The DevOps philosophy is fast gaining traction among CIOs; a recent IDG survey found that 54% of IT executives were using DevOps in some part of their businesses. By unifying development and operations, DevOps supports a faster, more responsive style of software development, where the inherent feedback loops work to resolve issues faster and promote innovation.
Yet DevOps isn’t a product you can buy and deploy, or even a methodology you can just roll-out. It’s a combination of technologies, tools, processes and cultural change, all of which need to be in place.
Tools are, of course, important. "We think in terms of how you develop, automate, test and deliver” says Ben Scowen, xPaaS Business Lead for Capgemini Cloud Platform, “and then, for operations, what are the tools we need to run and gather metrics, so we can feed back into the process to make our software better.”
This involves a wide range of development, operations and management tools, not to mention CI/CD tools and scheduling tools designed to optimise use of IT infrastructure. Automation is a crucial component, but companies need to realise that they don’t have to do everything themselves.
“Are you going to start at the microscopic level, automate everything and start building everything from scratch” Scowen questions, “or are you going to embrace and take PaaS, for example, or SaaS?” By using as high-level services as they can, he suggests, DevOps teams can avoid some of the heavy lifting involved in getting to the point where they’re automated and delivering successfully.
Beyond the tools, it’s a matter of process. For Amanda Clay, xPaaS Product Marketing Lead for Capgemini Cloud Platform, it’s about repeatability, standardisation and a rigour around processes, bringing development and operations together.
“If they’re using the same set of tooling they’ve got the same view of things” she says, “it’s all about the discipline and the governance of IT.” This means following pipeline templates, being prepared to relinquish control to automate, and working exclusively through the CI/CD pipeline rather than tinkering manually. DevOps isn’t about trying to find fault in live systems, but building software that works and, and if it doesn’t work, fixing it. It’s about having the tools in place and then relying on them.
Most of all, making DevOps work involves a change of culture, moving from big, six-month release cycles to releases every week or even every day. It’s too easy for a DevOps implementation to be confined by outdated approval processes; not every organisation can handle the shift, which requires sizable investments without any immediate promise of ROI.
“You have to go from the top” says Scowen, “because you’ve got a nervous workforce and nervous delivery managers and you’ve got to take them on a journey.” The trick, Clay suggests, is to start small with something that supports automation. “If you start with something small, and show the value of that, then that builds your business case. Hopefully, that can evolve and spread across the business and start adding more value.” Here working with partners and using established services and methodologies can help accelerate the route to ROI. And the struggle is worth it.
“The results are amazing” says Scowen. “Our team can click a button and build a platform in hours and scale it in milliseconds.”
Want to know more?
- Download The automation advantage report that shows how cloud automation is enabling companies to accelerate their delivery of applications.
- Read the Cloud native comes of age report to find out how cloud-native applications are enabling business agility and innovation.