When I think about the challenges of DevOps transformations, I love the viewpoint expressed by Ryan Harmon, strategic accounts manager at Trace3: “DevOps is a complex maze that has many leaders frustrated. In many large enterprises, ‘DevOps’ is simply a buzzword they are striving to achieve but struggle with what the end state looks like or even where to start. There are many layers across the organization that are instrumental to truly executing on DevOps. DevOps is not something you go get a quote for and simply buy. It’s an evolving journey.”
In my previous article, The Time Has Come for an Engineering Approach to DevOps, I agree with Ryan.
Too many organizations launch into a DevOps transformation initiative without considering the full scope of what it takes to engineer a successful DevOps solution. DevOps is more than “just” a CI/CD pipeline or toolchain, as complex and challenging as this alone may be to accomplish, for even one application. DevOps is more than simply a culture change, as difficult as that always is.
There are 24 DevOps practice categories that need to be mastered to realize and evolve well-engineered, high-performance enterprise DevOps solutions. Not all 24 need to be mastered to begin the DevOps transformation, but, ultimately, all of them do need to be mastered before the solution will achieve the high performance levels to which many enterprises aspire. The good news is, successful enterprises have demonstrated that the investment in people and technologies required for the journey is worthwhile, and a strategic approach will get you there systematically if you follow a disciplined engineering implementation roadmap.
As explained in my book, Engineering DevOps, the core of every successful DevOps practice includes nine core pillars. These are the pillars to focus on as you start on your DevOps journey. The following paragraphs illustrate requirements for each of the nine pillars.
Leadership – Leaders set an inspiring DevOps directional vision for the organization, and proactively stimulate and sponsor team activities that help progress toward goals.
Collaborative culture – The organization encourages cross-functional collaboration, shared responsibilities and works to reduce restrictive boundaries between organizational functions.
Design for DevOps – Products are architected to support service-oriented, modular, independent packaging, testing and releases, in accordance with the principles of twelve-factor apps.
Continuous integration – Changes to software code and artifacts needed for releases are frequently merged, built, tested and packaged into executable artifacts for application releases.
Continuous testing – Software tests that are most important to verify the results of pipeline stages including pre-flight, integration, regression, performance and release acceptance tests are automated as much as possible.
Elastic infrastructure – Infrastructure-as-a-service (IaaS) resources (i.e., software, computing machines, storage and networks) are integrated into a toolchain for automating builds and testing environments to efficiently support variable workload demands.
Continuous monitoring – Applications, pipeline tools and infrastructure components are instrumented with metrics, and the collected data is analyzed to manage health and performance.
Continuous security (DevSecOps) – Security practices, processes and technologies for applications, databases, pipelines and infrastructures are integrated into the culture and pipelines to assure security.
Continuous delivery – Release artifacts are validated, prepared for deployment to production and may be automatically deployed to production once acceptable deployment measures are achieved.
A Comprehensive Engineering Blueprint
While the above nine practice categories are the core of every successful DevOps solution, 14 additional practice categories are essential parts of a comprehensive engineering blueprint for accomplishing high-performance DevOps.
DevOps infrastructure practices are used continuously by the nine core practice categories as follows:
Version management systems – Every stage in the end-to-end DevOps value stream depends on keeping track of versions of software source code, configuration data and executable images using source code management systems, artifact repositories and configuration management systems.
Infrastructure-as-a-Service – Virtualized cloud and dedicated resources must be served up ephemerally to meet variable workload and storage demands, which requires mastering engineering practices for orchestrating dedicated and cloud, hybrid cloud and multicloud services, and deployment of applications to the infrastructure.
DevOps pipeline practices, which constitute a layer just above the nine core practices, include two additional categories that make use of the nine core DevOps practices.
Application release automation – Directing and monitoring software changes through the CI/CD pipeline requires process orchestration and data analysis working at a level above the CI/CD pipeline itself.
Value stream management – A deployed service, composed of multiple application microservices and data services, requires orchestration and analysis of a federated set of services, end-to-end, across the value stream from planning through to operations.
Management Practices
DevOps management practices include four categories that make use of the DevOps Pipeline practices, as follows:
Service catalogs – Organized collections of software stacks, tools and pre-configured pipelines, made available through portals, enable DevOps-as-a-Service operations and simplify access, configuration, control and evolution of DevOps resources needed by enterprise application teams.
Disaster prevention and recovery – Proactive disaster monitoring coupled with high-availability configurations and disaster mitigation practices, such as automated processes, work together to minimize the blast radius of disastrous events, and orchestrate rapid restoration of essential DevOps services, when needed.
Site reliability engineering (SRE) – A combination of practices which include service level objectives (SLOs), error budgets, monitoring, work-sharing policies, toil reduction, proactive anti-fragility practices and blameless post-mortems help to “shift left” production wisdom from operations into development, thus improving the efficiency, readiness, security and resiliency of applications and infrastructures in production.
Governance – The scale and scope of enterprise DevOps solutions are controlled using governance policies-as-code implemented through service catalogs, management dashboards and error budget policies.
Project Practices
DevOps project practices include four categories that guide enterprise adoption and evolution of DevOps, as follows:
Adoption – Enterprise-wide DevOps adoption strategies that engineer and build DevOps capabilities in logical phases. This begins with a small set of representative applications that serve as a model for other applications; this method has been shown to be more successful than “boil the ocean” strategies that attempt to transform all applications at the same time.
Model applications – Applications – that can serve as a model for greater enterprise adoption – are selected and monitored according to criterion that spotlight the value of DevOps for other applications.
Return on investment (ROI) – DevOps investments must compete with other IT projects for resources. An ROI model that demonstrates measurable cost savings is an approach that has proved to achieve greater numbers of DevOps project approvals compared to other methods.
Continuous improvement – Competitive forces and industry trends require that DevOps solutions continue to evolve to keep up with enterprise requirements.
The Human Element
Human development practices include two categories, as follows:
DevOps engineering – The development of DevOps engineer roles, skills and the structure of DevOps engineering within an organization is critical to the success of designing, building, maintaining and evolving DevOps solutions.
Continuous learning – A DevOps continuous learning strategy is critical for the development and transfer of DevOps knowledge needed to develop and continuously improve DevOps for the enterprise.
In summary, the importance of practice categories – and specific practices within each of the categories – will evolve as each application does, alongside the changing needs of the enterprise. DevOps is complex, but a comprehensive engineering strategy that considers the 24 DevOps practice categories described in this article can help enterprises achieve the high-performance DevOps solutions they need to justify the investment and to meet business goals.
