The zEnterprise pedigree includes reliability, availability, and industrial-strength computing. Sites add to this value chain when they develop and deploy quality applications that work right the first time, with minimal chance of failing or creating situations that generate business risk. With batch windows continuing to shrink and even morph into “background batch processes” that coexist with online processing, data centers are virtually running non-stop, with little room for any downtime.
The need for continuous system processing and availability that never fails is increasingly focusing the spotlight on testing and Quality Assurance (QA).
“Industry is starting to turn a corner and realize there’s a lot more to testing than how they have been accustomed to thinking about it,” says Mike Browne, a distinguished engineer and the executive who heads up IBM’s hardware, systems, and software testing. “It used to be that IT tested an application for a few days at the component or application level, verifying that all the features and functions worked. But if you’re building a car, you don’t just put all the parts together and sell it. You make sure the engine, the transmission, the drive train, etc., all work together, because a car is a collection of systems that make something work. Cars today also have to interact with each other so there are various and expanding levels of integration that must be systematically tested.”
The growing interrelationships between software, hardware, and networks in heterogeneous computing environments warrant new thinking about integration and systems testing.
“zEnterprise has always had a high bar to reach when it comes to constant availability and applications that work right, but we have noticed recently that customers are becoming even more demanding,” says Rich Prewitt, IBM senior technical staff member in z/OS Test, Design and Engineering Architecture. “The goal now is five nines availability with no more than 15 total minutes of downtime per year for a zEnterprise.”
Still, while consciousness of new quality and performance goals is growing, sites remain mired in older testing methodologies and approaches.
“They do really well when they’re testing the ‘grooved’ path they’re accustomed to testing,” says Prewitt. “When things work well, they work really well. The bottom line is that they’re good at ensuring we haven’t broken anything in the main line of functionality.”
Prewitt says that test organizations in enterprises today are focused on system resiliency and the availability of platforms in their data centers. To improve results, testers know they have to come up with new and innovative ways to test systems. For example, when a site encounters a failure that occurs deep in the software that could have a ripple effect at other levels of system and application, they could face second and even third orders of effects for recovery.
“This is why, when you’re testing, you need to spend considerable time testing for and ensuring availability,” says Prewitt. “Some of this testing involves identifying the ‘soft spots’ of a system and deliberately injecting problems to see how the system will respond. At the end of the day, your goal is to eliminate as many potential business impacts from the system as possible.”
Why Testing Is Challenging
