Success Patterns for Enterprise DevOps and Scaled Agile Framework
A few months ago, I found myself in a tense and challenging meeting at one of our large enterprise customers. On one side of the table was a group of developers and the lead architect for a mission critical business application, and on the other was the Scaled Agile Framework (SAFe) Program Lead and Agile Scrum Master. As a representative of the DevOps transformation team, I was there to share insights and recommendations and answer any pipeline or automation capability questions that might arise. The topic was the timing of key application features that would be released in the upcoming months. In theory, this should have been a typical planning meeting with open discussion of the pros, cons, opportunities and limitations of the proposed changes, eventually reaching consensus on the new backlog.
In reality, things were going very differently.
The appdev and engineering resources were frustrated. This was the third re-plan in the past few months, and the administrative toll was adding up. Recent advancements in the organization’s Continuous Integration (CI) and Continuous Deployment (CD) pipelines had enabled faster and more frequent releases, but the organization seemed to be moving in the opposite direction: large releases with longer testing cycles due to increasing cross-portfolio integrations and dependencies. Business value-creating features were being delayed or cut to allow time for program-wide testing, manual validation, and multi-level approval activities. The delivery team felt powerless against the tide of increasing overhead and top-down control on their release cycle.
Ability versus Reality
The meeting highlighted a key misstep in the customer’s Enterprise DevOps transformation program, and unfortunately it was not unique. They were capable of moving fast, but the organization was creating barriers to that velocity in the form of increased administration, oversight, and control. A chart from the recently published State of DevOps Report for 2019 illuminates just how common this is.
From this diagram, it’s clear that a majority of organizations *can* release quickly (65% of those surveyed indicated being capable of releasing more than once per day or “on demand”, as shown in the left column), but far fewer actually *do* (only about half of those with the capability, or about one third of the global total, as shown in the right column).
What is causing this divide in capabilities and outcomes?
For this organization, leadership had lost focus on the vision of its DevOps transformation and allowed a complex bureaucratic structure to create new barriers in the flow of releasing software. A recent evolution of their Agile delivery practices towards the Scaled Agile Framework (SAFe, for short) was the unrecognized culprit.
SAFe: A Framework for Scaling Agile Practices
Here’s a quick background on SAFe for the uninitiated. SAFe emerged in 2011 as a “knowledge base of proven, integrated principles, practices and competencies to implement Lean, Agile and DevOps at scale”. It provides concepts and guidelines that help large organizations move away from Waterfall-based delivery models and towards continuous improvement. SAFe’s standards and structure give large organizations the tools to manage and guide enterprise-scale Agile delivery at the Portfolio, Product and Team level. (Watch a 3-minute video explaining SAFe here.)
Used appropriately, SAFe drives business agility and enables leaders with key insights and metrics that aid strategic decision-making for complex application portfolios. Used ineffectively, the increased visibility and controls that SAFe gives leadership can actually increase the distance, divide and distinction between teams by creating an entry-point for the fixed, linear mindsets of the past (i.e., forcing the org into sequentially flowing from requirements to analysis, design, coding, testing, and finally release). Unintentionally, this customer had allowed the latter to happen. Rather than using SAFe to reframe the narrative on how software was built, tested and delivered, they were repackaging the inefficiencies of sequential testing and validation that had existed in their legacy delivery model. The result is that the customer’s implementation of SAFe was limiting, if not outright preventing, the business outcomes that their DevOps teams were producing.
Navigating Digital Transformation Challenges
SAFe is the first step many organizations take to move themselves away from Waterfall and considering the key benefits of top-level visibility and control that it provides, it’s not surprising that “old” ways of thought might creep back into practice. In this case, management was overzealous with these new fine-grained controls, and was applying them to drive complex new validation activities. Without realizing it, they were directly cutting into most critical outcome of DevOps: team empowerment.
This instance highlights the challenges many organizations face when trying to change the fundamental mindsets and principles within an organization, especially when a transformation is seeded internally. Thankfully, there are specific steps that can help overcome these hurdles and realize meaningful improvements in the velocity and quality of software development activities.
Start with the Right Definition
First, it’s critical to start with the right shared definition of “Enterprise DevOps”. For some organizations, Enterprise DevOps is simply considered to be an incremental step on their Agile evolution. At Dell Technologies, we go much farther than that. We consider Enterprise DevOps to be the groundswell of automation, integration and orchestration capabilities that fundamentally change the way we understand and deliver business value for the business. We find that organizations who apply this definition are inherently more prepared to identify and remove barriers in the flow of creating measurable business outcomes.
Know Your Metrics and Set Improvement Targets
Next, organizations must establish consensus on the intended outcomes of their Enterprise DevOps program. But organizations must also consider the how they will measure success looking forward. What metrics will be tracked and how? Consider the key transformation outcomes presented below. What aspirational targets in these areas will be used to define success? Communicating and socializing these targets is extremely important because it helps stakeholders understanding and see the value of what they are doing, which ultimately leads to better outcomes.
According to SAFe, the framework should produce a 30-75% improvement in time to market compared to traditional Waterfall delivery. Organizations must determine if that improvement target is good enough, entertaining such questions as:
Will moving from a 4-month to a 2-month delivery cycle keep the company ahead of its competitors? Will it produce the agility necessary to handle to industry disruption? Will it foster a positive culture that attracts or retains the technical and engineering talent necessary for long-term growth?
In answering these questions, organizations should consider also the potential impact of fully-empowered Enterprise DevOps teams. Massive shifts in application velocity and quality are achievable through automated, integrated pipelines, self-service resources that enable users with the services they need on-demand, and a leftward shift of testing earlier into the development cycle stages of development. Complex application dependencies, which often hamstring traditional release processes, can be proactively managed through a combination of architecture refactoring, code simplification, feature toggles, integrated testing, and other proven Enterprise DevOps practices.
The High-Performing DevOps Organization
When considered in totality, the capabilities of Enterprise DevOps will deliver orders of magnitude improvements over past performance. The table below, from the 2017 State of DevOps Report, quantified the results achieved by high performing DevOps organizations.
Third, Dell Technologies recommends that enterprises create a Value Stream of the complete feature delivery process, all the way out to production. This helps every stakeholder understand where value is being created for the business, and provides a natural lens to identify, define and act on improvement opportunities. During this exercise, it’s helpful to describe both the AS-IS baseline (which may be achieved through exercises like maturity assessments and Value Stream Mapping), and the desired TO-BE future state. This further strengthens the transformational vision by informing specific focus areas for change in the flow of business value.
Treat Digital Transformation as a Journey, Not an Outcome
Finally, organizations must be honest and open to hard discussions about mistakes and missteps on their transformation journey. Learning from failures, and taking the appropriate steps to not repeat them, is critical. A recent empirical study by Comparative Agility found that one of the most significant drivers of positive outcomes of Agile transformation is whether an organization acted on findings from retrospectives in a timely manner. This is powerful, because it implies that success is not an end-state, but a moving target against which organizations must constantly be measuring and correcting themselves.
Stepping back into that customer meeting, it was clear that the organization needed to reassess what they wanted to achieve with their Enterprise DevOps transformation. SAFe was helping to scale Agile practices and centralize and unify decision-making, but it was also inadvertently adding artificial upper limits on the velocity of application development. To move forward, these limitations needed to be discussed, understood, and appropriately challenged. Advocates of the increased testing and manual approval cycles needed exposure to the capabilities and potential of their Enterprise DevOps teams, so that they could evaluate whether the quality and control requirements that they were pursuing could be achieved in a less disruptive way.
Taking the right steps to build revolutionary Enterprise DevOps capabilities and empower teams is not easy. Doing so against the headwind of an unfocused SAFe implementation can be even harder. Dell Technologies recognizes the power of real Enterprise DevOps transformation, and has proven experience achieving it. We know that DevOps teams are ready to release business value safely and quickly, but they must be enabled with a pathway to success in the Enterprise.
When your organization is ready to take the next step, I recommend you connect with Dell Technologies’ DevOps Services team to learn more.