Atos recently briefed NelsonHall on M-Frame, IP the company has recently launched focusing on model-based testing (MBT) and integrating it with test data creation, performance, and security testing. M-Frame looks to remove several inhibitors that exist around MBT: complexity of the different MBT software tools, and lack of a business case for building and maintaining the UML diagrams used for MBT activities.

MBT: A Promising Concept….

When the concept of MBT was introduced over a decade ago, it promised to transform the software testing services industry by automating the full testing lifecycle, from creation of test scenarios (based on UML diagrams) to test scripting and execution, avoiding the creation of manual test cases. Manual test cases are easy to write but can be ambiguous and are an extra step between requirements collection and test case scripting.

However, the adoption of model-based testing has been relatively limited.

In addition to the inhibitors mentioned above, the creation of UML diagrams requires organizations to invest in a new set of testing-related , and these, of course, need to be maintained. The business case of MBT thus relies on two variables: test cases that do not change dramatically release after release, and the elimination of manual testing.

The industry has not been inactive in attempting to remove these issues: several services vendors have invested in creating diagrams based on business process standards, typically focusing on specific industries (e.g. the telecom service provider sector with its eTom standards), and applications (e.g. core banking).

And yet MBT adoption has been limited.

Atos’ M-Frame Value Proposition

In response, development teams at M-Frame have focused on web applications and on automating the full test process; reducing training time for learning how to use MBT software tools; and using open source software, along with proprietary Atos’ accelerators, including its Atos AI Test Engine.

Schematically, M-Frame includes the following steps

• Crawling of a web site to identify its structure, hyperlinks, and call outsourcing (using yWorks’ yEd software product and Microsoft Excel). This step provides a state transition model used for the creation of diagrams
• Generating path/test cases, using GraphWalker open source software. Output is in Microsoft Excel
• The creation of manual test cases (using Atos AI Test Engine=, or automation into Selenium
• If automated: the next step is inputting of test data, using manual intervention, using Atos’ synthetic test data (through repositories or creation based on AI Test Engine)
  - At this stage, M-Frame also offers a statistical pairwise approach to reduce the number of test cases, based on test data combination analysis (using Microsoft PICT)
  - M-Frame proceeds to load capability determination to calculate the number of servers required to execute Selenium, based on the number of test cases
• Execution of test cases is then provided directly into Selenium (or Sikuli), or through the creation of Selenium test scripts (using Steady States Queueing techniques), which are then used for execution
  - This step is complemented by a backend systems test using Fiddler.

Atos estimates that overall, the above steps require one day of work, of which half is spent on test data entry. Implementation time for M-Frame is about one day. Training takes two to three days.

M-Frame, was released in early 2017; to date Atos has two organizations using M-Frame. Atos provides M-Frame as part of its services and does not sell it as a standalone tool

Atos feels that with M-Frame it can finally change the dynamics of adoption in MBT. The company is continuing to invest in M-Frame and its investment roadmap for M-Frame includes:

• Technical enhancement including iframe tag recognition, and further support of mobile drivers
• Integration with HPE ALM for test management
• Development of predictive models based on defect reports
• Combination of Atos’ own MBT tool, TEMPPO, with M-Frame, focusing on data.

Removing Investment in Artifact Creation and Maintenance

So, will M-Frame change adoption of MBT, as Atos predicts?

It can potentially shorten test activities dramatically. Also, M-Frame should decrease diagram creation and maintenance effort and cost very significantly. Another positive feature is that the platform is built on open source software, enabling Atos to offer it to clients with no license costs.

We will be monitoring the adoption of M-Frame to see how and in what conditions the tool has most impact.