Disclosure of Invention
One of the technical problems to be solved by the invention is to provide a performance testing method based on an accurate service model, which can improve the efficiency of performance testing by realizing the accurate service model.
One of the technical problems to be solved by the invention is realized in the following way: a performance test method based on an accurate service model comprises the following steps:
step 10, providing transaction data through an APM, automatically acquiring the transaction data through a timing batch running task, and storing the transaction data under a specified directory path of a file server according to a certain format;
step 20, storing the transaction data into a database;
step 30, extracting relevant system data in the transaction data for visual display;
step 40, generating a corresponding business model by manually screening modeling intervals of daily trade peak time or analyzing the trade data through automatic analysis rules to obtain the corresponding business model;
and 50, outputting a service model result, carrying out accurate construction of a test model according to the service model and executing a corresponding performance test flow according to the processing capacity of the highest transaction amount evaluation system of the service model.
Further, the step 30 specifically includes: and extracting relevant system data in the transaction data, and displaying the relevant system data through a Grafana visual graphic display tool, wherein the relevant system data comprises daily transaction total amount, daily transaction amount per minute, daily transaction curve graph and daily peak period transaction duty ratio.
Further, the step 40 further includes:
step 41, selecting a manual analysis mode or an automatic analysis mode according to the requirement to obtain a corresponding service model, if the manual analysis mode is adopted, entering a step 42, otherwise, entering a step 43;
step 42, according to the displayed related system data, manually screening modeling intervals of a minute peak time point and a daily transaction peak time period to respectively obtain a minute peak service model and a daily service model data;
step 43, automatically acquiring a minute peak time point and a minute peak transaction amount in the transaction data, generating minute peak service model data, and calculating a time period with the maximum average minute transaction amount as a daily modeling interval to generate daily service model data;
step 44, matching the minute peak service model data with the existing minute peak service model data, if the matching is successful, updating the data in the existing minute peak service model, otherwise, creating a minute peak service model; and simultaneously, matching the daily service model data with the existing daily service model data, if the matching is successful, updating the data in the existing daily service model, otherwise, creating a daily service model.
Further, the step 43 further includes:
step 431, automatically executing an SQL script through a timing task to obtain a minute peak time point and a minute peak transaction amount in APM data, wherein the obtaining mode of the minute peak time point is as follows: inquiring each minute time point and each minute transaction amount within 24 hours of the whole day from a database storing APM production data through an SQL script, grouping the minute time points and arranging the minute time points in descending order of the minute transaction amount, and taking the minute time point with the maximum minute transaction amount as a minute peak time point, wherein the obtaining mode of the minute peak transaction amount is as follows: inquiring the total transaction amount of the minute peak time point from a database storing APM production data through an SQL script;
step 432, analyzing the transaction type of the minute peak, the transaction amount, the ratio of the transaction amount to the total transaction amount of the minute peak, grouping according to the transaction type, and arranging according to the descending order of the transaction amount to generate minute peak service model data;
step 433, setting a dividing line, obtaining a peak interval set of the current day according to the dividing line, and selecting a peak interval with duration exceeding a preset number of minutes to obtain a long-time interval set;
step 434, calculating the average minute trading volume of each interval in the interval set, and taking the interval with the maximum value of the average minute trading volume as a daily modeling interval;
and 435, analyzing the transaction types, the transaction amounts and the total transaction amount ratio of the transaction amounts and the daily modeling interval in the daily modeling interval through SQL, grouping according to the transaction types, and arranging according to the transaction amount descending order to generate daily business model data.
Further, the minute peak service model and the day service model in the step 44 are respectively matched in the following manner:
setting the difference value of the transaction number as X, the difference value of the transaction type number as Z, and the difference value of each transaction as Y;
judging whether the transaction number difference value of the current model and the existing model is smaller than X, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Z, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Y, if yes, judging that the current model and the existing model are the same, taking a group of model data with larger transaction amount as updated model data, and ending the flow; otherwise, judging the model as different models, and creating a new model.
Further, the method further comprises: setting an automatic new model and an automatic reminding function of model update, and outputting an updated or newly-built service model report through corresponding communication means when the service model is updated or newly-built, wherein the communication means comprise communication software, an e-mail or a company OA system.
The second technical problem to be solved by the invention is to provide a performance test system based on an accurate service model, which can improve the efficiency of performance test by realizing the accurate service model.
The second technical problem to be solved by the invention is realized in the following way: a performance testing system based on an accurate business model, comprising:
the data acquisition module is used for providing transaction data through the APM, automatically acquiring the transaction data through a timing batch running task and storing the transaction data under a specified directory path of the file server according to a certain format;
the data warehouse-in module is used for storing the transaction data into a database;
the data display module is used for extracting relevant system data in the transaction data to carry out visual display;
the business model creation module is used for generating a corresponding business model by manually screening modeling intervals of daily transaction peak time or analyzing transaction data through automatic analysis rules to obtain the corresponding business model; and
and the performance test module is used for outputting a service model result, evaluating the processing capacity of the system according to the highest transaction amount of the service model, carrying out accurate construction of the test model according to the service model, and executing a corresponding performance test flow.
Further, the data display module specifically includes: and the related system data comprises daily transaction total amount, daily transaction amount per minute, daily transaction curve graph and daily peak period transaction duty ratio.
Further, the service model creation module further comprises a selection module, a manual module, an intelligent module and a matching module;
the selection module is used for selecting a manual analysis mode or an automatic analysis mode according to the requirement to obtain a corresponding service model, executing the manual module if the manual analysis mode is adopted, and executing the intelligent module if the manual analysis mode is not adopted;
the manual module is used for respectively obtaining a minute peak service model and a daily service model data by manually screening modeling intervals of a minute peak time point and a daily transaction peak time period according to the displayed related system data;
the intelligent module is used for automatically acquiring a minute peak time point and a minute peak transaction amount in the transaction data, generating minute peak service model data, calculating a time period with the maximum average minute transaction amount as a daily modeling interval, and generating daily service model data;
the matching module is used for matching the minute peak service model data with the existing minute peak service model data, if the matching is successful, updating the data in the existing minute peak service model, otherwise, creating a minute peak service model; and simultaneously, matching the daily service model data with the existing daily service model data, if the matching is successful, updating the data in the existing daily service model, otherwise, creating a daily service model.
Further, the intelligent module specifically includes:
automatically executing an SQL script through a timing task to obtain a minute peak time point and a minute peak transaction amount in APM data, wherein the obtaining mode of the minute peak time point is as follows: inquiring each minute time point and each minute transaction amount within 24 hours of the whole day from a database storing APM production data through an SQL script, grouping the minute time points and arranging the minute time points in descending order of the minute transaction amount, and taking the minute time point with the maximum minute transaction amount as a minute peak time point, wherein the obtaining mode of the minute peak transaction amount is as follows: inquiring the total transaction amount of the minute peak time point from a database storing APM production data through an SQL script;
analyzing the transaction type of the minute peak value, the transaction amount and the total transaction amount ratio of the minute peak value, grouping according to the transaction type, and arranging according to the transaction amount in a descending order to generate minute peak value business model data;
setting a dividing line, acquiring a peak interval set of the current day according to the dividing line, and selecting a peak interval with duration exceeding a preset number of minutes to obtain a long-time interval set;
calculating the average minute trading volume of each interval in the interval set of the long time, and taking the interval with the maximum value of the average minute trading volume as a daily modeling interval;
and analyzing the transaction types, the transaction amounts and the total transaction amount ratio of the transaction amounts and the daily modeling interval in the daily modeling interval through SQL, grouping according to the transaction types, and arranging according to the transaction amount descending order to generate daily business model data.
Further, the minute peak service model and the day service model in the matching module are respectively matched in the following mode:
setting the difference value of the transaction number as X, the difference value of the transaction type number as Z, and the difference value of each transaction as Y;
judging whether the transaction number difference value of the current model and the existing model is smaller than X, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Z, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Y, if yes, judging that the current model and the existing model are the same, taking a group of model data with larger transaction amount as updated model data, and ending the flow; otherwise, judging the model as different models, and creating a new model.
Further, the system also comprises a notification module for setting an automatic new model and an automatic reminding function of model update, and when the service model is updated or newly built, the updated or newly built service model report is output through a corresponding communication means, wherein the communication means comprises communication software, an e-mail or a company OA system.
The invention has the advantages that:
1. the transaction index setting is realized, the multi-factor multi-index can be dynamically set according to the actual situation to analyze and judge the service model, the actual application of the accurate service model in the performance test scene is realized, and a real and reliable basis is provided for the performance test scene to more accord with the actual service scene of production
2. The method has the advantages that the automatic analysis of the model script is realized, the dynamic update of the performance service model of the current monitoring system is realized, the continuous update of the model, the new model and the daily comparison function are realized after one modeling, the manual complex operation is reduced, the repeatability and the hysteresis of manual comparison are avoided, the automatic update operation is realized, the accurate analysis and the model taking efficiency of the service model are obviously improved, the analysis time of the service model is reduced, and the management of the service model is facilitated.
3. And the service model report is updated and output regularly, so that the functions of dynamic tracking and automatic reminding of the service system model result are realized, and the communication cost of the performance test engineer for modeling the production and the fetch of operation staff and development staff is reduced.
Detailed Description
As shown in fig. 1, the performance testing method based on the accurate service model of the present invention includes:
step 10, providing transaction data through an APM, automatically acquiring the transaction data through a timing batch running task, storing the transaction data under a specified directory path of a file server according to a certain format, for example, dividing the format of the transaction data according to a system, separating each data field through 'I', and storing the transaction data in a file with a specified format; the step 10 further includes: and the downloading function is provided for the performance tester to automatically download data for analysis and processing, and the acquired data is deleted.
Step 20, storing the transaction data into a database, such as a MySQL database, and providing a data basis for the subsequent manual analysis verification of the service model and the accurate automatic analysis performance test service model;
step 30, extracting relevant system data in the transaction data for visual display; so as to conveniently display related system data for performance testers in real time;
step 40, generating a corresponding business model by manually screening modeling intervals of daily trade peak time or analyzing the trade data through automatic analysis rules to obtain the corresponding business model;
and 50, outputting a service model result, carrying out accurate construction of a test model according to the service model and executing a corresponding performance test flow according to the processing capacity of the highest transaction amount evaluation system of the service model. The number of the models output in the step can be one or more (the number of the models is determined according to the actual conditions and types of the minute peak service model and the daily peak time period service model), and the processing capacity of the system is evaluated according to the highest transaction amount of the models, so that the accurate construction of the test model can be completed according to the service model through the subsequent performance test.
Preferably, the step 30 specifically includes: and extracting relevant system data in the transaction data, and displaying the relevant system data through a Grafana visual graphic display tool, wherein the relevant system data comprises daily transaction total amount, daily transaction amount per minute, daily transaction curve graph and daily peak period transaction duty ratio. The visualization tool Grafana is a cross-platform open source measurement analysis and visualization tool, and can perform visual display of APM data by inquiring collected data and then performing visual display, so that a function of providing visual page display for manually analyzing a service model is realized, and a performance tester can conveniently analyze an application system service model.
Preferably, the step 40 further includes:
step 41, selecting a manual analysis mode or an automatic analysis mode according to the requirement to obtain a corresponding service model, if the manual analysis mode is adopted, entering a step 42, otherwise, entering a step 43;
step 42, according to the displayed related system data, manually screening modeling intervals of a minute peak time point and a daily transaction peak time period to respectively obtain minute peak service model data and daily service model data, and displaying transaction types and transaction occupation ratios of the minute peak time point and the peak time period;
step 43, automatically acquiring a minute peak time point and a minute peak transaction amount in the transaction data, generating minute peak service model data, and calculating a time period with the maximum average minute transaction amount as a daily modeling interval to generate daily service model data; according to the manual analysis business model experience, designing an automatic analysis business model rule, and completing an automatic analysis accurate business model according to APM actual transaction amount data;
step 44, matching the minute peak service model data with the existing minute peak service model data, if the matching is successful, updating the data in the existing minute peak service model, otherwise, creating a minute peak service model; and simultaneously, matching the daily service model data with the existing daily service model data, if the matching is successful, updating the data in the existing daily service model, otherwise, creating a daily service model.
Preferably, the step 43 further includes:
step 431, automatically executing an SQL script through a timing task to obtain a minute peak time point and a minute peak transaction amount in APM data, wherein the obtaining mode of the minute peak time point is as follows: inquiring each minute time point and each minute transaction amount within 24 hours of the whole day from a database storing APM production data through an SQL script, grouping the minute time points and arranging the minute time points in descending order of the minute transaction amount, and taking the minute time point with the maximum minute transaction amount as a minute peak time point, wherein the obtaining mode of the minute peak transaction amount is as follows: inquiring the total transaction amount of the minute peak time point from a database storing APM production data through an SQL script;
step 432, analyzing the transaction type of the minute peak, the transaction amount, the ratio of the transaction amount to the total transaction amount of the minute peak, grouping according to the transaction type, and arranging according to the descending order of the transaction amount to generate minute peak service model data;
step 433, setting a dividing line, obtaining a peak interval set of the current day according to the dividing line, and selecting a peak interval with duration exceeding a preset number of minutes to obtain a long-time interval set; for example, take 80% of the minute peak transaction amount (the value can be adjusted according to the need) as a dividing line, take a daily peak section as shown in fig. 2 in a peak section set (1 or N sections can be obtained), intercept an ascending section (the abscissa is a minute time axis) when the current day exceeds the dividing line, and intercept rules of the abscissa of the ascending section (i.e. the peak section) are: taking the interval set of time abscissas of the nth (N > =1 and odd) and the (n+1) th times exceeding 80% of the minute peak transaction amount (e.g. 06:20-09:05 peak interval, 14:30-17:30 peak interval) according to the minute time axis; selecting a set of interval formabilities with duration exceeding a preset minute M (M is configurable) to eliminate short time intervals of steep rise or dip peaks;
step 434, calculating the average minute trading volume of each interval in the interval set, and taking the interval with the maximum value of the average minute trading volume as a daily modeling interval; the interval average minute transaction amount is equal to the time abscissa interval transaction total amount divided by the time abscissa interval minute number;
and 435, analyzing the transaction types, the transaction amounts and the total transaction amount ratio of the transaction amounts and the daily modeling interval in the daily modeling interval through SQL, grouping according to the transaction types, and arranging according to the transaction amount descending order to generate daily business model data.
Preferably, as shown in fig. 3, the minute peak service model and the day service model in the step 44 are respectively matched in the following manner:
setting the difference value of the transaction number as X, the difference value of the transaction type number as Z, and the difference value of each transaction as Y;
judging whether the transaction number difference value of the current model and the existing model is smaller than X, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Z, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Y, if yes, judging that the current model and the existing model are the same, taking a group of model data with larger transaction amount as updated model data, and ending the flow; otherwise, judging the model as different models, and creating a new model.
The transaction number difference value is the difference between the transaction number of the current model and the transaction number of the existing model divided by the transaction number of the existing model, the transaction type number difference value is the difference between the transaction type number of the current model and the transaction type number of the existing model divided by the transaction type number of the existing model, and the transaction duty ratio difference value is the difference between the transaction duty ratio of the current model and the transaction duty ratio of the existing model.
In general, a model corresponds to a service scene, if the model is the same, that is, the model which indicates that the service scene has a corresponding model, only part of data in the existing model needs to be updated, and if the model is different, that indicates that the service scene has no corresponding model, a new model needs to be created. By the matching mode, the current minute peak service model is matched with the existing minute peak service model, and the current day service model is matched with the existing day service model, so that the accuracy of model data in different service scenes can be ensured.
Preferably, the method further comprises: setting an automatic new model and an automatic reminding function of model update, and outputting an updated or newly-built service model report through corresponding communication means when the service model is updated or newly-built, wherein the communication means comprise communication software, an e-mail or a company OA system. By updating the existing model and creating the automatic reminding function of the new model, the communication cost of the performance test engineer for production and fetch modeling from operation and maintenance personnel and development personnel can be reduced.
As shown in fig. 4, a performance test system based on an accurate service model of the present invention includes:
the data acquisition module is used for providing transaction data through the APM, automatically acquiring the transaction data through a timing batch running task, storing the transaction data under a specified directory path of the file server according to a certain format, for example, dividing the format of the transaction data according to a system, separating each data field through 'I', and storing the transaction data in a file with a specified format; the data acquisition module further includes: providing a downloading function for performance testing personnel to automatically download data for analysis and processing, and deleting the fetched data;
the data storage module is used for storing the transaction data into a database, such as a MySQL database, and providing a data basis for the follow-up manual analysis and verification of the service model and the accurate automatic analysis performance test service model;
the data display module is used for extracting relevant system data in the transaction data to carry out visual display; so as to conveniently display related system data for performance testers in real time;
the business model creation module is used for generating a corresponding business model by manually screening modeling intervals of daily transaction peak time or analyzing transaction data through automatic analysis rules to obtain the corresponding business model; and
and the performance test module is used for outputting a service model result, evaluating the processing capacity of the system according to the highest transaction amount of the service model, carrying out accurate construction of the test model according to the service model, and executing a corresponding performance test flow. The number of the models output in the step can be one or more (the number of the models is determined according to the actual conditions and types of the minute peak service model and the daily peak time period service model), and the processing capacity of the system is evaluated according to the highest transaction amount of the models, so that the accurate construction of the test model can be completed according to the service model through the subsequent performance test.
Preferably, the data display module specifically includes: and the related system data comprises daily transaction total amount, daily transaction amount per minute, daily transaction curve graph and daily peak period transaction duty ratio. The visualization tool Grafana is a cross-platform open source measurement analysis and visualization tool, and can perform visual display of APM data by inquiring collected data and then performing visual display, so that a function of providing visual page display for manually analyzing a service model is realized, and a performance tester can conveniently analyze an application system service model.
Preferably, the service model creation module further comprises a selection module, a manual module, an intelligent module and a matching module;
the selection module is used for selecting a manual analysis mode or an automatic analysis mode according to the requirement to obtain a corresponding service model, executing the manual module if the manual analysis mode is adopted, and executing the intelligent module if the manual analysis mode is not adopted;
the manual module is used for respectively obtaining a minute peak service model and a day service model data by manually screening modeling intervals of a minute peak time point and a day transaction peak time period according to the displayed related system data, and displaying transaction types and transaction occupation ratios of the minute peak time point and the peak time period;
the intelligent module is used for automatically acquiring a minute peak time point and a minute peak transaction amount in the transaction data, generating minute peak service model data, calculating a time period with the maximum average minute transaction amount as a daily modeling interval, and generating daily service model data; according to the manual analysis business model experience, designing an automatic analysis business model rule, and completing an automatic analysis accurate business model according to APM actual transaction amount data;
the matching module is used for matching the minute peak service model data with the existing minute peak service model data, if the matching is successful, updating the data in the existing minute peak service model, otherwise, creating a minute peak service model; and simultaneously, matching the daily service model data with the existing daily service model data, if the matching is successful, updating the data in the existing daily service model, otherwise, creating a daily service model.
Preferably, the intelligent module specifically includes:
automatically executing an SQL script through a timing task to obtain a minute peak time point and a minute peak transaction amount in APM data, wherein the obtaining mode of the minute peak time point is as follows: inquiring each minute time point and each minute transaction amount within 24 hours of the whole day from a database storing APM production data through an SQL script, grouping the minute time points and arranging the minute time points in descending order of the minute transaction amount, and taking the minute time point with the maximum minute transaction amount as a minute peak time point, wherein the obtaining mode of the minute peak transaction amount is as follows: inquiring the total transaction amount of the minute peak time point from a database storing APM production data through an SQL script;
analyzing the transaction type of the minute peak value, the transaction amount and the total transaction amount ratio of the minute peak value, grouping according to the transaction type, and arranging according to the transaction amount in a descending order to generate minute peak value business model data;
setting a dividing line, acquiring a peak interval set of the current day according to the dividing line, and selecting a peak interval with duration exceeding a preset number of minutes to obtain a long-time interval set; for example, take 80% of the minute peak transaction amount (the value can be adjusted according to the need) as a dividing line, take a daily peak section as shown in fig. 2 in a peak section set (1 or N sections can be obtained), intercept an ascending section (the abscissa is a minute time axis) when the current day exceeds the dividing line, and intercept rules of the abscissa of the ascending section (i.e. the peak section) are: taking the interval set of time abscissas of the nth (N > =1 and odd) and the (n+1) th times exceeding 80% of the minute peak transaction amount (e.g. 06:20-09:05 peak interval, 14:30-17:30 peak interval) according to the minute time axis; selecting a set of interval formabilities with duration exceeding a preset minute M (M is configurable) to eliminate short time intervals of steep rise or dip peaks;
calculating the average minute trading volume of each interval in the interval set of the long time, and taking the interval with the maximum value of the average minute trading volume as a daily modeling interval; the interval average minute transaction amount is equal to the time abscissa interval transaction total amount divided by the time abscissa interval minute number;
and analyzing the transaction types, the transaction amounts and the total transaction amount ratio of the transaction amounts and the daily modeling interval in the daily modeling interval through SQL, grouping according to the transaction types, and arranging according to the transaction amount descending order to generate daily business model data.
Preferably, the minute peak service model and the day service model in the matching module are respectively matched in the following modes:
setting the difference value of the transaction number as X, the difference value of the transaction type number as Z, and the difference value of each transaction as Y;
judging whether the transaction number difference value of the current model and the existing model is smaller than X, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Z, if yes, judging whether the transaction type number difference value of the current model and the existing model is smaller than Y, if yes, judging that the current model and the existing model are the same, taking a group of model data with larger transaction amount as updated model data, and ending the flow; otherwise, judging the model as different models, and creating a new model.
The transaction number difference value is the difference between the transaction number of the current model and the transaction number of the existing model divided by the transaction number of the existing model, the transaction type number difference value is the difference between the transaction type number of the current model and the transaction type number of the existing model divided by the transaction type number of the existing model, and the transaction duty ratio difference value is the difference between the transaction duty ratio of the current model and the transaction duty ratio of the existing model.
In general, a model corresponds to a service scene, if the model is the same, that is, the model which indicates that the service scene has a corresponding model, only part of data in the existing model needs to be updated, and if the model is different, that indicates that the service scene has no corresponding model, a new model needs to be created. By the matching mode, the current minute peak service model is matched with the existing minute peak service model, and the current day service model is matched with the existing day service model, so that the accuracy of model data in different service scenes can be ensured.
Preferably, the system further comprises a notification module for setting an automatic new model and an automatic reminding function of model update, and when the service model is updated or newly built, the updated or newly built service model report is output through a corresponding communication means, wherein the communication means comprises communication software, an email or a company OA system. By updating the existing model and creating the automatic reminding function of the new model, the communication cost of the performance test engineer for production and fetch modeling from operation and maintenance personnel and development personnel can be reduced.
According to the invention, a normalization system for continuously updating the performance test service model is formed by constructing a performance test accurate service model from related transaction service volume statistics, service volume analysis, service transaction volume peak value accurate positioning, manual auxiliary verification analysis, performance model automatic analysis comparison, service model modeling, dynamic tracking reminding and model updating, the accurate matching of the production multi-service model is realized from service analysis to model dynamic updating and automatic comparison, the types of the service model are enriched, a real and reliable basis is provided for the performance test scene to more accord with the actual production service scene, and the overall efficiency of the performance test is improved.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that the specific embodiments described are illustrative only and not intended to limit the scope of the invention, and that equivalent modifications and variations of the invention in light of the spirit of the invention will be covered by the claims of the present invention.