Features
Automation Puts the 'A' in Agile Testing
Test automation delivers agile's promises
Mar. 9, 2009 10:30 AM
In the agile community there are many different opinions about the value of automated testing. While many argue about automation's place in agile, the business realities of geographic separation, multiple teams, and limited resources dictate a more pragmatic approach. The process of software delivery has always centered on the idea that teams must manage quality if they are to avoid operational risk. Since nothing about agile is pre-determined or dictated, it comes as no surprise that businesses are hesitant to implement it as a trusted business process. Using tools to automate a newly adopted agile process becomes yet another variable that can complicate business decisions around software delivery.
These fears need to be overcome - automated testing, when integrated into the development process in a practical manner, plays a critical role in helping companies support both new and existing processes. When investigating agile testing methods, guidelines and first-hand experience can help to ease concerns around automated testing methods in environments of all sizes. Test automation allows for increased flexibility and repeatability, delivering on agile's promise of increasing the efficiency and quality of software delivery - through it all, automation puts the ‘A' in Agile testing.
Automation & Agile Software Delivery
In the world of agile software delivery, requirements are raised, changed, and reprioritized in much shorter timescales, meaning the functionality must be delivered in less time, but must still be fully tested to ensure that it continues to meet the needs of end users and the business. In an agile process, testing is continuous and imperative, since features aren't "done" until all associated test cases pass. As software volume and complexity increases, it's clear that manual processes will fail to deliver the fast and repeatable results necessary to keep the promises of an agile proposition. Without the ability to automate and quickly repeat tests, quality will inevitably suffer as testing is marginalized to achieve on-time delivery.
Regardless of whether your development team is considering going agile, making the transition, or already operating in an agile mode, it's essential to consider how technology can support this new way of working - particularly in the area of testing. There's room for improvement in all major test processes including: test automation, test management, and requirements management.
In terms of challenges, those interested in implementing agile testing automation are facing a number, such as operating agile and traditional testing methods side-by-side, identifying tools that are usable and flexible enough for an agile approach, allowing specific non-technical and non-testing members to actively contribute to the testing process, enabling open source integration, and enabling test-driven (or model-driven) development. Overwhelming? Yes. Insurmountable? Definitely not!
Why Automated Testing for Agile Development?
Let's dive a little deeper into the reasons why testing needs to be automated in an agile delivery environment. The people involved in testing are part of the delivery team - not an isolated group that developers give the code to as a final step to release. Ideally, testers sit side-by-side with the developers, who, as they create code, pass it to testers early and continually throughout the process for evaluation against acceptance criteria. Since capability is built iteratively and the team needs to maintain velocity, the code assets have to be verified quickly. For agile to truly succeed, functional testing has to be quick, iterative, and responsive.
- The Need for Speed: Accelerating the Code-and-Test Process - Automation enables testers to create simple, reusable scripts that they can deploy to save time and increase the consistency of testing across similar user stories, story points, or requirements in and across projects. Tests can be developed from the user story to drive the functional capabilities and then run rapidly and iteratively. The speed of automation significantly lightens the workload of testers and eliminates the need for late night and weekend testing marathons that can burn teams out.
- The Need for Repeatability: Doing the Same Tests and Scripting Tests against the Right Acceptance Criteria - With agile development, regression testing should be done at the close of every new iteration - in some cases this means daily. Regression testing requires that 1) you do the same tests every time you test a particular piece of code and 2) that the test is scripted against the acceptance criteria of each respective user story. Whenever code changes (or is extended to include new capability), you need to rerun all functional tests for all user stories up to the latest change to ensure other user stories weren't impacted inadvertently.
Repeatability is nearly impossible to achieve with manual testing due to human error, variability, and inconsistency. People simply can't remember exactly which tests they ran for each piece of code for the last iterative cycle - and even one oversight can cause problems in the final code. Not to mention the fact that with large code bases the time needed to manually test usually exceeds the time allotted for the iteration. But with automated, repeatable functional and regression testing, one can execute tests consistently whenever necessary.
The automation element adds key benefits that can't be achieved with manual testing. For example, it can greatly accelerate the code-and-test process by supporting fast automated test scripts. Automation also ensures the repeatability of tests to maintain regression testing from sprint-to-sprint, iteration-to-iteration. It can also enhance test efficiency with robust yet flexible test management processes, helping customers avoid the inherent inaccuracies that manual processes inject in to the process.
Choosing the Right Test Automation Tools
Agile software delivery relies on the right mix of people, processes, and tools. The choice of testing tools becomes critical for reaping the efficiency and quality benefits of agile. One side of the coin shows that the manual testing processes popular with most businesses today can make it difficult - if not impossible - to implement and scale agile delivery methods. On the other side of the coin are those automated testing tools designed solely for use in traditional environments that force testers to work within a certain development practice to ensure test accuracy. With such choices, it's no wonder many companies fear adopting automated testing tools!
Time to trade in that outdated "coin" for real change: an agile delivery process requires a shift to truly agile testing tools. When researching tools for implementing automated testing, watch out for those that can only be initiated at defined points in the development process, require specific QA knowledge, or require that specific development practices be used to guarantee precision. Automated testing puts the ‘A' in agile, but only if the tools can deliver both flexibility and accuracy.
Conclusion
As an increasing number of enterprises embrace agile software delivery approaches they realize a new set of challenges. The realities of the enterprise - geographic separation of teams, cost reductions, diminishing timelines, and limited resources - require that organizations take a practical approach to agile software delivery. Modern automated testing solutions play a vital role in supporting this evolution by enabling agile teams to implement test strategies that are flexible, repeatable, and accurate. In the end, businesses can reduce the risk of releasing applications of unknown quality or reliability with agile test automation.
About Stephen WaltersStephen Walters is product marketing manager at Borland, bringing Borland’s LQM solutions and products to market. He has been working within the IT industry for 16 years. Stephen started out as a programmer of Pseudo computer systems, followed by programming using Basic, Pascal and C for three tier client server systems. He then spent some time working in a testing arena before his first work in Configuration Management 13 years ago.