System architecture management method based on link tracking

Abstract

The invention discloses a system architecture management method based on link tracking. The method comprises the following steps: collecting system operation related data; preprocessing the collected data, and integrating the data into structured data; calculating system performance indexes of the system according to the structured data, and perfecting architecture information; generating a system relation graph according to the architecture information in combination with a front-end framework; judging whether the current system is subjected to capacity expansion or capacity reduction or not according to the system performance indexes, if yes, carrying out capacity expansion or capacity reduction according to the system performance indexes, and analyzing the influence range of capacity expansion or capacity reduction operation according to a system relation graph so as to synchronously carry out capacity expansion or capacity reduction on a system associated with the current system. According to the invention, the maintenance of the architecture diagram is more accurate and real-time, the system health degree is supported by data, and the system expansion and contraction capacity has data basis and is more effective; and meanwhile, the whole-line automatic process can greatly reduce the labor cost.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a system architecture management method based on link tracing. Background Art

[0002] With the rapid development of the internet and cloud computing, distributed systems are becoming increasingly complex. In such systems, the call relationships between services are intricate and complex. Traditional architecture assets rely on manual maintenance, which cannot guarantee data accuracy and is labor-intensive. Furthermore, manual management capabilities are extremely limited, making it difficult to effectively manage system deployment architectures. Summary of the Invention

[0003] The purpose of the present invention is to provide a system architecture management method based on link tracing to address the shortcomings of the existing technology.

[0004] To achieve the above objectives, the present invention provides a system architecture management method based on link tracing, comprising:

[0005] Collect system operation related data;

[0006] Preprocess the collected data and integrate it into structured data;

[0007] Calculate system performance indicators of the system based on the structured data and improve the architecture information;

[0008] Generate a system relationship diagram based on the architecture information and in combination with the front-end framework;

[0009] Determine whether the current system is to be expanded or reduced based on the system performance indicators. If so, expand or reduce the capacity based on the system performance indicators, and analyze the scope of the impact of the expansion or reduction operation based on the system relationship diagram to synchronously expand or reduce the system associated with the current system.

[0010] Furthermore, the system operation related data includes system interface call information, system response time and concurrency.

[0011] Furthermore, an interface agent for data collection is embedded in the virtual machine template or the basic image template of the cloud platform to complete the deployment synchronously with the application deployment process.

[0012] Furthermore, the collected data is pre-processed including cleaning, and data deduplication is performed according to the request system abbreviation, response system abbreviation, request IP and response IP.

[0013] Furthermore, the system performance indicators include the number of requests per unit time and system health indicators;

[0014] The calculation method of the number of requests per unit time is as follows:

[0015] qps = sum(F) / min

[0016] Among them, qps is the number of requests per unit time, sum(F) is the total number of requests, and min is 1 minute;

[0017] The calculation method of the system health index is as follows:

[0018] Apdex = (the number of satisfied + 0.5 * the number of tolerable) / qps

[0019] Among them, Apdex is the calculated system health index. The number of satisfied is the total number of interface responses where the actual response time is less than the required value T of the service response time of the service. The number of tolerable is the total number of interface responses where the response time is between the required value T of the service response time of the service and 4 * the required value T of the service response time of the service.

[0020] Furthermore, the method for determining whether to scale out the current system according to the system performance index is as follows:

[0021] Judge whether the system health index Apdex is less than the set lower health threshold. If it is less than the set threshold, trigger an alarm, and judge whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is greater than the set upper ratio threshold. If so, scale out. The first set period is greater than the second set period;

[0022] The method for determining whether to scale in the current system according to the system performance index is as follows:

[0023] Judge whether the system health index Apdex is 1 within the set period. If it is 1 within the set period, judge whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is less than the set lower ratio threshold. If so, scale in.

[0024] Furthermore, the first set period is one year, the second set period is one month, the lower health threshold and the upper ratio threshold are both 0.8, and the lower ratio threshold is 0.4.

[0025] Beneficial effects: Compared with the relatively independent work of traditional development, operation and maintenance, and monitoring, the present invention organically combines the architecture maintenance of the application system with application operation and maintenance and system monitoring, realizing a closed-loop full-line implementation from architecture to operation and maintenance to monitoring. Relying on the actual system operation status, the maintenance of the architecture diagram is more accurate and real-time, the system health degree is supported by data, and the system scaling has data basis and is more effective; at the same time, the fully automated process can greatly reduce the labor cost. Brief Description of the Drawings

[0026] Figure 1 It is a schematic flowchart of a system architecture management method based on link tracing according to an embodiment of the present invention. Detailed Embodiments

[0027] The present invention will be further clarified below with reference to the drawings and specific embodiments. These embodiments are implemented on the premise of the technical solution of the present invention, and it should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention.

[0028] As Figure 1 shown, an embodiment of the present invention provides a system architecture management method based on link tracing, including:

[0029] Collect data related to system operation. Specifically, the data related to system operation includes system interface call information, system response time, and concurrency. An interface agent for data collection can be embedded in the virtual machine template or the base image template of the cloud platform to complete the deployment synchronously with the application deployment process. Data collection is completed during the system operation and pushed to the database of the present system. In addition, due to the large amount of data, a distributed database is preferably used in the system.

[0030] Preprocess the collected data and integrate it into structured data. The preprocessing of the collected data includes cleaning and data deduplication according to the abbreviations of the requesting system, the abbreviations of the responding system, the requesting IP, and the responding IP.

[0031] Calculate the system performance indicators of the system according to the structured data, and improve the architecture information, such as system information, subsystem information, service information, service node information. According to the requesting system and responding system fields, system relationship information can be maintained, etc. Data such as system relationship information and system information can be stored in the graph database. Specifically, the system performance indicators include the number of requests per unit time and the system health degree indicator;

[0032] The calculation method of the number of requests per unit time is as follows:

[0033] qps = sum(F) / min

[0034] Among them, qps is the number of requests per unit time, sum(F) is the total number of requests, and min is 1 minute.

[0035] The calculation method of the above system health index is as follows:

[0036] Apdex = (the number of satisfied requests + 0.5 * the number of tolerable requests) / qps

[0037] Among them, Apdex is the calculated system health index. The number of satisfied requests refers to the total number of interface responses whose actual response time is less than the required value T of the service response time of the business. The number of tolerable requests refers to the total number of interface responses whose response time is between the required value T of the service response time of the business and 4 * the required value T of the service response time of the business. The Apdex index is an industry standard developed by the Apdex Alliance for evaluating application performance. It can quantify the performance of an application and provide data reference for the stability monitoring and expansion of the system. The Apdex value is between 0 and 1, and the smaller the value, the worse the performance of the application system.

[0038] Generate a system relationship diagram according to the architecture information and in combination with the front-end framework. Specifically, according to the system access relationships stored in the database, for a certain system, the list of systems that call this system and the list of systems called by this system can be queried. Obtain the above data through the browser side, represent the call relationships with line segments, represent the systems involved with rectangles plus text, and construct a system relationship diagram on the display page.

[0039] Judge whether to expand or contract the current system according to the system performance index. If so, expand or contract according to the system performance index, and analyze the scope affected by the expansion or contraction operation based on the system relationship diagram to synchronously expand or contract the systems associated with the current system.

[0040] Specifically, the method for judging whether to expand the current system according to the system performance index is as follows:

[0041] Judge whether the system health index Apdex is less than the set lower threshold of the health index. If it is less than the set threshold, trigger an alarm, and judge whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is greater than the set upper threshold of the ratio. If so, perform expansion. Expansion means increasing the number of service nodes. The specific number of service nodes to be increased can be defined by oneself, and it can be combined with the actual CPU usage rate of the service nodes (e.g., >60%, which can be defined according to the actual situation) and the memory usage rate (>80%, which can be defined according to the actual situation) to increase the number of service nodes in a certain proportion.

[0042] The method for judging whether to contract the current system according to the system performance index is as follows:

[0043] Determine whether the system health indicator Apdex is 1 within the set period. If it is 1 within the set period, then determine whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is less than the lower limit of the set ratio threshold. If so, perform capacity reduction. Capacity reduction means reducing the number of service nodes, and it can also recover and reduce the number of service nodes in a certain proportion in combination with the actual CPU usage rate of the service nodes (e.g., <30%, which can be defined according to the actual situation) and the memory usage rate (<50%, which can be defined according to the actual situation).

[0044] Preferably, the above-mentioned first set period is one year, the second set period is preferably one month, the set period can be a quarter, the lower threshold of the health degree and the upper limit of the ratio threshold are both preferably 0.8, and the lower limit of the ratio threshold is preferably 0.4. The above parameters can all be adjusted according to the actual situation.

[0045] The above is only the preferred implementation manner of the present invention. It should be pointed out that for those of ordinary skill in the art in this technical field, the other parts not specifically described belong to the prior art or common general knowledge. Without departing from the principle of the present invention, several improvements and refinements can also be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A system architecture management method based on link tracing, characterized in that, Including: Collecting operation-related data of the system; Preprocessing the collected data and integrating it into structured data; Calculating the system performance metrics of the system based on the structured data and improving the architecture information; Generating a system relationship diagram based on the architecture information and in combination with a front-end framework; Judging whether to scale up or scale down the current system according to the system performance metrics. If so, scale up or scale down according to the system performance metrics, and analyze the scope affected by the scale-up or scale-down operation based on the system relationship diagram, so as to synchronously scale up or scale down the systems associated with the current system.

2. The system architecture management method based on link tracing according to claim 1, wherein, The operation-related data of the system includes system interface call information, system response time, and concurrency.

3. A system architecture management method based on link tracing according to claim 1, characterized in that, Embedding an interface agent for data collection in the virtual machine template or the base image template of the cloud platform to complete the deployment synchronously along with the application deployment process.

4. A system architecture management method based on link tracing according to claim 1, characterized in that, Preprocessing the collected data includes cleaning and deduplicating the data according to the abbreviations of the request system, the response system, the request IP, and the response IP.

5. A system architecture management method based on link tracing according to claim 1, characterized in that, The system performance metrics include the number of requests per unit time and the system health metric; The calculation method of the number of requests per unit time is as follows: qps = sum(F) / min where qps is the number of requests per unit time, sum(F) is the total number of requests, and min is 1 minute; The calculation method of the system health metric is as follows: Apdex = (the number of satisfied requests + 0.5 * the number of tolerable requests) / qps where Apdex is the calculated system health metric, the number of satisfied requests refers to the total number of interface responses with the actual response time less than the required value T of the service response time of the business, and the number of tolerable requests refers to the total number of interface responses with the response time between the required value T of the service response time of the business and 4 * the required value T of the service response time of the business.

6. A system architecture management method based on link tracing according to claim 5, characterized in that, The method for judging whether to scale up the current system according to the system performance metrics is as follows: Judging whether the system health metric Apdex is less than the set lower threshold of the health metric. If it is less than the set threshold, trigger an alarm, and judge whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is greater than the set upper threshold of the ratio. If so, scale up. The first set period is longer than the second set period; The method for judging whether to scale down the current system according to the system performance metrics is as follows: Judging whether the system health metric Apdex is 1 within the set period. If it is 1 within the set period, judge whether the ratio of the maximum number of requests per unit time in the first set period in the past to the average QPS in the second set period is less than the set lower threshold of the ratio. If so, scale down.

7. A system architecture management method based on link tracing according to claim 6, characterized in that, The first set period is one year, the second set period is one month, the lower threshold of the health metric and the upper threshold of the ratio are both 0.8, and the lower threshold of the ratio is 0.4.

Images