By Anupama Rathi, Associate Vice President, Head of DevOps COE, Infosys

DevOps has and continues to evolve, with organizations racing to implement new capabilities for competitive advantage and to respond to today's unique business challenges. Benefits, such as quality, reliability, time-to-market improvement, and cost savings have become mainstream in 2023, with continuous integration and continuous delivery (CICD) establishing itself as the new way of working. In fact, it's no wonder that 45% of Agile leaders, in a 2022 Gartner study, said that they consistently use DevOps across their organization.

Now, we've entered a new era of DevOps, with organizations moving towards mature adoption and investing to scale at the enterprise level. This shift has required organizations to first reflect on how DevOps was implemented in the past and assess for any necessary course correction. Leaders must make certain that CICD adoption has met the goals of the organization and derived all the benefits previously being pursued. From here, new KPIs and metrics need to be established so that DevOps success can continuously be measured.

Challenges in scaling DevOps and the need for continuous evaluation

To accurately assess past success, organizations must look back at how DevOps began. At its start, DevOps came about organically through the experimentation of application CICD automation. Organizations wanted to implement CICD automation across their applications, which inspired them to move onto other applications and then whole portfolios. This success inspired widespread DevOps adoption.

Despite changing the IT world, the organic success of DevOps implementation did create various challenges. When we assessed multiple organizations across various domains that claimed to have mature DevOps adoption, we found two key problems. First, some applications achieved quality and speed benefits, but not all. Second, DevOps implementation and maintenance exceeded the initial planned cost.

Infosys observed these challenges at a large automobile giant in North America that had federated and siloed DevOps implementations resulting in rapid increase in CICD maintenance costs. Inconsistent CICD implementations led to higher rework, longer cycle time and impacted reliability of applications.

We can trace these problems back to ‘federated DevOps' implementation due to organic DevOps adoption. Currently, thousands of CICD pipelines exist in an organization, with sometimes even one or more pipelines per application or system. This results in duplicate human efforts, as well as an increase in hardware and software consumption, to implement CICD for every application. Looking towards the future, we will see huge maintenance efforts and costs for these overwhelming amount of DevOps pipelines.

Federated DevOps requires each application to implement its CICD, making it harder to evaluate if these applications are adopting DevOps correctly at all desired automation stages. This prevents organizations from deriving all the benefits of DevOps, since some applications might bypass CICD stages at their convenience. Fewer applications use no-touch automation for better quality, speed and reliability, which has led to a non-standard DevOps adoption, hence why investments are falling short of planned benefits.

Getting DevOps back on track

To overcome these challenges, organizations must implement a centrally defined DevOps strategy with policies and practices that all DevOps CICD pipelines must implement mandatorily.  Minimizing the duplication of DevOps pipelines reduces the effort of maintenance, and any organization-level change required in DevOps strategy or tooling can be done centrally as a streamlined effort. Additionally, duplicate software and hardware investments are no longer needed, cutting costs and saving time.

Implementing easily customizable DevOps solutions will help navigate the complexity of the variety of technologies, deployment patterns and combinations of legacy and new-age systems that make up an enterprise. For example, template-based pipeline solutions can be implemented to support multiple technology stacks and infra-as-code modules to facilitate the choice of cloud as needed. As enterprises mature, customizable centralized solutions provide the flexibility required to adopt varied application-level situations with the continual application shift to higher DevOps maturity.

For example, at a North American automobile giant, Infosys built an enterprise DevOps platform integrating 40+ build and deployment tools across 10+ technologies, with one-click CI/CD automation for 150+ application teams. The dynamically scalable cloud-agnostic solution improved release frequency by 75% with zero downtime and resulted in a 10M+ USD cost saving.

With the continued evolution of DevOps, organizations must make sure they're getting the most out of their investments and reaping all the desired benefits. In summary, DevOps practices enforced via a central reusable solution framework will help comply with policies and practices to ensure applications' high quality, security, speed and reliability. This ensures that more applications adopt DevOps in their true sense, leading to full return on DevOps investment.

##

ABOUT THE AUTHOR

Anupama Rathi, Associate Vice President, Head of DevOps COE, Infosys Limited

Anupama has nearly three decades of IT experience and leads the DevOps practice for Infosys. She has helped many Infosys customers scale their DevOps adoption to higher maturity.