Removing Testing Bottlenecks in CI/CD and DevOps
Curiosity often discuss barriers to “in-sprint testing”, focusing on techniques for reliably releasing fast-changing systems. These solutions accelerate and optimise processes that prevent organisations from making incremental improvements to complex systems.
Common barriers to in-sprint testing include flawed software design, slow test case creation, repetitive test scripting and time-consuming test data provisioning. You can see how Curiosity solve these challenges in our latest infographic, and on our next webinar: In-Sprint Testing: Aligning tests and teams to rapidly changing systems.
Curiosity recently received a great question from a senior tester: How do these testing accelerators further apply to CI/CD and DevOps?
The question hits on a key point for CI/CD and DevOps: how can testing processes match the speed and automation of CI/CD and DevOps pipelines, avoiding testing bottlenecks when making incremental changes to complex systems?
This article summaries three perennial blockers to CI/CD and DevOps, discussing how solutions for “in-sprint testing” can also unlock Continuous Testing and Delivery. It concludes with a nod towards the concept of an “Open Testing Platform”, which leverages data analytics to make testing truly reactive to rapid change. This is because it makes testing proactive to rapid change.
IN-SPRINT TESTING: A KEY COMPONENT OF CI/CD AND DEVOPS
The below processes are time-consuming testing activities that are unreactive to change, and which struggle to ensure the test coverage needed to release quality software reliably. Fortunately, automated solutions are available for each. These solutions can be integrated into CI/CD pipelines and DevOps toolchains, rapidly and reliably testing fast-changing systems after incremental changes.
1. Unavailable, Incomplete, and Inaccurate Test Data
This is a big blocker for CI/CD and DevOps, as often the manual processes involved in test data provisioning clash with the automation and evolution built into CI/CD. Automated tests and test teams are frequently held up by lengthy waits for manual data provisioning. Incomplete data further undermines test coverage, while inaccurate data and data clashes lead to time-consuming test failures.
Test Data Automation is a concept (and product) that aims to make all these TDM processes reusable and reactive, integrating them into test automation frameworks, DevOps pipelines and CI/CD infrastructure. That means that tests trigger automated data “find and makes” as they are generated and run, equipping every test with the complete and consistent data.
Test Data Automation furthermore integrates processes like masking and subsetting, right-sizing data and ensuring compliance. This self-provisioning of data in turn removes one of the greatest bottlenecks in testing, and so one of the greatest barriers to CI/CD and DevOps
Curiosity’s most recent webinar, Data Breaks DevOps: Why you need automated test data in CI/CD, set out a strategy for allocating data on-the-fly within CI/CD pipelines.
2. Slow and Repetitive Test Case Creation and Scripting
Much of the same applies to test case creation and scripting. Today, both processes can be highly repetitive and time-consuming, while a lack of measurability means there is no guarantee of the coverage needed to deliver continuously and reliably.
Test Modeller auto-generates both test cases and scripts for a wide range of test case management tools and automation frameworks, using coverage algorithms to target testing where it’s needed in-sprint.
This automatically generates the smallest set of tests needed based on time and risk. Exposing the test generation to CI/CD tools and source control systems auto-generates the tests as part of CI/CD and DevOps pipelines. This replaces the time lost to repetitive and cumbersome test creation, enabling Continuous Testing and so Delivery.
You can see this automated and optimised approach to test creation live on Curiosity’s next webinar, In-Sprint Testing: Aligning tests and teams to rapidly changing systems.
3. Test Maintenance
Manual test scripting and data provisioning cause substantial bottlenecks in CI/CD and DevOps.
Manually created tests are usually brittle to change, while there is little traceability to source code or requirements. That means that there is little automation or when identifying which tests need updating or removing.
Meanwhile, data must be refreshed, repeating many of the test data bottlenecks discussed above. Test cases, scripts and data are furthermore rarely linked to one another, or to requirements, and so all need checking and updating separately to keep them aligned.
The advantage of generating tests and allocating data automatically from intuitive flowcharts is that you only need to update one single source of truth. Making a relatively quick changes to a single flowchart will ripple across all interrelated master flows and “subflows”, regenerating an up-to-date set of test cases and scripts. Test data “find and makes” similarly become reactive to changes as the models are updated.
FROM AUTOMATION IN TESTING TO TRULY CONTINUOUS TESTING
Automating test data allocation, test creation and maintenance accelerates and optimises testing processes often responsible for blockers during Continuous Delivery and DevOps. The automated processes can all furthermore be integrated with CI/CD infrastructure, rigorously and automatically testing fast-changing software with every rapid release.
This in-sprint testing becomes truly rapid and reactive when it is tied to what Curiosity call an “Open Testing Platform”.
An Open Testing Platform gathers and analyses data from across the whole application development lifecycle, seeking to expose significant changes as they occur. Exposing this data to a “model in the middle” then regenerates tests and data as changes occur:
An Open Testing Platform thereby provides an “Early Warning System” for testing, informing testers of changes in source control systems, system requirements, user behaviours, and more. It furthermore analyses changes to identify risk, prioritising the tests that should be run during that sprint. In other words, it informs testers of changes that have occurred and helps them identify the tests required to act on that information.
The last component of this “Inform, Act, Automate” paradigm is then the automation needed to run the requisite tests. This goes beyond test execution automation, encompassing the solutions described above. Having identified the tests needed to act on the latest changes, automated test generation combines with test data allocation to create the requisite in-sprint tests.
Tying this automation together with CI/CD and DevOps infrastructure begins making this process continuous, especially when all are embedded within an Open Testing Platform.
To return to the question driving this article, then: the role of automated test generation, data allocation and automated test maintenance go beyond “in-sprint testing”. They remove barriers to delivering rigorously tested software continuously, while also removing bottlenecks in DevOps pipelines.
An Open Testing Platform then makes testing truly reactive to rapid change. This is because it makes testing proactive to rapid change. You can learn more about the role of an Open Testing Platform in the latest Curiosity eBook, Building Sustainable Test Automation: Should you hire more people or buy “another testing tool”?
If you’re interested in removing test data bottlenecks within CI/CD, sign up for Curiosity’s next webinar, In-Sprint Testing: Aligning tests and teams to rapidly changing systems.
About The Author: Thomas Pryce, Communications Manager at Curiosity Software, has been in the software testing space for over 5 years.