Production Data vs. Synthetic Data: Which one to prefer for Software Testing?
- February 12, 2021
- Ray Parker
There is an increasing interest among QA professionals to use synthetic data generation for software testing. This increase is triggered by a requirement for data privacy or to meet the needs accelerated for agile and DevOps environments. However, one of the topmost reasons to use synthetic data is to have complete control over the variety of data required for maximizing test coverage.
While test data has been identified as a vulnerability for businesses that should adhere to privacy laws like HIPAA and GDPR that are designed to prevent exposure of sensitive information. Organizations need to practice meaningful change to accelerate the speed and accuracy of test data provisioning. For this purpose, an offshore testing company may prefer synthetic data, which can overcome the threat of exposing sensitive customer information.
Synthetic test data provisioning has become crucial in software testing in order to achieve success with the help of AI and new test automation technologies. As a result, test data can be the building block for organizations that are implementing continuous integration (CI) and continuous delivery (CD).
Challenges in the QA Process
So how can QA departments simultaneously maximize the speed, quality, and privacy of test data while reducing the cost and the complexity that comes with it?
Organizations are in dire need to address the challenges of keeping up with the speed of development as QA teams strive to achieve quality code and data privacy. Synthetic test data is preferable instead of masking production test data.
What is Production Test Data?
Production test data is a copy of a production (real-time) database that has been masked to represent data that is relevant to a test case. Production test data is accompanied by a test data management (TDM) system to prepare, control and use this data. Commercial TDM systems are expensive, so many organizations choose to develop their own processes tailored to their needs.
What is Synthetic Test Data?
Synthetic test data does not include any actual data from the production database. It is artificial data that is generated by a synthetic test data generation engine. Synthetic test data generation eliminates the need for data masking, as test data can be generated on-demand and without compromising sensitive customer information. Thus, teams can utilize synthetic test data using a self-service model.
Test Data Criteria
There are six factors often used to make a choice between the use of production and synthetic test data. Each factor is essential to eliminate the test data bottlenecks and to avoid the risk of a data security breach. Now let’s have a look at these essential test data criterion that can help differentiate between the two:
QA managers need to consider the time requirements for test data provisioning before beginning a testing project. Typically, it takes a few days to fulfill a request for test data to support a certain test environment. But what if this time could be substantially reduced from days to minutes? Synthetic test data simulates the real-world data and can be generated at a rate of thousands of rows per second. So synthetic test data generation eliminates the bottleneck of requesting production data from the team and also removes the need to mask the data. This model allows testers to provide their own data whenever they need it and discard it when they have completed their testing.
Cost is an important factor to consider when it comes to creating, managing, and archiving test data. Since production data needs to be prepared, managed, and stored, teams, need a TDM system. So they need to purchase a TDM system and bear its maintenance cost too. However, if synthetic test data is generated on demand, and there are more cost-effective solutions/tools available now than they were a few years ago that can significantly lower the cost of providing test data.
When provisioning production test data, testers have little control over the quality of data with respect to the factors like age, accuracy, variety, and value of data that they need to copy, mask, and subset. Software testing requires different permutations of data with negative test data. Testers may be forced to manually modify the production data into usable values for tests. But synthetic test data removes the effort that goes into creating a data subset. It is generated on a test data scenario and is able to quickly generate data with a complexity that is not possible to be performed manually.
QA teams also need to consider the privacy implications of the sources of test data. Test data provisioning should remove all PII, to avoid the high costs of a data breach. Production data requires data masking, but no masking process is foolproof. However, synthetic test data ensure 100% compliance with all security regulations throughout the testing cycle.
When choosing a source of provisioning of test data, QA managers should ensure it is easy for the testers to get the data they need for their tests. It should be a simple model that makes quality test data available to anyone at any time. Synthetic test data generation makes the process simple with platforms that allow real-time test data to be created on-demand by the QA team.
Test data should be versatile enough to be used by any testing tool or technology. The test data provisioning process should be adaptable to any testing environment, of any size, for any sector. It should be capable of working with large databases with different applications. Synthetic data is known for its versatility and can cater to many large databases on demand.
QA professionals are still concerned about the trade-offs. They still need to figure which approach is better and what is the right choice for their testing environments. These concerns set the stage for a great debate about whether to use production test data or synthetic test data in continuous testing environments. The above-mentioned differences can help QA teams working for an offshore testing company to make a better choice.