DevOps: Bringing development and operations together
What is DevOps?
Call it a culture, a movement, a philosophy. In short, DevOps is a partnership between development and operations that emphasizes communication, collaboration, and integration.
DevOps aims to unite the agile and innovative nature of software development with the core tenets of operations — security, stability, and a kick-ass customer experience.
In a DevOps culture, operations and IT professionals understand and participate in software development (applications) and likewise, the software developer learns and participates in IT operations (infrastructure).
From continuous delivery and faster time to market to lower failure rates and faster fixes, the technical and business benefits of adopting DevOps are multiple.
- Continuous delivery
- Less complex problems to fix
- Fewer failures
- Faster resolution of problems
- More time to focus on high-value features
- Increased customer satisfaction
- Reduced infrastructure costs
- Faster delivery of features
- Increase in sales
- More stable operating environments
History of DevOps
The DevOps movement started to coalesce some time between 2007 - 2008, when IT operations and software development communities got vocal about what they felt was a fatal level of dysfunction in the industry.
They railed against the traditional software development model, which called for those who write the code to be organizationally and functionally apart from those who deploy and support that code.
Developers and IT/Ops professionals had separate (and often competing) objectives, separate department leadership, separate key performance indicators by which they were judged, and often worked on separate floors or even separate buildings. The result was siloed teams concerned only with their own fiefdoms, long hours, botched releases, and unhappy customers.
Surely there’s a better way, they said. So the two communities got together and started talking – with people like Patrick Dubois, Gene Kim, and John Willis driving the conversation.
What began in online forums and local meet-ups is now a major theme in the software zeitgeist, which is probably what brought you here! You and your team are feeling the pain caused by siloed teams and broken lines of communication within your company.
You’re using agile methodologies for planning and development, but still struggling to get that code out the door without a bunch of drama. You’ve heard a few things about DevOps and the seemingly magical effect it can have on teams and think “I want some of that magic.”
The bad news is that DevOps isn’t magic, and transformations don’t happen overnight. The good news is that you don’t have to wait for upper management to roll out a large-scale initiative. By understanding the pillars of DevOps and making small, incremental changes, your team can embark on the DevOps journey right away. Let’s start with the CALMS framework.
First described by John Willis and Damon Edwards in 2010, CALMS stands for Culture, Automation, Lean, Measurement, and Sharing. Let’s talk about each in turn.
If we could sum up DevOps culture in one word, it’d be “collaboration” – and if we were allowed two words, they’d be “cross-functional collaboration.” (Ok, that’s more like three words.)
All the tooling and automation in the world are useless if they aren’t accompanied by a genuine desire on the part of development and IT/Ops professionals to work together. Because DevOps doesn’t solve tooling problems. It solves human problems. Therefore, it’s unlikely you’ll poke your head out of the cubicle one day, look around, and discover that teams at your company embody DevOps culture. But there are simple things you can do to nurture it.
Think of DevOps much like agile, but with the operations included. Forming project- or product-oriented teams to replace function-based teams is a step in the right direction. Include development, QA, product management, design, operations, and any other skill set the project requires. At Atlassian, we even embed marketing with our product teams.
Few things foster collaboration like sharing a common goal and having a plan to reach it together. At some companies, switching suddenly to project-based teams is too much, too soon. So take smaller steps. Development teams can – and should – invite appropriate members of the operations team to join sprint planning sessions, daily stand-ups, and sprint demos. Operations teams can invite key developers. It’s an agile and organic way to keep on the pulse of each other’s projects, ideas, and struggles. The time spent listening and cross-pollinating subject-area knowledge pays for itself by making releases and emergency troubleshooting far more efficient.
And speaking of emergencies, they’re an effective test of DevOps culture. Do developers, operations, and customer support swarm on the problem and resolve it as a team? Does everyone start with the assumption that their teammates made the best decisions possible with the information and resources they had at the time? Is the incident post-mortem about fixing processes instead of pointing fingers? If the answer is “yes,” that’s a good indication that your team functions with DevOps culture at its core.
Note that the most successful companies are on board with DevOps culture across every department, and at all levels of the org chart. They have open channels of communication, and talk regularly. They make sure everyone’s goals are aligned, and adjust as needed. They assume that keeping customers happy is just as much product management’s responsibility as it is the development team’s responsibility. They understand that DevOps isn’t one team’s job. It’s everyone’s job.
Teams that practice DevOps deploy 30x more frequently, have 60x fewer failures, and recover 160x faster.
Investing in automation eliminates repetitive manual work, yields repeatable processes, and creates reliable systems.
Build, test, deploy, and provisioning automation are typical starting points for teams who don’t have them in place already. And hey: what better reason for developers, testers, and operators to work together than building systems to benefit everyone?
Teams new to automation usually start with continuous delivery: the practice of running each code change through a gauntlet of automated tests, then packaging up successful builds and promoting them up toward production using automated deploys. As you might guess, this is not a quick and easy thing to set up, but the return on investment is well worth it.
Why? Computers execute tests more rigorously and faithfully than humans. These tests catch bugs sooner, allowing developers to fix them more easily. And the automated deploys alert IT/Ops to server “drift” between environments, which reduces or eliminates surprises when it’s time to release.
Another of DevOps’ major contributions is the idea of “configuration as code.” Developers strive to create modular, composable applications because they are more reliable and maintainable. That same thinking can be extended to the infrastructure that hosts them.
True, systems are always changing. But we can create a facade of immutability by using code for provisioning so that re-provisioning a compromised server becomes faster than repairing it – not to mention more reliable. It reduces risk, too. Both development and operations can incorporate new languages or technologies via the provisioning code, and share the updates with each other. Compatibility issues become immediately apparent, instead of manifesting in the middle of a release.
“Configuration as code” and “continuous delivery” aren’t the only types of automation seen in the DevOps world, but they’re worth special mention because they help break down the wall between development and operations. And when DevOps uses automated deploys to send thoroughly tested code to identically provisioned environments, “Works on my machine!” becomes irrelevant.
When we hear “lean” in the context of software, we usually think about eliminating low-value activities and moving quickly – being scrappy, being agile. Even more relevant for DevOps are the concepts of continuous improvement and embracing failure.
A DevOps mindset sees opportunities for continuous improvement everywhere. Some are obvious, like holding regular retrospectives so your team’s processes can improve. Others are subtle, like A/B testing different on-boarding approaches for new users of your product.
We have agile development to thank for making continuous improvement a mainstream idea. Early adopters of the agile methodology proved that a simple product in the hands of customers today is more valuable than a perfect product in the hands of customers six months from now. If the product is improved continuously, customers will stick around.
And guess what: failure is inevitable. So you might as well set up your team to absorb it, recover, and learn from it (some call this “being anti-fragile”). At Atlassian, we believe that if you’re not failing once in a while, you’re not trying hard enough.
We challenge our teams with big, hairy, audacious goals and make sure they have the autonomy and the resources to meet them. We hire smart, ambitious people and expect them to fail sometimes.
In the context of DevOps, failure is not a punishable offense. Teams assume that things are bound to go pear-shaped at some point, so they build for fast detection and rapid recovery. (Read up on Nexflix’s Chaos Monkey for an excellent example.) Postmortems focus on where processes fell down and how to strengthen them – not on which team member f'ed up the code. Why? Because continuous improvement and failure go hand in hand.
It’s hard to prove your continuous improvement efforts are actually improving anything without data. Fortunately, there are loads of tools and technologies for measuring performance like how much time users spend in your product, whether that blog post generated any sales, or how often critical alerts pop up in your logs.
Although you can measure just about anything, that doesn’t mean you have to (or should) measure everything. Start with the basics:
- How many story points did your team complete this sprint?
- How often do recurring bugs or failures happen?
- How long does it take to recover after a system failure?
- How many people are using your product right now?
- How many users did you gain / lose this week?
With a solid foundation in place, it’s easier to capture more sophisticated metrics around feature usage, customer journeys, and service level agreements (SLAs). The information you get comes in handy when it’s time for road mapping and spec’ing out your next big move.
All this juicy data will help your team make decisions, but it’s even more powerful when shared with other teams – especially teams in other departments. For example, your marketing team wants shiny new features they can sell. But meanwhile, you’re seeing high customer churn because the product is awash in technical debt. Providing user data that supports your roadmap – even if it’s light on features and heavy on fixes - makes it easier to build consensus and get buy in from stakeholders.
The future of DevOps
The gains teams have made in quality, time to market, and employee happiness mean DevOps is here to stay. We believe companies that value transparency and collaboration will continue to produce the best products and lead their industries.
Given the pace at which technology is changing, it’s nearly impossible to predict what the state of automation or monitoring might be ten years from now. But one thing is certain: technical problems will always be human problems at their core. DevOps’ focus on people and process means it will remain relevant for the current generation of software makers, and the generations that follow.
Before you go...
Tooling can't magically make your team "DevOps." But we happen to make some pretty cool tools that encourage collaboration
and make working with a team feel less like "work."