DevOps 101

In the years leading up to the 2009 DevOpsDays conference tour, the term "DevOps" was relatively unknown to the engineering and technology stratosphere. The inception of DevOps-oriented culture was provided by Patrick Debois at an agile infrastructure conference in 2008. During this conference, Patrick spoke of a highly collaborative development team he worked with during his tenure at a large enterprise. The most highly collaborative moments during this tenure were when there were site outages or emergencies. During these incidents, the developers and operations people seemed to be laser-focused and worked incredibly well together. This experience gave Patrick a yearning to encourage this behavior outside of non-emergency activities.

It was at the agile infrastructure conference that Pattick Debois was able to also connect with Andrew Shafer (who then worked at Puppet Labs, Inc.). These two soon found out that they shared many of the same goals and ideologies. In many senses, this chance encounter encouraged Patrick to continue to push the fledgling concept of DevOps forward. In future conferences, Patrick tried fervently yet unsuccessfully (at agile infrastructure conferences) to encourage a more collaborative approach to software development and delivery. While the idea was novel, the practical implementation of the idea never seemed to gain traction at the venues provided to Patrick.

It was in 2009 that Patrick Debois attended an O'Reilly Velocity conference, where he heard John Allspaw speak of how Ops and Dev could collaborate. From this speech, the idea of DevOps was seeded in his mind. Patrick decided to begin hosting a set of mini DevOpsDays conferences, which would eventually catapult the concept of DevOps into mainstream engineering cultures.

While there is yet to be a concise, one-line summary of everything that DevOps entails, there has come about a generally accepted agreement on the overarching concepts and practices that define DevOps: culture, automation, measurement, and sharing, or CAMS for short. The CAMS approach to DevOps was defined by Damon Edwards and John Willis at DevOpsDays in 2010. It is described in greater detail next.

Culture

One of the generally accepted concepts to arise out of the DevOps movement is a cultural one. With traditional IT organization being isolated from development, silos are a commonplace within organizations worldwide. In an effort to pave the way for rapid development and delivery, a fundamental change in organizational culture must take place. This would be done in an effort to promote collaboration, sharing, and a sense of synergy within the organization. This cultural change is indeed probably the most difficult aspect of a DevOps adoption in an organization.

Automation

Automating once-manual processes is critical for a successful DevOps transformation. Automation removes the guesswork and magic out of building, testing, and delivering software and enforces the codification of software processes. Automation is also among the more visible aspects of DevOps and provides one of the highest returns on investment (ROIs).

Measurement

Measuring successes and failures provides critical business data and helps pave the way for higher efficiency through effective change. This simply emphasizes that business decisions can be made through data and metrics rather than gut reactions. For a DevOps transformation to be a success, measuring things such as throughput, downtime, rollback frequency, latency, and other related operational statistics can help pivot an organization toward higher efficiency and automation.

Sharing

In stark contrast to the previously accepted paradigm of software development, sharing is pivotal for a successful DevOps transformation. This means that teams should be encouraged to share code, concepts, practices, processes, and resources. A successful DevOps-oriented organization may even go so far as to embed an operations employees or QA resource in the development team in order to facilitate autonomy and collaborative teams. Some organizations may also have shared or overlapping roles. This may be realized through some modern development techniques (TDD, BDD and so on).

At the time of writing, there are hundreds if not thousands of DevOps-specific tools. Such tools are designed to make the lives of engineering organizations better or more efficient. While the tools aspect of DevOps is important, it is important to not let a given tool define your specific organization's DevOps process for you. Once again this implementation CANNOT be achieved without applying the CAMS model first. Throughout the course of this book, we will reference and tutorialize an array of different tools and technologies. For you, specifically, it's important that you select and leverage the right tool for the right job.