We are a decade into the DevOps movement which began with the radical goal of breaking down the gap between the ‘Dev’ and ‘Ops’ silos in an organization. It’s worth stepping back and evaluating whether or not the movement has achieved those goals, and what obstacles still remain, ten years later.
As DevOps matured and gained traction in large organizations, while it may have succeeded in facilitating better communication between Dev and Ops, its definition expanded to include breaking down all silos in the organization, not just those between Dev and Ops. Since most organizations are broken into multiple silos, both functional and organizational, the single-minded focus on getting Dev and Ops to communicate and collaborate more effectively is no longer sufficient. It’s become increasingly clear that the issues weren’t just with Dev and Ops, but that all stakeholders who owned various functional areas of the application delivery pipeline needed to communicate and collaborate better. Analysts, architects, developers, testers, deployment engineers, QA specialists, DBAs (yes, DBAs too), and of course, IT operations, all needed to communicate and collaborate in order to achieve lean and efficient ‘flow’ of code, and hence, the flow of value, through application delivery pipelines.
Conway’s Law and Silos
Conway’s Law states, in essence, that the boundaries of components and subsystems within any large, complex system that’s being delivered, will reflect the boundaries of the teams responsible for creating and delivering those components, as well as the communication pathways between those teams. Instead of the architecture representing the ideal architectural decomposition of the system, the architecture ends up instead representing how teams are siloed and organized. Therein lies the challenge; as these teams are segmented by organizational boundaries, like business units, programs, or managerial hierarchies as well as by the functional roles of the various stakeholders, application architecture and the ability of teams to collaborate is also constrained. These silos are even more amplified in organizations with heavily outsourced or third-party-contracted resources, where vendors add further fragmentation to the already-siloed teams.
Achieving DevOps’ promised value flow, also known as Continuous Delivery,, requires applications and systems to have ‘loosely coupled’ architectures. Conway’s law, in a badly siloed and fragmented organization, prevents this architectural de-coupling from occurring at the right architectural ‘segmentation’ points and instead forces segmentation along organizational boundaries. This cancels out the promises of DevOps.
Software Delivery Management
The ability to manage value delivery across multiple interconnected application delivery pipelines has been referred to as ‘Software Delivery Management’ (SDM). Effective SDM requires organizations, especially management and leadership, to have visibility into the various software delivery teams, as well as the metrics and KPIs coming from various tools these teams use. They need to be able to correlate and analyze this data to identify inefficiencies in the delivery of value to understand where further investment can enable leaner and more efficient and delivery. In a heavily siloed organization, the silos prevent (or, at least, impede) this streamlined delivery from happening; there is a lack of visibility across functional silos.
Instead, there needs to be an integrated, standardized toolchain or ‘platform’ used by all the functional stakeholders working together to develop and deliver the application. Artifacts need to be able to flow from one functional area to the next, and from one tool to the next, seamlessly. Data needs to flow across the application delivery pipeline to the stakeholders who need the data, as and when needed. Functional silos impede this.
Organizational silos exacerbate this problem. Even with integrated pipelines in place, each pipeline interacts with several others. Application development, as well as application operation, is interdependent. Data must be visible and available across all application delivery pipelines throughout the organization, regardless of organizational boundaries. Finally, organizations must ensure technology stacks are interoperable; tools that do not allow data to be easily visible and ready for analysis, across these technology stacks, make the problems worse.
State of Software Delivery Management Research
The recent State of Software Delivery Management 2020 research report proves that organizational silos continue to remain a challenge to effective SDM.
- More than 60% of survey respondents said functional silos still exist in their organizations.
- More than 85% of survey respondents said silos impede the free flow of information across the organization.
This impact of the challenge silos present is directly reflected by the fact that more than 50% of the respondents said that they do not have access to good quality data from across the toolchain, and also do not have access to data from tools across multiple applications and projects.
Breaking Silos was the imperative that launched the DevOps movement a decade ago, and it’s unfortunate that, in most organizations, as our research reveals, these silos still exist and present challenges to the lean and efficient flow of value through application delivery pipelines.
Breaking down these silos – functional and organizational – needs to be a high-priority initiative for any leader undertaking the modernization and streamlining of their organization’s application delivery capability. It’s proven to be one of the most critical areas of impedance to an organization’s ability to become a lean and efficient application delivery organization, at any scale.
Accelerated Strategies Group conducted research into the state of Software Delivery Management during the spring and summer of 2020. Commissioned by CloudBees, this report is available at no cost at https://accelst.com/state-of-software-delivery-mgmt-exec-summary-2020/.