of discovering improper environment configurations only after deployment, breaking through the limitations of static monitoring and fixed strategies. The overall process is highly automated and intelligent, requiring less manual intervention, responding faster, and exhibiting stronger adaptability, thus significantly improving the ability of low-code platform applications to go from "rapid deployment" to "efficient operation."

[0011] 3. During application deployment, by quantifying the utilization of computing and storage resources (such as memory usage efficiency, CPU load, disk read / write speed, and storage expansion latency), the resource usage during application deployment can be accurately assessed. This multi-dimensional monitoring can comprehensively reflect the location of resource bottlenecks, thus providing a reliable basis for subsequent resource scheduling. Traditional deployment methods often use static resource allocation strategies, which can easily lead to resource waste or insufficiency, especially in environments with large load fluctuations. This invention, through real-time monitoring and dynamic adjustment of resource utilization, ensures accurate resource scheduling and avoids resource waste or performance bottlenecks caused by static configuration. Furthermore, by dynamically adjusting the resource pool and traffic allocation ratio, resource waste or insufficiency can be effectively avoided. When container response time is too long, shrinking the memory resource pool avoids excessive consumption of system memory, improves application performance and response speed, and ensures that the system can still run efficiently under high load. On this basis, continuous monitoring of the utilization of computing and storage resources and adaptive adjustments based on real-time feedback can ensure that optimal performance is maintained under different loads and operating conditions, enhancing the elasticity and stability of the application deployment process. Existing technologies may face the problem of failing to adjust resource configuration in a timely manner when load changes, leading to longer container response times or service instability. This invention, by dynamically adjusting the shrinking ratio of the resource pool and traffic allocation, can automatically optimize the deployment process, ensuring the platform maintains high efficiency under load fluctuations and significantly improving the stability and fault tolerance of application deployment. Simultaneously, by dynamically adjusting the resource pool memory, it further optimizes memory resource usage, effectively avoiding excessive resource consumption or waste, and ensuring efficient resource configuration. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A flowchart illustrating the low-code platform application deployment method based on AI intelligent scheduling provided in this embodiment of the invention; Figure 2 The following is a logic diagram for obtaining the fluctuation determination result of the low-code platform application deployment method based on AI intelligent scheduling provided in the embodiments of the present invention; Figure 3 A logic diagram for obtaining resource utilization detection results of the low-code platform application deployment method based on AI intelligent scheduling provided in this embodiment of the invention; Figure 4This is a schematic diagram of the structure of a low-code platform application deployment system based on AI intelligent scheduling provided in an embodiment of the present invention. Detailed Implementation

[0014] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0015] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0016] In the context of current digital transformation, low-code platforms are gradually becoming an important tool for enterprises to quickly build applications. However, as business scale and complexity increase, low-code platforms often face problems such as uneven resource allocation and unpredictable load fluctuations during the application deployment phase. Traditional deployment methods are usually based on static configuration, lacking dynamic monitoring and adjustment of resource utilization, leading to performance bottlenecks, resource waste, and even system stability issues.

[0017] The introduction of AI-powered intelligent scheduling technology enables automatic adjustment of system resource configuration based on real-time data, thereby improving resource utilization efficiency and system responsiveness. Specifically, by dynamically analyzing application load, resource requirements, and system performance, AI scheduling can optimize server resource allocation, reduce operational costs, and enhance system availability and scalability. For low-code platforms, intelligent scheduling is not merely a matter of allocating computing resources, but also involves load balancing, automatic scaling, and fault prevention in different environments.

[0018] like Figure 1 The diagram shows a flowchart of a low-code platform application deployment method based on AI intelligent scheduling provided in an embodiment of the present invention. The method includes the following steps: performing quality detection and judgment, load scheduling configuration deployment, and resource scheduling configuration deployment.

[0019] refer to Figure 1 The first step in the AI-based intelligent scheduling low-code platform application deployment method is execution quality detection and judgment. Specifically: S1, the execution efficiency of the application code generated by the low-code platform is tracked to obtain execution quality detection results. Based on the obtained execution quality detection results, it is determined whether there is a need for initializing application deployment parameters for configuring the database connection environment. By tracking and evaluating the execution efficiency of the generated code in real time, potential performance bottlenecks or errors can be identified in a timely manner. For example, if the generated code has low execution efficiency or excessive resource consumption, it can be corrected promptly through a feedback mechanism. This helps ensure that the generated code has higher quality when deploying the application, avoiding system performance instability or failures due to code problems. The specific steps are as follows: S11. Within the code execution tracing time interval, obtain the code execution error rate during the application code execution process using log analysis tools such as Prometheus, and compare the obtained code execution error rate with the stored maximum execution error rate.

[0020] It should be noted that the maximum execution error rate is stored in the platform application deployment database. This database is created first when designing the low-code platform application deployment method based on AI intelligent scheduling, and is specifically used to store core configuration information. This database stores various limits and mapping sets necessary for the operation of this method, such as the maximum execution error rate. The initial settings of these limits are not arbitrarily specified. Technical personnel can manually set, adjust, or fine-tune them at any time according to the specific performance of the method in actual testing, thereby ensuring that the method can be continuously optimized to achieve the best working state.

[0021] S12, if the code execution error rate is greater than the maximum execution error rate limit, the execution quality detection result will be recorded as unqualified, and a code execution exception prompt will be sent.

[0022] S13, otherwise, record the execution quality test result as qualified, and obtain the execution error rate deviation parameter, which reflects the degree of deviation between the code execution error rate and the maximum execution error rate limit, i.e., the difference between the maximum execution error rate limit and the code execution error rate.

[0023] S14. Compare the obtained execution error rate deviation parameter with the stored deviation control threshold. If the execution error rate deviation parameter is less than the deviation control threshold, it is determined that there is a need for application deployment parameter initialization settings. Then, the application deployment parameter initialization settings for configuring the database connection environment are performed and the fluctuation judgment result is obtained. Otherwise, it is determined that there is no need for application deployment parameter initialization settings.

[0024] In the application deployment process on low-code platforms, besides focusing on the code execution error rate, code execution response latency is also a key performance indicator. Code execution response latency directly impacts the user experience and system responsiveness. To ensure efficient application operation and system stability, it is essential to dynamically optimize system resource allocation and application deployment parameters by tracking and adjusting execution response latency. As a further step, tracking the execution efficiency of application code generated by the low-code platform also includes: First, the code execution response latency within the code execution tracing time interval is obtained through log analysis tools. When the deviation of the execution error rate from the parameter is not less than the deviation control threshold, the obtained code execution response latency is compared with the stored code execution latency setting value. If the code execution response latency is not greater than the code execution latency setting value, the application deployment parameter initialization requirement judgment is not updated. Otherwise, it is determined that there is a requirement for application deployment parameter initialization settings and the execution response deviation parameter is obtained.

[0025] Secondly, when the execution error rate deviation parameter is less than the deviation control threshold, the application deployment parameter initialization requirement judgment is not updated and the execution response deviation parameter is obtained. The execution response deviation parameter is used to reflect the degree of deviation between the code execution response delay and the code execution delay setting value, that is, the absolute value of the difference between the code execution response delay and the code execution delay setting value.

[0026] Next, if there is no need to initialize application deployment parameters, the code execution error rate and code execution response latency are monitored for the next code execution tracing time interval. Otherwise, the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter are input into the tracing time interval mapping table. Both the execution error rate deviation parameter and the execution response deviation parameter have been de-normalized, and the corresponding output is the tracing time interval adjustment amount used to set the next code execution tracing time interval. The tracing time interval mapping table is pre-trained using the historical results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter, as well as the tracing time interval adjustment amount set by professionals based on experience rules. It is used to describe the mapping relationship between the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter and the tracing time interval adjustment amount. The inverse proportional processing is the addition and reciprocal operation.

[0027] In the next code execution tracing time interval, the adjustment value of the tracing time interval is set based on the obtained tracing time interval adjustment value. That is, the step size corresponding to the tracing time interval adjustment value is set as the length of the next code execution tracing time interval, and the code execution error rate and code execution response latency of the time interval are monitored.

[0028] During the deployment and operation of application code generated on a low-code platform, a two-dimensional monitoring mechanism combining multi-dimensional analysis of code execution error rate and response latency, with deviation parameters serving as the trigger for decision-making, can significantly improve application performance, deployment flexibility, and operational efficiency. Furthermore, the introduction of dynamically adjusted monitoring cycles optimizes the system's monitoring strategy. When the system is running stably, the monitoring cycle can be appropriately extended to avoid performance overhead from frequent monitoring; conversely, when the system exhibits abnormal deviations, the monitoring cycle automatically shortens to ensure higher-frequency error detection and performance optimization. This adaptive monitoring mechanism enhances the system's resilience and adaptability, providing a more flexible response to load fluctuations, sudden traffic surges, and other business changes.

[0029] The second and third steps of the low-code platform application deployment method based on AI intelligent scheduling are load scheduling configuration deployment and resource scheduling configuration deployment. Specifically: S2, if there is a demand, after initializing the application deployment parameters, the load request fluctuations during deployment are judged, and the dynamic load scheduling configuration is triggered based on the fluctuation judgment result, and the resource utilization detection result is obtained; S3, if there is no demand, after obtaining the fluctuation judgment result, an instruction for judging the resource utilization during deployment is sent, and the resource utilization detection result is used to determine whether to trigger the resource scheduling configuration for improving resource utilization by adjusting the deployment strategy parameters.

[0030] Furthermore, the application deployment parameter initialization settings include idle connection timeout settings and connection recycling time settings. Specifically, setting the idle connection timeout effectively optimizes connection resource management in the connection pool and improves overall system performance. By dynamically adjusting the idle connection timeout based on execution error rate and response latency, the system can promptly release unused connections, thereby preventing a large number of idle connections from consuming valuable resources in the connection pool. This setting helps reduce unnecessary resource consumption when the system load is light, while ensuring that there are always enough valid connections available for rapid use when the load is heavy, ensuring the stability of system response speed and processing capacity. Through this flexible idle connection management, system resource waste caused by too many idle connections can be avoided, and the system's resource utilization and overall performance can be effectively improved.

[0031] The idle connection timeout setting is specifically as follows: the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter are input into the corresponding output connection idle timeout adjustment time of the connection idle timeout mapping set. Based on the output connection idle timeout adjustment time, the idle connection timeout in the connection pool is set, that is, the idle connection timeout is set as the connection idle timeout adjustment time. The connection idle timeout mapping set is pre-trained using historical results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter, as well as the connection idle timeout adjustment time set by professionals based on empirical rules. It is used to describe the mapping relationship between the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter and the connection idle timeout adjustment time.

[0032] Furthermore, setting the connection recycling time ensures that invalid or expired connections in the connection pool are promptly recycled and reallocated. Dynamically adjusting the connection recycling time based on real-time feedback from execution error rates and response latency effectively prevents unused connections from consuming resources, ensuring the system always has sufficient and healthy connection resources. Under high concurrency or resource constraints, appropriately shortening the connection recycling time can quickly release useless connections, avoiding system latency or performance bottlenecks caused by prolonged connection occupation; under low load, extending the connection recycling time reduces frequent connection destruction and recreation operations, lowering system overhead. This setting enhances the system's resilience and adaptability, enabling applications to dynamically adjust connection pool strategies according to actual business needs, ensuring maximum utilization of connection resources and optimal performance output.

[0033] The connection recycling time setting is as follows: Input the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter to the connection pool recycling time mapping set. The corresponding output is the connection pool recycling adjustment time used to set the initial connection recycling time, i.e., setting the connection recycling time as the connection pool recycling adjustment time. The connection pool recycling time mapping set is pre-trained using historical results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter, as well as connection pool recycling adjustment times set by professionals based on empirical rules. It describes the mapping relationship between the results of the inverse proportional processing of the execution error rate deviation parameter and the arithmetic mean processing of the execution response deviation parameter and the connection pool recycling adjustment time.

[0034] like Figure 2The diagram shown illustrates the logic for obtaining the fluctuation judgment result in the low-code platform application deployment method based on AI intelligent scheduling provided in this embodiment of the invention. The corresponding logic is as follows: The obtained request rate fluctuation judgment value is compared with the maximum request rate fluctuation limit. If the request rate fluctuation judgment value is greater than the maximum request rate fluctuation limit, the fluctuation judgment result indicates abnormal load request fluctuation. Otherwise, the load request fluctuation judgment value is obtained based on the coupling result of the request rate fluctuation judgment value and the concurrency effect correction factor. It is then determined whether the obtained load request fluctuation judgment value is greater than the request rate fluctuation baseline limit. If yes, the request failure rate is obtained. If and only if the request failure rate is less than the request failure set value, the fluctuation judgment result indicates normal load request fluctuation; otherwise, the fluctuation judgment result indicates abnormal load request fluctuation. If no, the response time standard deviation is obtained. If and only if the response time standard deviation is less than the response fluctuation set value, the fluctuation judgment result indicates normal load request fluctuation; otherwise, the fluctuation judgment result indicates abnormal load request fluctuation. As a further solution, refer to... Figure 2 To determine load request fluctuations during deployment and obtain the fluctuation determination results, the specific steps are as follows: M1 acquires a request rate fluctuation determination value during the deployment period for determining request rate fluctuations. The request rate fluctuation determination value is represented by the ratio of the standard deviation of the request rate to the average request rate during the deployment period. The standard deviation of the request rate and the average request rate are obtained through log analysis tools.

[0035] M2 compares the obtained request rate fluctuation judgment value with the stored request rate fluctuation maximum limit; if the request rate fluctuation judgment value is greater than the request rate fluctuation maximum limit, it indicates that the fluctuation judgment result is abnormal load request fluctuation.

[0036] In M3, if the request rate fluctuation judgment value is not greater than the maximum request rate fluctuation limit, a concurrency effect correction factor is introduced to correct the amplification effect of different concurrency levels on load request fluctuations. The load request fluctuation judgment value is obtained based on the coupling result of the request rate fluctuation judgment value and the concurrency effect correction factor. The concurrency effect correction factor is set by professionals according to industry standards, and the coupling result is the product of the request rate fluctuation judgment value and the concurrency effect correction factor. It is important to understand that different concurrency levels have different impacts on system load. By introducing the concurrency effect correction factor, the system can dynamically correct the impact of request rate fluctuations, making load fluctuation judgment more accurate. For example, when the number of concurrent users is high, the system will automatically adjust the fluctuation judgment according to changes in concurrency, thereby avoiding misjudgments caused by the concurrency effect amplifying request fluctuations.

[0037] M4 determines whether the obtained load request fluctuation judgment value is greater than the request rate fluctuation baseline limit: if so, the request failure rate used for request failure judgment is obtained through log analysis tools. If and only if the request failure rate is less than the request failure setting value, it indicates that the fluctuation judgment result is that the load request fluctuation is normal; otherwise, it indicates that the fluctuation judgment result is that the load request fluctuation is abnormal.

[0038] M5, if not, then obtain the standard deviation of response time for request response determination through log analysis tools. If the standard deviation of response time is less than the response fluctuation setting value, it indicates that the fluctuation determination result is that the load request fluctuation is normal; otherwise, it indicates that the fluctuation determination result is that the load request fluctuation is abnormal.

[0039] Based on this, further, the dynamic load scheduling configuration is triggered according to the fluctuation determination result. The specific process is as follows: N1, if the fluctuation determination result is that the load request fluctuation is normal, then dynamic load scheduling configuration will not be triggered, and resource utilization during deployment will be determined to obtain resource utilization detection results; if Figure 3 The diagram shown illustrates the logic for obtaining resource utilization detection results in the low-code platform application deployment method based on AI intelligent scheduling provided in this embodiment of the invention. The corresponding logic is as follows: obtain the resource utilization quantification parameters during the deployment period to obtain the resource utilization reflection value used to quantify the resource utilization rate during the deployment period. If the resource utilization reflection value is less than the minimum resource utilization limit, the resource utilization detection result is recorded as unqualified. The output published traffic percentage and resource pool shrinkage ratio are obtained to trigger resource scheduling configuration. Otherwise, the resource utilization detection result is recorded as qualified. Resource utilization indicators are continuously monitored, calculated, stored, and fed back.

[0040] refer to Figure 3 The resource utilization monitoring results are obtained using the following methods: N11, obtain resource utilization quantification parameters that reflect resource utilization during the deployment period. The resource utilization quantification parameters include computing resource utilization indicators that characterize computing resource utilization performance and storage resource utilization indicators that characterize storage resource utilization performance.

[0041] Specifically, the computing resource utilization index is represented by the arithmetic mean of the memory usage efficiency ratio and the CPU load deviation ratio. The memory usage efficiency ratio is the ratio of memory usage efficiency to the minimum memory usage efficiency limit. The CPU load deviation ratio is the inverse of the absolute value of the difference between the CPU load reference setting and the CPU load. The storage resource utilization index is represented by the arithmetic mean of the disk read / write rate ratio and the storage expansion latency deviation. The disk read / write rate ratio is the ratio of disk read / write rate to the minimum disk read / write rate limit. The storage expansion latency deviation ratio is the inverse of the inverse of the absolute value of the difference between the storage expansion latency reference setting and the storage expansion latency. The CPU load reference setting, minimum memory usage efficiency limit, minimum disk read / write rate limit, and storage expansion latency reference setting are obtained from the platform application deployment database. Memory usage efficiency, disk read / write rate, storage expansion latency, and CPU load are obtained from application performance management tools (such as Datadog) and have all been de-unitized.

[0042] N12, based on the obtained weights, performs weighted coupling processing on the resource utilization quantification parameters to obtain a resource utilization reflection value used to quantify the resource utilization rate during deployment; the weighted coupling processing is to multiply the calculated resource utilization index and the storage resource utilization index with their corresponding weights and then add them together, where the calculated resource utilization index corresponds to the calculated resource utilization weight, and the storage resource utilization index corresponds to the storage resource utilization weight.

[0043] N13 compares the obtained resource utilization response value with the minimum resource utilization limit of the storage. If the resource utilization response value is less than the minimum resource utilization limit, the resource utilization detection result is recorded as unqualified and resource scheduling configuration is triggered. Otherwise, the resource utilization detection result is recorded as qualified. The system continuously monitors and calculates resource utilization indicators and storage resource utilization indicators and provides feedback.

[0044] The specific process for resource scheduling and configuration is as follows: N131 inputs resource utilization reflection values ​​into the deployment traffic allocation mapping table to output the published traffic percentage. Based on the obtained published traffic percentage, it sets the initial traffic allocation ratio during application deployment, i.e., sets the allocated traffic ratio to the published traffic percentage. It also uses log management tools to obtain container response times during application deployment. The deployment traffic allocation mapping table is pre-trained using historical resource utilization reflection values ​​and published traffic percentages set by professionals based on experience rules. It describes the mapping relationship between resource utilization reflection values ​​and published traffic percentages. Adjusting the published traffic percentage based on the resource utilization reflection values ​​allows the system to allocate more resources and traffic to critical application instances under high load conditions, avoiding excessive or insufficient container resource consumption. Reasonable traffic allocation can improve the overall system throughput and ensure that critical services are not overwhelmed during peak traffic periods.

[0045] Adjusting traffic allocation based on resource utilization data allows for real-time balancing of system resources, preventing overload of certain containers or services due to excessive traffic. If a container's response time is too long or resource utilization is too high, traffic will automatically decrease to prevent overload and ensure stable application operation.

[0046] N132: If the container response time is not greater than the set maximum container response time limit, a resource scheduling configuration completion prompt is sent. Otherwise, the arithmetic average of the inversely proportional processing results of the container response time and resource utilization response value is input into the resource pool shrinkage ratio mapping table to output the resource pool shrinkage ratio, thereby setting the shrinkage of resource pool memory resources. Both the container response time and resource utilization response value have been de-normalized. The resource pool shrinkage ratio mapping table is pre-trained using the arithmetic average of historical inversely proportional processing results of container response time and resource utilization response value, and resource pool shrinkage ratios set by professionals based on experience rules. It describes the mapping relationship between the arithmetic average of the inversely proportional processing results of container response time and resource utilization response value and the resource pool shrinkage ratio. The shrinkage setting of resource pool memory resources can automatically reclaim idle memory resources when the system load is low or the response time latency is too long, avoiding unnecessary memory occupation by containers or services. This effectively reduces resource waste and ensures that memory resources are only used for the containers that need them, thereby improving the overall system performance.

[0047] By dynamically shrinking the memory resources in the resource pool, excessive memory usage and resource waste can be avoided. When the container response time exceeds the set maximum limit, the system will adjust the allocation of memory resources as needed, thereby ensuring that memory is used effectively and that there is not too much unused idle memory.

[0048] N2, if the fluctuation judgment result is that the load request fluctuation is abnormal, then the dynamic load scheduling configuration is triggered to adapt to the load change by dynamically adjusting the scaling strategy parameters, obtain the resource utilization detection result and determine whether to trigger the resource scheduling configuration.

[0049] The specific process for dynamic load balancing configuration is as follows: N21 inputs the load request fluctuation judgment value into the scaling ratio mapping model and the scaling rate mapping model, respectively, and outputs the scaling adjustment ratio for setting the scaling ratio and the scaling adjustment rate for setting the scaling rate. The scaling ratio mapping model is obtained by pre-training using historical load request fluctuation judgment values ​​and the scaling adjustment ratio set by professionals based on experience rules. The scaling rate mapping model is obtained by pre-training using historical load request fluctuation judgment values ​​and the scaling adjustment rate set by professionals based on experience rules.

[0050] N22 sets the initial scaling ratio and initial scaling rate based on the obtained scaling ratio and scaling rate, respectively, that is, sets the scaling ratio to the scaling adjustment ratio and the scaling rate to the scaling adjustment rate, and obtains the fluctuation judgment result for the next deployment period.

[0051] N23. ​​If the fluctuation determination result in the next deployment period is that the load request fluctuation is normal, the load request fluctuation determination value will be continuously monitored. Otherwise, the load fluctuation change amount, which reflects the degree of change of the load request fluctuation determination value corresponding to the two deployment periods, will be input into the scaling operation cooling mapping model to obtain the scaling cooling adjustment time, so as to set the initial cooling time for scaling operation. The load fluctuation change amount is the difference between the load request fluctuation determination values ​​corresponding to the two deployment periods. The scaling operation cooling mapping model is obtained by pre-training using historical load fluctuation change amounts and the scaling cooling adjustment time set by professionals based on experience rules.

[0052] Dynamic load balancing configuration helps the system intelligently respond to load fluctuations by adjusting key parameters such as scaling ratio, scaling rate, and cooldown time in real time, ensuring that resource allocation always meets actual needs. First, the scaling ratio setting dynamically adjusts the system's resource allocation based on load request fluctuations, ensuring that the system can expand resources promptly under high load and automatically reduce them under low load, avoiding resource waste and improving resource utilization. This not only improves the system's adaptability to different load scenarios but also optimizes overall resource management, ensuring stable operation even under high concurrency.

[0053] Secondly, the scaling rate setting determines the speed at which resources are expanded and contracted, balancing the system's responsiveness and stability under load fluctuations. A reasonable scaling rate avoids overreaction to load changes and reduces the oscillations that may be caused by frequent scaling operations, thus ensuring smooth system operation. By controlling the scaling rate, the system can gradually adapt to load changes, avoiding performance fluctuations caused by excessively rapid resource expansion or contraction, and ensuring that applications can provide fast response even under high loads.

[0054] Furthermore, the calculation of load fluctuations and the setting of cooldown times further optimize scaling operations, preventing frequent resource adjustments due to excessive load fluctuations. This mechanism reduces unnecessary resource expansion or contraction by delaying overreactions, ensuring that the system does not perform unnecessary operations when load changes are small, thereby reducing the risk of resource waste and configuration errors. Simultaneously, the cooldown time setting enhances system stability, preventing excessive system oscillations caused by frequent scaling operations, ensuring that the system gradually adjusts resources and smoothly responds to load changes during load fluctuations.

[0055] When load request fluctuations are determined to be normal, the system continuously monitors and adjusts resource configuration based on the actual load. Through continuous monitoring and feedback, the system can optimize resource scheduling based on real-time data, improve prediction accuracy, and thus more effectively anticipate future load changes. This intelligent dynamic load scheduling allows the system to maintain efficient and stable operation under different load scenarios, while improving user experience, reducing maintenance workload, and ensuring the rational use of resources. Overall, dynamic load scheduling configuration, through a flexible scaling mechanism, improves the system's resource utilization efficiency, stability, and scalability.

[0056] like Figure 4 The diagram shown is a structural schematic of a low-code platform application deployment system based on AI intelligent scheduling provided in an embodiment of the present invention. The low-code platform application deployment system based on AI intelligent scheduling provided in an embodiment of the present invention includes: an execution quality detection and judgment module, a load scheduling configuration and deployment module, and a resource scheduling configuration and deployment module.

[0057] Correspondingly, the execution quality detection and judgment module is used to track the execution efficiency of the application code generated by the low-code platform in order to obtain the execution quality detection results. Based on the obtained execution quality detection results, it is determined whether there is a need for initialization settings of application deployment parameters for configuring the database connection environment. The load scheduling configuration and deployment module is used to determine the load request fluctuations during deployment after the application deployment parameters are initialized if there is a demand. Based on the fluctuation determination results, it determines whether to trigger dynamic load scheduling configuration and obtains resource utilization detection results. The resource scheduling configuration deployment module is used to send an instruction to determine the resource utilization during deployment after obtaining the fluctuation judgment result if there is no demand. Based on the obtained resource utilization detection result, it determines whether to trigger the resource scheduling configuration to improve resource utilization by adjusting the deployment strategy parameters.

[0058] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0059] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0060] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0061] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0062] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0063] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0064] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for deploying low-code platform applications based on AI intelligent scheduling, characterized in that, Includes the following steps: The execution efficiency of application code generated by the low-code platform is tracked to obtain execution quality test results. Based on the obtained execution quality test results, it is determined whether there is a need for initialization settings of application deployment parameters for configuring the database connection environment. If there is a demand, the load request fluctuations during the deployment period will be judged after the application deployment parameters are initialized. Based on the fluctuation judgment results, it will be determined whether to trigger dynamic load scheduling configuration and obtain resource utilization detection results. If there is no demand, after obtaining the fluctuation judgment result, an instruction is sent to determine the resource utilization during the deployment period. Based on the obtained resource utilization detection result, it is determined whether to trigger resource scheduling configuration to improve resource utilization.

2. The low-code platform application deployment method based on AI intelligent scheduling according to claim 1, characterized in that, The specific steps for tracking the execution efficiency of application code generated by the low-code platform are as follows: Within the code execution tracing time interval, obtain the code execution error rate during the application code execution process, and compare the obtained code execution error rate with the stored maximum execution error rate limit; If the code execution error rate exceeds the maximum execution error rate limit, the execution quality test result will be recorded as unqualified, and a code execution exception message will be sent. Otherwise, the execution quality test result will be recorded as qualified, and the execution error rate deviation parameter, which reflects the degree of deviation between the code execution error rate and the maximum execution error rate limit, will be obtained; The obtained execution error rate deviation parameter is compared with the stored deviation control threshold. If the execution error rate deviation parameter is less than the deviation control threshold, it is determined that there is a need for application deployment parameter initialization settings. The application deployment parameter initialization settings used to configure the database connection environment are then performed and the fluctuation judgment result is obtained. Otherwise, it is determined that there is no need for application deployment parameter initialization settings.

3. The low-code platform application deployment method based on AI intelligent scheduling according to claim 2, characterized in that, The tracking of the execution efficiency of application code generated by the low-code platform also includes: Obtain the code execution response latency within the code execution tracing time interval; When the execution error rate deviation parameter is not less than the deviation control threshold, the obtained code execution response delay is compared with the stored code execution delay setting value: if the code execution response delay is not greater than the code execution delay setting value, the application deployment parameter initialization requirement judgment is not updated; otherwise, it is determined that there is a requirement for application deployment parameter initialization settings and the execution response deviation parameter is obtained. When the execution error rate deviation parameter is less than the deviation control threshold, the application deployment parameter initialization requirement judgment is not updated and the execution response deviation parameter is obtained. The execution response deviation parameter is used to reflect the degree of deviation between the code execution response delay and the code execution delay setting value. If there is no need to initialize application deployment parameters, continue to monitor the code execution error rate and code execution response latency of the next code execution tracing time interval; otherwise, input the result of the inverse proportional processing of the execution error rate deviation parameter and the result of the arithmetic mean processing of the execution response deviation parameter into the tracing time interval mapping table, and output the tracing time interval adjustment amount used to set the next code execution tracing time interval. In the next code execution tracing time interval, the adjustment is set based on the obtained tracing time interval adjustment.

4. The low-code platform application deployment method based on AI intelligent scheduling according to claim 2, characterized in that, The application deployment parameter initialization settings include idle connection timeout setting and connection recycling time setting; The idle connection timeout setting is specifically as follows: the result of the inverse proportional processing of the execution error rate deviation parameter and the result of the arithmetic mean processing of the execution response deviation parameter are input into the corresponding output connection idle timeout adjustment time of the connection idle timeout mapping set, and the idle connection timeout time in the connection pool is set based on the output connection idle timeout adjustment time. The connection recycling time setting is specifically as follows: input the result of the inverse proportional processing of the execution error rate deviation parameter and the result of the arithmetic mean processing of the execution response deviation parameter to the connection pool recycling time mapping set, and output the connection pool recycling adjustment time used to set the initial connection recycling time of the connection pool.

5. The low-code platform application deployment method based on AI intelligent scheduling according to claim 2, characterized in that, The specific steps for determining load request fluctuations during deployment and obtaining fluctuation determination results are as follows: During the deployment period, a request rate fluctuation determination value is obtained for determining request rate fluctuations. The request rate fluctuation determination value is expressed as the ratio of the standard deviation of the request rate to the average request rate during the deployment period. The obtained request rate fluctuation judgment value is compared with the stored maximum request rate fluctuation value; If the request rate fluctuation judgment value is greater than the maximum request rate fluctuation limit, it indicates that the fluctuation judgment result is abnormal load request fluctuation. If the request rate fluctuation judgment value is not greater than the maximum request rate fluctuation value, a concurrency effect correction factor is introduced to correct the amplification effect of different concurrency on load request fluctuation. The load request fluctuation judgment value is obtained based on the coupling result of the request rate fluctuation judgment value and the concurrency effect correction factor. Determine whether the obtained load request fluctuation value is greater than the request rate fluctuation baseline limit: If so, obtain the request failure rate used for request failure determination. If and only if the request failure rate is less than the request failure set value, it indicates that the fluctuation determination result is that the load request fluctuation is normal; otherwise, it indicates that the fluctuation determination result is that the load request fluctuation is abnormal. If not, obtain the standard deviation of the response time used for request response determination. If the standard deviation of the response time is less than the response fluctuation setting value, it indicates that the fluctuation determination result is that the load request fluctuation is normal; otherwise, it indicates that the fluctuation determination result is that the load request fluctuation is abnormal.

6. The low-code platform application deployment method based on AI intelligent scheduling according to claim 5, characterized in that, The specific process for determining whether to trigger dynamic load scheduling configuration based on the fluctuation determination result is as follows: If the fluctuation determination result is that the load request fluctuation is normal, the dynamic load scheduling configuration will not be triggered, and the resource utilization during the deployment period will be determined to obtain the resource utilization detection result. If the fluctuation determination result indicates abnormal load request fluctuation, then the dynamic load scheduling configuration, which is used to dynamically adjust the scaling strategy parameters to adapt to load changes, is triggered. The resource utilization detection result is obtained and it is determined whether to trigger the resource scheduling configuration.

7. The low-code platform application deployment method based on AI intelligent scheduling according to claim 6, characterized in that, The specific process for configuring dynamic load balancing is as follows: The load request fluctuation judgment value is input into the scaling ratio mapping model and the scaling rate mapping model respectively, and the corresponding outputs are the scaling adjustment ratio for setting the scaling ratio and the scaling adjustment rate for setting the scaling rate. Based on the obtained scaling ratio and scaling rate, perform the corresponding initial settings and obtain the fluctuation judgment results for the next deployment period. If the fluctuation determination result in the next deployment period is that the load request fluctuation is normal, the load request fluctuation determination value will be continuously monitored. Otherwise, the load fluctuation change amount, which reflects the degree of change of the load request fluctuation determination value between the two deployment periods, will be input into the scaling operation cooling mapping model to obtain the scaling cooling adjustment time, thereby setting the initial cooling time for scaling operations.

8. The low-code platform application deployment method based on AI intelligent scheduling according to claim 6, characterized in that, The resource utilization monitoring results are obtained using the following specific method: Obtain quantitative parameters of resource utilization to reflect resource utilization during the deployment period. These quantitative parameters include computing resource utilization indicators to characterize computing resource utilization performance and storage resource utilization indicators to characterize storage resource utilization performance. The computing resource utilization index is represented by the arithmetic average of the memory usage efficiency ratio and the CPU load deviation ratio, and the storage resource utilization index is represented by the arithmetic average of the disk read / write rate ratio and the storage expansion latency deviation. The resource utilization quantification parameters are weighted and coupled based on the acquired weights to obtain a resource utilization reflection value used to quantify and measure resource utilization during deployment. If the resource utilization response value is less than the minimum resource utilization limit, the resource utilization detection result is recorded as unqualified, triggering resource scheduling and configuration; otherwise, the resource utilization detection result is recorded as qualified, and resource utilization indicators are continuously monitored, calculated, stored, and fed back.

9. The low-code platform application deployment method based on AI intelligent scheduling according to claim 8, characterized in that, The specific process for resource scheduling configuration is as follows: Input resource utilization reflection values ​​into the deployment traffic allocation mapping table to output the published traffic percentage. Based on the obtained published traffic percentage, set the initial allocated traffic ratio during the application deployment process and obtain the container response time during the application deployment process. If the container response time is not greater than the set maximum container response time limit, a resource scheduling configuration completion prompt is sent; otherwise, the result of the arithmetic average of the inverse proportional processing of the container response time and resource utilization response value is input into the resource pool shrinkage ratio mapping table to output the resource pool shrinkage ratio, thereby setting the shrinkage of resource pool memory resources.

10. A low-code platform application deployment system based on AI intelligent scheduling, employing the low-code platform application deployment method based on AI intelligent scheduling as described in any one of claims 1-9, characterized in that, include: The system executes a quality detection and judgment module, a load scheduling configuration and deployment module, and a resource scheduling configuration and deployment module. The execution quality detection and judgment module is used to track the execution efficiency of application code generated by the low-code platform in order to obtain execution quality detection results. Based on the obtained execution quality detection results, it is determined whether there is a need for initialization settings of application deployment parameters for configuring the database connection environment. The load scheduling configuration deployment module is used to determine the load request fluctuations during deployment after initializing the application deployment parameters if there is a demand, and then determine whether to trigger dynamic load scheduling configuration based on the fluctuation determination result, and obtain the resource utilization detection result. The resource scheduling configuration deployment module is used to send an instruction to determine the resource utilization rate during the deployment period after obtaining the fluctuation judgment result if there is no demand, and to determine whether to trigger the resource scheduling configuration to improve the resource utilization rate based on the obtained resource utilization detection result.

Images