DevOps has become a cornerstone of the IT department in organizations wishing to speed up software development and deliver new applications, faster. Now that the cloud is ubiquitous, this once-niche field is transforming how many businesses operate and bringing the days of clunky, cumbersome IT rollouts to an end.
DevOps will continue to evolve in the coming years as companies identify new IT challenges to be solved, particularly as they look to automation to optimize or even remove processes that stand in the way of business agility. That's according to a new report from tech analyst firm Forrester, which says tech leaders need to prepare for "big-picture changes" as DevOps changes the face of business innovation and transformation.
The principles of DevOps are based on the concept of cohesion and joining up people and processes for better outcomes. As such, companies need to break down silos and invest in training if they hope to create successful cross-functional product teams.
Forester believes that DevOps teams will redefine themselves according to their purpose, rather than their function. For example, feature (business/end-user aligned), enabling (coaching and internal consulting), platform (servicing the developers on the feature teams), and complicated subsystem (mainframe or speciality hardware). This makes it easier for teams to understand their role and function, which can be important for addressing design challenges.
Engineers will also become a shared resource, as opposed to being assigned to individual teams, Forrester predicts, particularly as IT teams ditch specialist infrastructure teams in favour of platform teams with a broader range of software-engineering tooling, configuring, and monitoring responsibilities. This means DevOps teams will get better tools, too.
Practices will overtake processes
Achieving goals or outcomes within a business typically involves following a series of steps in a particular order. This makes it easy for delays to be introduced, as each step often requires a person or department to sign it off. Handovers can be clumsy, and in the process, goals can become muddled or misaligned.
Forrester predicts that DevOps will help businesses shift their focus from process to practice in the future, which will be centred around agreed consensus on organizational priorities and how things are done within the company. These will be "optimized for speed, with greater automation around governance, compliance, security, and standard operations" and minimal human interaction.
Risk management, for example, remains a lengthy and often burdensome process, even if it does justify more manual oversight than some other IT processes. In the future, Forrester predicts that companies will be able to leverage machine learning to analyse findings from the development process to provide automated risk-change analysis, significantly speeding up the software release and deployment process.
Platforms will consolidate, extend and deepen
DevOps requires a rich set of technologies that can integrate and operate harmoniously. As businesses' requirements evolve, so too will the requirements of the DevOps market and the tools required to support business transformation.
Forrester believes that organizations' current pick-and-mix approach to DevOps tools and tech stacks will give way to a more unified approach, whereby each team uses the same end-to-end integrated software delivery platforms (ISDPs). Likewise, organizations will begin looking at ways to simplify development for tightly integrated enterprise apps, which don't tend to play nicely with the type of custom software that can speed up app development and delivery.
A couple of proposed solutions here include low-code/ no-code tools, which can allow even non-tech employees to create business apps, as well as machine learning and automation. Forester notes that challenges stand in the way here: low-code platforms aren't well integrated with DevOps pipelines currently, meanwhile testing out machine learning on critical business applications can be risky – something MLOps (machine learning plus operations) aims to solve.