I was interviewed recently by the folks at Ranger4 for their #DevOpsFriday5 question series. Since June 2014 (when I was interviewed) I have published a couple of things which expand on the original answers, so I have outlined these here. The questions were:
- What’s your preferred definition of DevOps?
- When people ‘do’ DevOps, what’s the most common mistake you see them make?
- How do you recommend an organisation new to DevOps start?
- What’s your prediction for what DevOps will look like in 2020?
- Where do you like to go to get a DevOps hit?
1) What’s your preferred definition of DevOps?
The real focus of DevOps is surely increased operability of the software systems we work with. As the rate of change to these systems is so high (and the changes so valuable), also because these systems are becoming more distributed and complex, the scope for operational faults is significantly higher than with the more static systems of the past. Therefore highly effective collaboration between Dev teams, Ops teams, and other teams helps to ensure that the new end-user features operate well when deployed, reducing the impact and extent of transitory failures.
I recently wrote a more extended blog post called ‘A useful working definition of DevOps‘ based on my work with clients organisations since 2012.2) When people ‘do’ DevOps, what’s the most common mistake you see them make?
Many people have come to understand ‘DevOps’ as mostly just infrastructure automation or release management. They might start a ‘DevOps Team’ whose responsibility is to “automate stuff”, but if that team remains isolated from other teams, it will become a new silo, pushing Dev and Ops even further apart. So beware of the permanent ‘DevOps Team’; instead, use a short-lived team to reach a better place and then dissolve the team into product-aligned teams.
Another common problem present in many organisations is that the Development and (in particular) Operations teams are viewed as ‘execute only’ rather than high-skill and ‘value-add’. I expandon why this works against DevOps and software operability in my eBook ‘Software Operability: a DevOps Cornerstone‘.3) How do you recommend an organisation new to DevOps start?
Give space for – and encourage – daily collaboration between Ops people and Dev people: arrange joint pizza sessions, hold a joint social, collaborate on incident triage, organise metrics-focused ‘hack days’, fund a decent project/programme to make logging a first-class concern. All these things will help to build trust between Ops folk and Dev folk, and improve your software at the same time, reducing outages and time needed for diagnosis. Everyone wins.
4) What’s your prediction for what DevOps will look like in 2020?
The practice of developers working highly collaboratively with operations folk will have become the norm, as will software-defined infrastructure and monitoring as a first-class concern. Organisations whose main revenue stream is enabled through software systems that do not follow these practices and principles will probably not be around in 2020.
5) Where do you like to go to get a DevOps hit?
DevOps Days is a great event series. XP Day in London is inspirational. DevOps Weekly is a useful newsletter. Meetup.com has many active DevOps meet up groups. Helping clients to tackle DevOps problems always throws up new problems and solutions.
The book Build Quality In – a collection of Continuous Delivery and DevOps experience reports – has recently been published via LeanPub. Steve Smith and I are editors, and we’re contributing our experiences along with around 20 other practitioners around the world. 70% of royalties from the book are being donated to Code Club. We expect the book to be valuable resource for people trying to adopt and sustain DevOps and Continuous Delivery as good practices for their organisation.
Matthew Skelton 