DEVOPSdigest

Vanson Bourne recently conducted a survey of organizations reliant on the mainframe with a surprising finding: deployment frequency for new mainframe features is expected to increase by 41%, compared to 38% for non-mainframe applications.

Why would deployment frequency on the mainframe be higher than other platforms?

Customer-facing, distributed, web and cloud-based apps all continue to be highly reliant on the mainframe. A single back-end mainframe system could be supporting multiple front-end systems. Because the rate of continuous improvement of systems of engagement must be at least as fast as the rate of improvement of systems of record, the urgency of mainframe development work is increasing. And, organization must deliver these changes faster without compromising quality.

But there's something slowing them down — testing. Testing is critical to producing high-quality, high-functioning software. However, it remains a development function where manual, error-prone methods are threatening the quality, velocity and efficiency of mainframe software roll-outs. Across platform types, the need to automate testing is vital and the current state is a real problem, as evidenced by other Vanson Bourne survey findings:

■ 90% of application development managers say that automating more test cases could be the single most important factor in their success as the pressure on IT to accelerate innovation increases.

■ 80% of application development leaders say it is inevitable that unless they can automate more test cases, bad code will make its way into production.

Manual testing is a bottleneck on all platforms, though especially the transactional mainframe which is largely responsible for making applications "work." Consider the following:

■ 77% of application development managers find it increasingly difficult to simultaneously increase quality, velocity and efficiency to meet business innovation goals when developing and testing mainframe application code.

■ Only 7% of organizations automate the execution of test cases on mainframe code to support their ability to accelerate innovation.

■ 53% of application development managers say the time required to conduct thorough testing is the biggest barrier to integrating the mainframe into Agile and DevOps.

Clearly there's a disconnect between the perceived importance of automated testing on the mainframe, and the number of organizations actually doing it — and the disconnect is arguably more dangerous for the mainframe than other platforms for several reasons. Many of the biggest and most successful organizations in the world rely on mainframes, and as more mainframe experts retire, there is a potential skills shortage and loss of customer-specific application experience looming. This means there will be fewer and less experienced people available to develop and execute tests; it also means that slow, cumbersome, manual testing processes may continue to be a real deterrent as organizations vie to attract newer skilled developers to mainframe careers.

Workloads on the mainframe are also expected to increase substantially — according to BMC's most recent State of the Mainframe survey, 93% of respondents believe in the combined long-term and new workload strength of the platform. This means unless something changes with the current testing status quo, the chances for bad mainframe code to make its way into production will only increase, thus potentially wreaking havoc.

Ultimately, Vanson Bourne's survey shows that current manual approaches to testing are not working. The lack of automated testing, particularly on the mainframe, is a primary area for concern given the potential for exacerbating the mainframe skills shortage and causing mission-critical transactions to break. More automated testing on the mainframe is the key to avoiding these problems, not just for mainframe applications, but for the end-to-end transactional applications they support.