DevOps models deliver significant benefits
Moving to adopt DevOps methodologies can deliver significant and measurable results in as little as six weeks, says Shridhar Mittal, GM of the CA Application Delivery Business Unit.
"DevOps, as a methodology, is rapidly gaining traction globally, because enterprises can simply no longer afford to take months to get applications to market," he says.
"As enterprises become increasingly dependent on software and applications to drive competitive advantage, they are under pressure to deliver new applications faster. But amid the need for speed, quality often suffers."
Mittal explains that developers are challenged by the demand to deliver quality applications fast in a highly complex environment, which may include a distributed, heterogeneous architecture and distributed teams collaborating globally.
"Theoretically, you might want to get the new code to market the same day. But compliance and quality control processes may slow release to the point that it can take six to 12 months to release. This situation is untenable, and has informed the need for new methodologies. Effective DevOps models can reduce this time to market to weeks - or even days," he says.
However DevOps is neither a system to acquire nor a big bang approach, Mittal says, adding that moving to DevOps methodologies can be done in 'consumable chunks,' each of which is driven by value.
"We have developed a value release model in which you grab the low-hanging fruit to see immediate value. Our work with partners and customers, in addition to our own research, has resulted in a model through which enterprises can move towards a full DevOps methodology in a phased manner."
The first step tackles the delays and complexity that occurs during development, he says. "Our 'shift left' model uses service virtualisation to support improved engineering and testing much earlier in the development process, so minimising time spent in application testing. This can speed up time to market by about 30% and reduce bugs by up to 60%-80%," he says.
The next step entails intelligent, model-based release automation, which also addresses the need for speed - with a 30-40% improvement in time to market, as well as improving quality.
Third is the automation of the testing itself. "If developers are doing a lot of manual, performance and integration testing, they need to look to continuous validation for improved quality and time to market. In these complex environments, testing is no longer an event," says Mittal.
Finally, he says, is a feedback layer in which developers learn from production and bring a production-like environment to play earlier in the process.
Mittal will dive down into the processes, challenges and benefits of CA's DevOps solutions during the upcoming CA IT Management Symposium Africa 2014. For more information about this event, click here.