Test plan・Test environment and environment preparation

26/05/202120/10/2021Test qualityy

The test plan also describes the environment for testing and its preparation.

When making a software test plan , the content of the test and the personnel plan of the test worker who performs the test come to mind first, but just as important as planning the test environment . To test software, you need to prepare a lot of test equipment and use it . At the time of the test plan, it is necessary to carefully consider the preparation of the test environment, such as how much equipment will be prepared by what time and where to arrange the equipment , and write it in the test plan. In this article, we’ll take a closer look at what equipment you need for your test environment.

How many target hardware to run the software will be used from when

The target hardware for executing the software to be tested is the most important test equipment, without which the test will not start. However, when developing a new product for embedded devices, preparing the target hardware is the most difficult task in preparing the test environment. Since it is a new product, it is often the case that hardware is also newly developed at the same time, but in that case , surely incorporate in the hardware development plan how many target hardware is required at what time for software testing. Let’s get it.

How many target hardware at what time to perform various tests such as a test for checking the function, an overload aging test for checking the stability, and a large-scale test for checking the operation at the maximum configuration. Based on each test plan, identify the required number of units and the required period , make a plan for arranging the target hardware, and allow some margin for those units and timing, and develop the hardware. It is good to have it incorporated into the plan.

Software testing is based on the premise that there is product-level quality target hardware that operates stably . However, when developing hardware at the same time, it is often difficult to have a stable number of target hardware. In such a case, it may be possible to make up for the shortage of target hardware by sharing a small number of target hardware with multiple testers . In that case, the person in charge of the test must be divided into three teams, and each team must use the target hardware for eight hours a day, which may require a night shift system for the test . To prevent this from happening, allow some margin in planning the number of target hardware and when to obtain it.

Which version of the software to be tested will be passed to the test team at what time

Software testing doesn’t start without the target hardware, but there’s actually something more important than that, it 's the software you’re testing . To add a little more words, it is the software to be tested that has a quality that is worth testing .

Unfortunately, software design and implementation rarely goes according to the original plan. Usually, in the course of development to add or change the specifications Toka, the occurrence of technical problems at Toka, design and implementation of the process is delayed is often. In that case, the day when the test is scheduled to start in the test plan has come, but the software to be tested has some basic functions that have finally started to work, and some functions have not yet worked. It can happen. Of course, it’s not a decent test.

The test team cannot take measures against the delay in the design / implementation process, but in order to reduce the impact on the test process as much as possible, design / implement the software at the stage such as the α version before the test version software. obtain from the team, which part function if test whether the start is likely to, such as the re-adjustment of test dates that the, because should be incorporated into the test plan is also to, is one of how to respond. For that purpose, consult in advance about what version of software will be handed over from the design / implementation team to the test team, and write it down in the test plan.

Clearly the version of the server connected to the upper level and the terminal connected to the lower level

Some of the products under test can be used alone, but recently internet connectivity has become a natural feature. When testing a product, it is often connected to a higher-level server or a lower-level peripheral device or terminal to run the product under test .

For the upper server, peripheral devices, and terminals to be prepared as a test environment , clarify the hardware revision and the installed software version . In the case of a problem caused by cooperation with a server or device connected to the upper or lower level, it may not be reproduced if the hardware revision or software version is different . Reproducibility is important for investigating defects, so to ensure reproducibility, clarify the hardware revision and software version in the plan, or remember to record it during the actual test. Let’s go.

Is the communication path connecting the upper server and the lower terminal only a clean one?

In the case of the connection device on the premise of the network is, the level of the server or to connect in any communication path is important as also a test environment. If you want to check the simple function, you can use a clean communication path , but if you want to check the stability, you should consider preparing a communication line that is dirty to some extent like the production environment. prize.

A clean communication path is one in which transmission errors and communication delays hardly occur . If you build a test environment in one room and the connection cable between the test target and the server is short, the communication path will not be affected by noise and the signal will not be attenuated in the communication path, so it is very beautiful. It becomes a communication path.

However, the communication path actually used is quite dirty because the signal is attenuated by a long cable and the signal is interrupted for a moment due to noise caused by electromagnetic waves in the middle of the communication path . Of course, since it is a digital signal, recovery by error detection, error correction, and retransmission works at various layers of the protocol, so the communication data itself will not be lost.

However, digital signals that have passed through dirty communication channels will suffer from reduced responsiveness, delayed arrival, and short-term packet loss as a result of these error corrections and recovery . As a result, some error handling is executed even in the software under test.

Dirty similar to the communication channel of such reality communication channel was whether to prepare as a test environment, the test subject of soft confirmation of stability changes in the test design of whether over how much of the cost for to confirm I will come. Therefore, at the test planning stage, it is good to consider whether to prepare a dirty communication path while considering the test environment.

The same can be said for cables that connect lower peripherals and terminals . If you connect it with an expensive and short cable, it will be a beautiful communication path, but if you connect it with an inexpensive long cable and run a vacuum cleaner near the cable, you can make a fairly dirty communication path. Again, it’s a good idea to consider whether to have a dirty channel at the test planning stage.

Do you use auxiliary devices or software that generate errors for communication partners?

If the device under test operates by connecting to a higher-level server or lower-level peripheral device or terminal, a server, peripheral device or terminal is also required. It is also necessary to prepare these real things, but in order to test the quasi-normal system and abnormal system with the cooperation function with the upper server and lower peripheral devices and terminals, various from the server, peripheral devices and terminals You will need some kind of error response .

If you want to generate the required error response at the required timing, it is quite difficult in the real world and you will need some kind of simulator device or software for simulation . If it is a general communication protocol, you can use it because a commercially available test device has a function to generate an error packet, and if it is a special communication protocol, you may need to make your own simulation software for the test environment. not. In any case, consider the need for these simulation devices and simulator software at the test planning stage.

What to do with the location, power supply and office for testing work

As an environment used for software testing, if the number of target hardware and the upper server and lower peripheral devices and terminals are specifically identified, a wide range of test work locations will be required to install them and perform test work. You can roughly understand. At the same time, you can see how much power capacity you will need.

In particular, if you want to perform a maximum configuration test as a large-scale test , you need a place to install the maximum configuration equipment and a power supply capacity to operate it at the same time, but if you do not carefully consider it at the planning stage, the location What happens when the breaker trips when I start testing for nothing.

It is also important to record the test results in the test work and start tracking any defects found. For that purpose, the office work place for organizing the test work results and inputting them into the system is also required according to the number of testers . Let’s consider these at the stage of test planning.

How many test personnel do you need to carry out the test?

This doesn’t need to be explained again. The test design designers and, carry out the test work test the person in charge and, for the test administrators responsible for the management of the design and test, what kind of experience and testing with the skill personnel is what people need or that of, necessary from the test plan that you created Estimate the number of people and make a personnel plan. Whether or not we can secure personnel according to the personnel plan we have made is another matter. .. ..

Once you have a test plan, the next step is to manage your test results.

In this article, I have introduced what you should write in the test plan based on what Father Gutara has experienced and seen. Once you have a test plan, it’s time to start testing. The important thing here is the management of defects found in the test, so let’s introduce the management of defects in the next article.

Next : Test plan・Decide how to classify defects and bugs and the items to be recorded

Back : Test plan・Test metrics (Part 2): Test result management

Return to :  Testing is the only way to guarantee software quality