With the expansion of the short-term ecosystem and business development, the decision-making scenarios undertaken by data in the business are becoming more and more diversified, and some data has been applied to high-risk scenarios such as asset loss and high customer complaints, so the requirements for data quality, especially for high-risk scenarios, are very high.
Many QA data BP teams face the following pain points in the assurance process:
1.There is no standardized process, and the management and control ability is weak
The R&D QA manpower ratio is as high as 20:1, so QA adopts a hierarchical guarantee strategy, and QA manpower is inclined to high-risk needs and asset changes, but each business has its own definition of the hierarchical guarantee process, and the landing method is mostly offline communication manual constraints, lacking standardized processes
Demand-based R&D testing relies on manual judgment and mutual notification in risk identification, test content, QA testing, etc., and there is a risk of missing errors in high-risk requirementsThe change control strategy of assets is different, and the reviewer can choose at will, and the reviewer only makes a judgment based on coderview information, so the overall management and control ability is weak.
2.The testing process is inefficient and the test management is messy
The R&D self-test and QA testing process is mostly in manual SQL writing mode, which is inefficient and the test process is scattered, and it needs to be switched between multiple platforms, lacking one-stop testing tools
The design of test cases is mostly based on personal experience, scattered in various test reports, and there is no unified use case managementThe test reports of each business are different, and most of them are output according to the personal experience of R&D and QA, which is not readable and lacks unified test management.
So how to do the test platform?
Starting from the requirements of the whole process test management capability, we summarize the test process into the following parts:
Step 1: Test admittance, test content, smoke test.
Step 2: Test process, case generation, unit test, integration test, regression test.
Step 3: Test approval, test report, impact evaluation.
Step 4: Go online and transfer the case to monitoring.
Morse Data Test Platform 1Version 0 has been launched, which basically realizes the automation capability of the whole process of QA testing, and the current implementation effect is as follows:
Extension-based Morse Data Test 1Since the 0 version was launched in early July, it has initially had the ability to standardize the process and test the platform, and rough statistics have saved about 30% of the manpower investment of QA students in the testing process.
In the future, we will continue to focus on the stability construction of the platform, the support capabilities of real-time data testing, and the ability to generate test scenarios based on general experience, semantic analysis and contract intelligence, and develop to test intelligence in the future.