Bank software testing workload analysis and evaluation methods

As an emerging profession, banksSoftware TestingThe knowledge system integrating banking business and software testing is becoming more and more important in the domestic banking industry. The testing team is also growing. Currently, large commercial banks in China all have their own dedicated testing teams.

As a cost center, the testing department must share the profit of the business department. With the development and growth of the testing team, bank decision makers need to seriously consider the configuration problem between the business development needs and the size of the testing department. Therefore, scientific testing is required.WorkMeasure the test taker's knowledge about the evaluation methods.

According to the human resources configuration model of the IT industry, the ratio of testers to developers is 1/3-1/5. According to the actual configuration of China's banking industry, the number of testers is almost unable to meet such requirements.

Testing staff's work t how to analyze and evaluate to ensure that the testing staff quantity configuration is scientific. This is a problem that needs to be studied in the software testing of banks.

I. Features of bank Testing

Unlike professional testing companies, banking software testing is different from the testing work of professional testing companies due to factors such as organizational structure, human resource management model, system complexity, and special requirements of banking services. large.

1. Different organizational structures lead to different testing responsibilities

As a functional department within the Bank, testing is undoubtedly the core responsibility of the testing department. In other words, the acceptance of the test task is not optional. Regardless of the version quality and test conditions, the test should be carried out on schedule;

The professional testing company determines whether to undertake the project based on the complexity of its personnel structure and version. Therefore, the responsibility of the Bank testing department is different from that of the professional testing company.

2. The human resource cost accounting methods are different.

As a self-financing testing enterprise, the cost accounting of professional testing companies will calculate human resource costs based on factors such as the evaluation quality of testing software and the goal and duration of ensuring product quality; the cost accounting of bank testers does not directly correspond to the project. The human investment of bank testers is relatively fixed in a specific period.

The differences in the concept of cost determine the depth of tests for different versions and projects.

3. Integration of banking work and Systems

The difficulty of the test depends on the high risk. The special commodities such as currency operating in the Bank determine that the test should not only achieve the goal of perfection of product functions, but also prove the availability and security of the system. Therefore. The test involves a wide range of content;

The computer system structure of banks is complex. Multiple applications have high Disaster Tolerance and strong relevance. It determines that the test process is abnormal and complex. Compared with testing companies, they have higher requirements on the professional level of testers, that is, bank testers are compound talents familiar with banking and testing technologies.

4. the continuous operation of the Bank determines the tester

With the continuous operation and expansion of banks, the demand of business departments is also changing. New requirements generate new functions, which lead to new test requirements. However, the testing objects are optimized and improved in part or individually when the main functions remain unchanged.

To this end. Testers are required to be proficient in banking and the testers need to be relatively fixed. For example, a specialist who is engaged in gold testing is difficult to perform accounting testing in the short term, if the business system architecture of the bank does not change, the test should be conducted based on long-term division of labor and continuous development.

5. Different testing objectives and standards

The testing objectives to be completed by the Banking testing department should not only verify the completeness of the functions, but also conduct adaptive testing. That isFunction TestingIn addition, comprehensive tests must be carried out to meet performance and capacity requirements. The goal is that software products meet the production needs of the business department. Do not generate incidents of production safety responsibilities.

In view of the characteristics of the banking software testing department, the workload evaluation of banking software testing is different from that of professional software testing companies.

2. Common Software Testing workload evaluation methods

The test workload is affected by many factors, such as the test content, test method, quality requirements, and test phase. The differences in test workload are very large. This article mainly describes the following methods to evaluate the test workload.

1. DelPhi Method

The elPhi method is a qualitative evaluation method developed by experts based on estimation of workload based on specific work experience. The specific evaluation process is as follows:

(1) The Head of the workload evaluation team should provide the experts with the project specifications and estimation forms:

(2) organize experts to discuss in detail the factors related to scale:

(3) Fill in the estimation form anonymously by experts;

(4) summarize the experts' opinions and return the conclusions to the experts:

(5) Experts discuss large differences in estimation;

(6) The experts re-assess the differences until the differences gradually contract and reach an agreement.

The oelPhi method is a targeted evaluation method without historical data and is easy to operate. It is used to evaluate the workload of the new test project and can be used to test the workload budget, the test plan and guidelines are also prepared.

The disadvantage of the elPhi method is that the accuracy is not high. The experience and style of the Expert Group members and their different personalities will lead to a large gap in the evaluation results.

2. Proportional Evaluation Method

Evaluate the testing workload in proportion based on the number of tasks undertaken by the development. The empirical workload of development and testing in the industry is allocated to 80%-65% of the total workload of development, and 20%-35% of the total workload of testing. The proportional evaluation method is the workload allocation based on the full life cycle model of software. This is a quantitative result from the summary and analysis of a large amount of historical data.

Estimate the test workload based on the Development workload is relatively accurate. This method is suitable for the full life cycle of software computing when software development companies undertake software development projects.

The disadvantage is that the development team and test team basically match the maturity of this method. Once there is a difference in maturity, there is a large gap between the most evaluated results of the work.

3. WBS Evaluation Method

WBS (WorkBreakdownStrueture, a work breakdown structure) breaks down a project into deliverables or into smaller, manageable, and normal components, until work and deliverables are defined at the level of the work package. The procedure is as follows:

(1) layer-by-layer decomposition of test items:

(2) The final decomposition is an action that cannot be further divided;

(3) estimate the time required for each action:

(4) summarize the workload step by step:

(5) Calculate the final test workload. WBs evaluation method is a more accurate workload evaluation method, which can be used to test the workload settlement activities. W-S is an important part of project management activities. The WBS evaluation method is the most accurate method for evaluating the current test workload. It not only completes workload evaluation. At the same time, the preparation of the test work plan has been completed, and the application of this method to evaluate the work volume can achieve the same effect.

However, this method has its obvious disadvantages: First, it takes a lot of time and effort to compile the WBS, and the investment is huge. Second, if the WBS compilation is unreasonable, the evaluation error will be very large; third, in the absence of workload quota data, the workload of a unit action can only be estimated because the Unit action does not have the corresponding labor volume data. The workload estimate is slightly different, and the final summary results will be significantly different.

There are still many methods to evaluate the test workload. For example, the Project Experience Comparison Evaluation Method evaluated based on previous testing experience or historical data is not listed here because each workload evaluation method has its own advantages and disadvantages. Based on the comprehensive comparison of various test workload evaluation methods and the work characteristics of the bank test, this paper designs a new test workload evaluation method-test work quota evaluation method.

Iii. Evaluation of testing work quota

The test work quota evaluation method is to break down test tasks into non-sharable activities. Work hours can be calculated based on actual work days or simulated operations, and a work-hours quota table is prepared. A workload evaluation method is used to calculate the test workload.

1. Split the test task into specific activities. Based on the WBS Method of project management, the test project is divided into various test actions, and then the test actions are subdivided into unclassified activities.The Bank's adaptive testing project is roughly divided into the following actions.

1) Test the forward action. Measure the test taker's knowledge about project design, R & D, coding, unit, integration, and system testing. study business requirements and software requirements in detail and compile test cases based on application transformation and interface transformation. These actions can be divided into the following activities: 1. Project Development Survey; 2. Demand Analysis and Evaluation; 3. Case Design and compilation, you can write a single case based on a specific transaction.

2) test plan action. Plan the project in detail, compile the test plan, discuss, review the Plan, and release and implement the plan. Decomposed activities include: 1. Overall Planning of various environments; 2. Preparation of plans for various environments; 3. discussion and revision of plans; 4. Promotion and implementation of plans.

3) test preparation. Test Document compilation and review, test environment preparation and configuration, parameter installation and data migration. Decomposed activities include: first, writing test cases; second, reviewing and training test cases; and third, configuring and debugging the test environment: iv. Compilation, check and installation of parameter text. V. Compilation of the migration text, and check of the migration and migration results.

4) test implementation. This is the most frequently used and core operation maintenance testing environment during the testing process, including functional testing, non-functional testing, regression testing, routine testing, and patch testing. Activities that can be decomposed during testing are divided into individual activities by test cases or transactions.

5) project production operations. The project production operation is the final stage of the test project. The most critical actions of the test project can be divided into the following: first, the preparation of the production plan; and second, the compilation of the production documentation: 3. production verification scheme and implementation verification: 4. Production support.

2. Calculate the work-hours quota for testing activities

The so-called work-hour quota refers to the standard work-hour quota required for a single activity. It is the time consumption of the unit activity that is finally calculated through the induction and analysis of historical data, there are many ways to generate the quota for collecting historical data. In this paper, two quota calculation methods are proposed.

1) The work realism quota method records the testing staff's actions and the time corresponding to the actions on site to summarize the total time consumption of all actions of the realistic object, and combines the time consumption of the same activity. Select valid time and invalid time, and compare and analyze the action analysis results. Finally, calculate the work-hour quota for the unit activity.

The team members record realistic data in a relatively standard, real, and accurate manner, but time-consuming and laborious: the actual information entered by employees is relatively poor in terms of standardization and information authenticity. However, after the information is collected and summarized, the data can be corrected by methods such as comparative method, comparative method, and comparative method to achieve the expected results. The specific implementation steps of the Work realistic quota method are as follows.

① Compile a realistic worksheet. The Worksheet should contain activity names, activity categories, work hours, and activity associated objects.

② Realism. Describe the activities of testers on site and record the time invested in individual activities.

③ Summary statistics. Summarize all the realistic data to get the time consumed by each activity. Add the activity items not compiled in the worksheet to improve the realistic table.

④ Data analysis. This is the most complex task in a realistic activity. Three elements need to be analyzed:

First, the authenticity of the data is based on the actual information provided by the employee. Some information may be inaccurate, such as exaggerating the labor cost. This can be determined by comparing the discrete degree of the average time consumption of the same activity;

Second, data complexity analysis. For example, the test time of a single transaction. The labor cost difference between online transactions and batch transactions is very large. Batch transaction testing, some of which can be completed in batches, and some of which can be completed in batches. The complexity of the same activity varies greatly;

The third is to analyze the correlations and logical relationships between services and analyze the normal workload range through logical relationships.

⑤ Calculate the work-hour quota. Based on the actual working hours. After the invalid working hours are eliminated, the hourly quota for a single activity is obtained.

⑥ Improve the work-hour quota. Regularly revise the work-hour quota based on the maturity and productivity development of the testing team.

(2) The simulation operation quota method extracts sample operators from various testing activities, simulates various testing activities, and records the operation time of various testing activities, and finally sums up the standard time of each activity. Precautions for implementing simulated operations:

First, the scope of sample selection. Because the test team is composed of personnel at all levels, the sample points should cover the personnel at all levels;

Second, the operation skill of the sample. Because the quota level needs to be calculated, the operation skill of the sample should be at the average deviation level of the test team, and should not be too high or too low, otherwise, the quota standards will be affected;

The third is the test time selection, because the work-hour quota is restricted by the environment. The simulated operation time should be selected during the test. In this way, the impact of the environment on the quota can be reproduced.

The specific implementation steps of the simulated operation method are as follows.

① Prepare a simulated activity table to list the test activities in detail.

② Select an appropriate amount of samples at each level of the test team. It should be noted that different test activities can be completed by different sample operators.

③ Perform a time test on the testing activities. Take the average time of the sample activity at each level, and calculate the Working Hours quota of the personnel at this level. Take the average time of all samples for the activity to calculate the Working Hours quota of the activity.

④ Data correction: Due to the minor differences in the simulated operation process, the level of the quota will be greatly affected. Therefore, we must modify the test data based on the specific conditions of the simulated operation.