What is DevOps? … in bullet points, quotes and tweets

Share Button

Interest in DevOps is booming. I feel like I heard it mentioned as the motivation for some decision at least once a week last year, and it climbed its way into the headlines of most of the newsletters I receive from LinkedIn, Twitter, InfoQ, etc.

Army soldiers pulling hard on a rope in a tug of war. Teams working together in a DevOps environment concentrate on all pulling together in the same direction.

I thought I understood it. It just means Dev and Ops collaborating closer, right? But as it gained more and more attention, I realised it must be a movement, not just a simple idea, so I set out to discover what DevOps was really about.

This blog is my summary of what I’ve learned from reading about DevOps over the last few months. There are heaps of resources and lots of people making great observations, so what I’ve done is try to distil lots of salient points into bullet-point format to make it easy for people to pick up as much as possible in a short read.

If you’re interested in reading my sources for yourself, they’re all listed at the end.

Why does DevOps exist? What are the problems that it’s trying to solve?

  • Some of the problems in existing IT organisations are:
    • Conflict between teams and inefficiency
    • Buggy releases
    • Failed deploys
    • Teams with orthogonal priorities
    • Teams inadvertently creating more manual work for each other
    • Complexity of application deploys
    • Inconsistency of management controls and interfaces
    • Dev having a “not my problem” attitude to Production
    • Manual administration doesn’t scale

There's SO MUCH CODE. You need help. Now! All you want to know is: "Who's the Expert?!" Get the Atlassian Stash plugin that tells you.

  • Some of the causes:
    • Dev wants change, Ops wants stability
    • Separated by management structure, geography
    • Different tool sets
    • Pressure to deliver features above improving processes
  • An Agile Dev team releasing on a different cycle to the Ops deploy cycle can result in a Waterfall organisation, in particular extending the feedback loop.
  • What we really want is “business agility”, not just software agility, and “IT alignment” – the whole of IT focusing on key business priorities and strategies.
  • One of the big problems cited is the idea of Dev “throwing code over the wall” to Ops, i.e. minimal coordination between the teams.
  • Some write about “the long, bitter inter-tribal warfare that has existed between Development and IT Operations, as well as between IT and ‘the business'”. (That’s not my experience, but I’ve always worked in small companies where everyone has the “let’s get stuff done” attitude.)
  • There’s a long-standing assumption (frequently proven) that IT projects will run late and won’t deliver what was expected/needed.
  • Many IT groups have a fear of change, a perception that deployments are risky, and issues from misaligned development and production environments.
  • Software projects are actually IT projects, and IT projects have a lot in common with manufacturing processes, especially with regard to coordination between stages to ensure consistent delivery without creating a backlog.

What does DevOps mean? How does it propose to solve these things?

  • It doesn’t mean combining Dev and Ops into one team, or just getting Devs to do Ops work, or automating a bunch of production config.
  • It’s about unified process, making the whole Biz -> Dev -> Ops process one machine that pushes value and feedback in both directions. Really key here is having Ops give feedback to Dev and having Dev do something about it.
  • Ops is potentially a stakeholder in every Dev story – invite them to add requirements. Ops is also a Customer that can produce Dev stories.
  • Dev is potentially a stakeholder in every Ops story. Dev is a Customer that can produce Ops stories.
  • Both teams are responsible for understanding, considering and accommodating the other’s needs.
  • DevOps encourages early and frequent engagement between teams.
  • Operational requirements are as important as functional ones.
  • DevOps is primarily around communication, but some argue that automation trumps communication.
  • It’s a culture change, which may require a change to the way success is measured so that success is when both teams deliver, i.e. Dev can’t be achieving if Ops is not, and vice versa.
  • It’s about unified tooling: making sure Dev, QA and Ops use the same tools as much as possible, the same versions of underlying platforms and software, the same model of what to change and how. Every tool choice needs to consider the impact on the end to end process.
  • Automation is key: machines are much better than humans at executing long lists of complex tasks quickly, in the right order and without error.
  • Is it just Ops doing Agile? It’s similar: Agile software fixes business to Dev misalignment, DevOps fixes Dev to Ops misalignment.
  • However some see it as being a bit grander: Agile aims to solve efficiency problems in one IT function: development. DevOps aims to solve efficiency problems between and across all IT functions.
  • DevOps is NOT a role and that idea is dangerous because it allows cross-functional issues to just become someone else’s problem. You wouldn’t “hire an Agile”.  You can’t “hire a DevOps”.
  • DevOps is not chiefly about better tools. It’s about solving the IT alignment problem, of which tools play a part in the solution. Be clear on what your problems are and which ones you are fixing first before jumping to tool discussions.
  • DevOps is not just about Dev and Ops, but that is a place where many organisations have a huge disconnect so that’s where discussions of mending IT back into one function have begun.
  • DevOps can turn an organisation’s IT capability into a competitive advantage.
  • Just as software teams have software design patterns to aid communication and prevent wheel re-invention, teams that manage infrastructure and deployment need infrastructure and deployment patterns.
  • Organisations that go a long way down the DevOps path often end up with Ops becoming providers and maintainers of platforms that Dev use as they please, without requiring intervention from Ops. So they’re doing internal Platform as a Service (PaaS), which probably includes but also extends a “Private Cloud”.
  • There is a related term ‘NoOps’ which is supposed to embody this idea of a private PaaS and the idea that developers never interact with the Operations team. Others have said this is just an extreme implementation of DevOps and calling it ‘NoOps’ is misleading as there is obviously still an Ops team taking care of the PaaS.
  • As more systems move towards being large collections of interacting hosts, managing the system manually becomes impossible and automatic management must become part of the application itself, a feature set that may be developed by Ops, Dev, or most likely a combination.
  • The idea of “Infrastructure as code” is emerging as a key pattern where infrastructure operations are executed by automated, repeatable tools using shared and version-controlled configuration, rather than being performed by hand and ad hoc.
  • The aim of infrastructure automation is not to remove the need for Ops people, but to allow them to spend less of their time fiddling with tools and more time delivering value to the business.
  • The two main (competing) tools in the Infrastructure as Code space seem to be Chef and Puppet, both tools for automating the configuration of large numbers of hosts.
  • There are lots of things Devs can learn from Ops and Ops can learn from Dev. Each team needs to learn to not be afraid of new ideas.
  • Dev need to take more responsibility for their applications running in Production. Some suggest that Dev and Ops share pager duties.
  • Like Agile development, DevOps is about continuous improvement: don’t try to implement a huge plan for monitoring and automating everything, but make frequent incremental advances towards the next most important thing.
  • As ‘polyglot programmers’ is a trend encouraging Devs to arm themselves with many skills, ‘sysadmin coders’ is a trend for Ops to do the same: understand a few tools well, but also have many varied tools on the belt ready to be whipped out for the right situation.
  • I think you’re more likely to find a plethora of multidisciplinary, “can do” people in startups, where people HAVE to take on multiple, varied responsibilities or the company doesn’t run. Larger organisations, on the other hand, tend to encourage people to specialise, become experts, and often not “interfere” with things where they’re not an expert. (That’s why consultants are brought in, right?)
  • Through communication and shared goals, the whole IT organisation can be pulling in the one direction.
  • Getting started with DevOps chiefly involves a change of attitude: a willingness to collaborate, focus on what the business needs and help both teams to move in the same direction together.

Best Quotes from Articles I’ve Read

These are some quotes from the literature that I thought were such gold that I knew I couldn’t improve them so I just copied verbatim:

  • “Currently, DevOps is more like a philosophical movement, not yet a precise collection of practices, descriptive or prescriptive”
  • DevOps is “a framework of ideas and principles designed to foster cooperation, learning and coordination between development and operational groups”
  • DevOps is “the concept that developers and operations need to work together as part of a team towards common goals rather than being two warring factions who hate each other”
  • “A model where teams work together to produce consistent, high-quality software that delivers business value using a set of processes that are integrated at every step together”
  • “Success at scale depends on the ability to consistently build and deploy reliable software to an unreliable hardware and platform that scales horizontally”
  • “The end goal of infrastructure as code is to perform as many infrastructure tasks as possible programmatically. Key word is “automation.”
  • “Server configuration, packages installed, relationships with other servers,… should be modelled with code to be automated and have a predictable outcome, removing manual steps prone to errors.”
  • “What this means is that managing, designing, deploying and testing infrastructure should be done in an analogous fashion to how we do these same things with software. The code that  builds, deploys and tests our infrastructure should be committed to source-control in the same way as the code that builds, deploys and tests our software is.”
  • “Continuous deployment is a prerequisite for the high deploy rates characterized by DevOps, and is therefore a needed skillset for the modern DevOps practitioner.”

Quotes from the Tyro Team

I asked a couple of people at work how they’d define DevOps in one or two sentences:

  • “An attempt to break the introversion of both groups.”
  • “DevOps recognises another system in business that needs to be adjusted to new ways of doing business. It improves IT service delivery agility by harvesting synergies, reduce friction points, re-thinking and re-imagining processes and encouraging system thinking.”
  • “It feels like the line separating Devs and ops is becoming less defined. To be able to meet today’s requirements of continuous delivery, high availability and security, the people involved need an understanding of each other’s jobs and a big picture perspective.”

Tweets

I’ve been keeping an eye out for DevOps definitions on Twitter and also pinged a bunch of people asking them to define DevOps in a tweet:

My Definition of DevOps

After soliciting so much information from other people who’ve generously contributed their thoughts to the web, I felt I should probably try to summarise what I’ve collected into my own succinct definition. The best I can do at the moment is this:

DevOps is about getting everyone in the organisation (particularly Development and IT Operations) pulling in the same direction by (a) setting common goals and priorities, (b) sharing the responsibility for success and failure across the organisation, and (c) constantly improving communication channels, processes, infrastructure and tools to remove bottlenecks and allow a consistent flow of value to the business’ customers.

DevOps at Tyro

Even though I’ve only recently been grasping the depth of what DevOps is, at Tyro Payments, with Agile company ethos and a very smart Ops team, we’ve been making some big strides down the DevOps path over the past couple of years. Here are some of the highlights:

  • Dev and Ops sit 5 metres from each other and we freely and regularly interrupt each other during the day.
  • Both teams are under the CIO and we meet weekly to discuss cross-team priorities and interaction.
  • In the rare circumstances that production incidents occur, Dev and Ops come together to have all hands at the pumps, collaborating, troubleshooting and executing as one team.
  • We release our software every two weeks, and we’ve coordinated Dev, Ops and QA so that it only takes two business days to get released software deployed into Production. It’s not exactly 10 deploys a day, but it’s pretty impressive for a highly-regulated financial institution.
  • The Dev team takes chief responsibility for the quality of the software (i.e. testing everything), with QA providing an awesome suite of alternative regression tests that act as the Production vanguard.
  • The majority of the task of deploying applications is automated using Maven repositories and an in-house webapp that remotely controls deploys to our production servers.
  • The majority of our configuration is internal to (and deployed with) the apps, except for small numbers of temporary overrides that are verified by Dev. We run pre-release tests that ensure the configuration for the Prod environment has everything that Dev environments have.
  • One or two Devs pair with the Ops guys while they do the deploy.
  • We recently sent a Dev “on secondment” to Ops to work on a major infrastructure project for a couple of months.
  • The Ops team is currently working on migrating applications to “infrastructure as code” servers that can be rebuilt from config without manual steps.

Have I Missed Something Important?

Obviously I haven’t read every blog post about DevOps, or even delved that deep into the details of implementation. My aim here was simply to get a grasp on the base definition and the major steps being proposed as important pieces of solving the puzzle.

Having said that, if there’s things I’ve omitted that you think are crucial to understanding DevOps, please add a comment below. If you’d like to have a go at defining what DevOps solves and how in a tweet, I’d love to receive more of these at @evolvable.

Share Button

Want to learn more?

Here’s a selection of books that either explain DevOps and I.T. transformation in more depth or describe the practices that are being used for implementation. Many of them are authored by those who have been instrumental in the DevOps movement – guys who are not just on the speaking circuit but at the coalface pioneering the next phase of IT delivery.

Upcoming Book

The DevOps Cookbook is still being written but I think it looks really promising:

“The intent behind The DevOps Cookbook Project is to catalog what the ‘high performing DevOps organizations’ all have in common, and then provide prescriptive guidance so that other organizations can replicate their results.”

Sources

DevOps on Wikipeida

What is DevOps? by Damon Edwards

DevOps is not a technology problem. DevOps is a business problem. by Damon Edwards

Use DevOps to Turn IT into a Strategic Weapon by Damon Edwards

The Rise of DevOps with Jesse Robbins (interviewed by Harry Brumleve)

What Is This Devops Thing, Anyway? by Stephen Nelson-Smith

What DevOps means to me by James Turnbull

What is DevOps? by Mike Loukides

What is DevOps not? by Michael Brunton-Spall

The Top 11 Things You Need To Know About DevOps by Gene Kim

Does Automation Replace Humans? by John Willis

Concise Introduction to Infrastructure as Code by Dmitriy Samovskiy

Infrastructure as Code by Carlos Sanchez

Deployment management design patterns for DevOps by Alex Honor

Test-Driven Infrastructure with Chef – Book Review by James Betteley

The Phoenix Project: A Novel About IT, DevOps, and Helping Your Business Win

Where To Learn More About Concepts In “The Phoenix Project” by Gene Kim

The DevOps Cookbook Project

The State of Open-Source Monitoring by Jason Dixon

Image Credit: Labour Day Events by Donald Watkins

7 thoughts on “What is DevOps? … in bullet points, quotes and tweets

  1. Hello Graham – great collection of quotes and ideas about DevOps – would be interesting to explore the concept of DevOps with Security – I’ve seen some refer to it as DevSecOps or DevOpsSec – either way, the idea of marrying security with DevOps so that it is integrated throughout the SDLC is a good concept. And it should be done in a way to support agile.

    Thanks, Mark

    IT/CIO Thought Leader
    Twitter @mtroester
    Blog http://www.sonatype.com/people/

    • It’s a good point, Mark. Security is so important in everything these days, especially as so many modern systems are publically accessible.

      In my experience, Ops guys tend to have spent more time studying security than Dev guys, so that is where I usually head for security advice (though as Devs I believe we need to be learning more and more about these issues). At Tyro, this is effected by Ops potentially being a customer in every Dev story – the “early and frequent engagement” directive. If what Dev are working on is in any way new from an architecture or user capability perspective, we’ll include some Ops guys to ensure we have the right protections in place.

      So, in my mind, Sec is everyone’s responsibility, but presently Ops have the most chops and Dev need to be active in availing themselves of that. Your view seems to suggest you’ve seen environments where ‘Sec’ is a different role from Dev or Ops. Is that right? If that’s the case, I’d definitely be bringing them into the tent as early as possible.

  2. Pingback: Michael T. Nygard on 'Five Years of DevOps: Where are we Now?' at YOW!Evolvable Me

  3. Pingback: 'Avoiding Speedbumps on the Road to Microservices' by Scott Shaw (Notes from YOW! 2014)

  4. Pingback: Microservices at Tyro: An Evolutionary Tale (Presentation)

  5. Pingback: Healthcare DevOps Challengedaniel kivatinos

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.