{"id":242040,"date":"2023-08-22T13:25:25","date_gmt":"2023-08-22T11:25:25","guid":{"rendered":"https:\/\/soflab.pl\/uncategorized-kategoria-do-usuniecia-en\/manual-or-automated-generation-of-the-test-data\/"},"modified":"2024-08-05T11:26:50","modified_gmt":"2024-08-05T09:26:50","slug":"manual-or-automated-generation-of-the-test-data","status":"publish","type":"post","link":"https:\/\/soflab.pl\/en\/quality-management-en\/manual-or-automated-generation-of-the-test-data\/","title":{"rendered":"Manual or automated generation of the test data?"},"content":{"rendered":"

[et_pb_section fb_built=”1″ _builder_version=”4.21.0″ _module_preset=”default” custom_margin=”-108px|||||” da_disable_devices=”off|off|off” global_colors_info=”{}” theme_builder_area=”post_content” da_is_popup=”off” da_exit_intent=”off” da_has_close=”on” da_alt_close=”off” da_dark_close=”off” da_not_modal=”on” da_is_singular=”off” da_with_loader=”off” da_has_shadow=”on”][et_pb_row _builder_version=”4.26.1″ _module_preset=”default” custom_padding=”0px|||||” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_column type=”4_4″ _builder_version=”4.21.0″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”][et_pb_text _builder_version=”4.26.1″ _module_preset=”default” global_colors_info=”{}” theme_builder_area=”post_content”]<\/p>\n

How to optimally generate test data and is it better to do it manually or automatically? The most common answer is \u201cit depends\u201d. So let’s consider what affects the choice of method for generating test data.<\/p>\n

Methods of generating test data<\/h2>\n

Certainly, the main factor determining the easiest ways of obtaining the needed test data is the type of data we\u00a0need. Simple data, such as\u00a0a\u00a0login or\u00a0password, may be\u00a0generated manually and it is a\u00a0relatively simple method of providing the variables required for the application testing. However, manual generation of data may be\u00a0laborious and error-prone, it may be\u00a0quite difficult to\u00a0insert the PESEL or\u00a0IBAN numbers without errors. The tool we\u00a0can use in\u00a0such cases are simple data generators, which can \u2018make up\u2019 the whole sets of e.g. personal data, starting from the name, to\u00a0PESEL or\u00a0bank account numbers, which meet the requirements of the algorithms used for their verification. The situation gets more complex when we\u00a0need to\u00a0store the inserted data for potential future use. One example that comes to\u00a0my\u00a0mind may be\u00a0the customer data file. We\u00a0provide the name, login or\u00a0other data necessary to\u00a0create the account. Such data, stored in\u00a0the application database, may be\u00a0re-used in\u00a0later operations. Thus, by\u00a0using simple data to\u00a0create the customer file and storing it in\u00a0the system, we\u00a0obtain a\u00a0customer data set. Very often such sets are given individual IDs, used for identification in\u00a0other systems. We\u00a0can complicate the situation even more, by\u00a0adding various products or\u00a0services to\u00a0individual customers. These products and services will also have their IDs and will remain in\u00a0specific relation to\u00a0other modules or\u00a0systems. This way we\u00a0build quite extensive sets of data. Multiple manual generation of such sets, containing the customers\u2019 data and services is certainly both time-consuming and monotonous. Therefore, in\u00a0the cases when we\u00a0need a\u00a0lot of varied data, it may be\u00a0generated automatically or\u00a0fed from a\u00a0database of readily available data sets.<\/p>\n

<\/blockquote>\n

Automated data generation tools save time<\/o:p><\/h2>\n

Developing an automated data-generation tools saves both time and work needed for manual generation of the needed data, however it requires certain effort related to\u00a0the development itself and maintenance of such tool. An alternative solution may be\u00a0feeding the data into our system from a\u00a0ready-made database of customers and services. The solution offering the widest spectrum of available data is import of the production data. The data obtained in\u00a0this way will reflect the starting point of our application in\u00a0full, which, in\u00a0turn, will ensure full coverage of the required test cases. When anonymized, such data sets are the best type we\u00a0can obtain. But what if we\u00a0do\u00a0not have access to\u00a0the production database which we\u00a0could use, because such database does not exist or\u00a0is unavailable? In\u00a0such cases, if we\u00a0know the database requirements , we\u00a0can build the database using adequate tools which can supplement the database of our system by\u00a0the required data sets. This will ensure the optimum way to\u00a0build a\u00a0potentially large database for testing, without the need to\u00a0anonymize it.<\/p>\n

 <\/p>\n

Data generation method depends on\u00a0the amount of required data<\/o:p><\/h2>\n

We\u00a0have not answered the question about the manual or\u00a0automatic data generation yet, but we\u00a0have already got a\u00a0basis for this answer. First of all, it depends on\u00a0the amount of needed data. Manual generation of a\u00a0small amount of even the most complex data will always be\u00a0more economical in\u00a0terms of workload and cost, than developing automated scripts to\u00a0generate the data or\u00a0building the system database. This way we\u00a0get the first answer \u2013 the method depends on\u00a0the amount of data; if we\u00a0need little data, it is more reasonable to\u00a0generate it manually.<\/p>\n

What if we\u00a0need large amounts a\u00a0data? We\u00a0have two options here. One of them is to\u00a0develop an automated script to\u00a0generate the synthetic data, the other \u2013 to\u00a0feed the required data sets into our application. Generating the data by\u00a0the automated scripts is slower than feeding it from the databases \u2013 often it is a\u00a0simulation of the user\u2019s action. It is also more susceptible to\u00a0the changes in\u00a0application and requires more maintenance, as\u00a0the changes in\u00a0the application code are more frequent than in\u00a0the data structure. Automated-script data generation has also one big advantage over the data feed. When in\u00a0our test environment the relations between individual data are specific and unique, the feed of the external data will not always be\u00a0possible, as\u00a0it may interfere with the continuity of these relations. Automated data generation, through the simulation of user\u2019s actions, resolves this problem, as\u00a0in\u00a0such cases we\u00a0generate sequential data in\u00a0a\u00a0given environment, from start to\u00a0finish, maintaining the continuity of relations.<\/p>\n

Now we\u00a0can answer, when to\u00a0generate automated data.
<\/strong>We\u00a0do\u00a0it when we\u00a0need a\u00a0lot of data in\u00a0an environment which requires data continuity. This does not mean than feeding large amounts of data is never useful. In\u00a0the situations when we\u00a0can ensure re-usability of the fed data, we\u00a0should consider such option. Feeding a\u00a0database will always be\u00a0faster and offer more variety than development of automated tools. Thus, we\u00a0can say that firstly we\u00a0should consider the feed of the test data. Only if this solution does not guarantee the required continuity of data relations, we\u00a0should move towards the automated scripts.<\/p>\n

 <\/p>\n

Soflab G.A.L.L. enables fast data feed and maintaining the relations between the test environment data<\/o:p><\/h2>\n

There is also a\u00a0solution which combines the benefits of data feed with the possibility to\u00a0maintain the relations between the data from the databases and our test environment \u2013 offered by\u00a0the G.A.L.L. application.<\/a><\/strong> It can take a\u00a0representative sample of our test environment as\u00a0a\u00a0model and feed our database with the needed test data maintaining its variety. The solution ensures an efficient process for generating any amount of reliable test data, improving the efficiency of test processes without compromising the security of real data.<\/p>\n