I like to joke that I went to my first DevOps conference by accident. It’s actually pretty-much true. While browsing for interesting reading back in late 2014, I came upon the website of Configuration Management Camp, aka #cfgmgtcamp, in Ghent.
For those of us whose careers have mainly revolved around the IT Service Management world, the phrase “Configuration Management” tends to evoke the ITSM process, as set down in the ITIL framework, of tracking and recording all of the “configuration items” which underpin the services delivered by the IT function of an organisation. My past life as a professional services consultant often led me into CMDB projects, in a range of complex scenarios such as national emergency service radio networks, the NHS, and large global payment networks. It was something of a personal speciality.
As a Product Manager, part of my portfolio is the IT Asset Management tooling in our enterprise ITSM suite. I’m very much a “get out and talk with practitioners” person, and Configuration Management has a big crossover with ITAM, and the conference I’d found was free and relatively local, and…
I’d actually realised my mistake soon after my impulsive registration. The “Configuration Management” to be discussed in Belgium was defined somewhat differently. This conference belonged to the emergent agile infrastructure crowd, open-source arm of that community specifically, talking about tools like Chef and Puppet and Ansible and lots of others that I’d only barely heard of.
I went anyway. It was a great decision, and I learned a huge amount about a rapidly emerging and very important set of technologies and practices which I felt were going to be critical to the future of enterprise computing. I got to see luminaries like Luke Kanies, Michael Coté and Andrew Clay Shafer, who remain incredibly important people in the DevOps movement. I’ve been going annually since.
I wrote a blog, at the time, and I thought it might be interesting to share it again in full:
Thoughts on #cfgmgmgtcamp, and why ITSM needs to take note
Posted on February 2, 2015
Mention the role of “Configuration Manager” at an ITSM conference, and then use the same description at an Infrastructure Management conference, and your respective audiences will visualise completely different jobs*.
So, it was with some curiosity that I arrived this morning for the first day of the Configuration Management Camp in Ghent.
This particular event falls squarely into the infrastructure camp. It’s the realm of very clever people, doing very clever things with software-defined infrastructure. A glance at the conference sponsors makes this very clear: it includes Puppet Labs, Chef, Pivotal, and a number of the other big (or new) names in orchestrated cleverness.
While this is not the ITSM flavour of Configuration Management, however, today’s conference really made it clear that this new technology will become more and more relevant to the ITSM community. In short, ITSM should make itself aware of it.
The tools here have underpinned the growth of many recent household-name startups: including those internet “unicorns” like Facebook and Uber which have risen from zero to billions. They’ve enabled rapid, cloud-driven growth in a brand-new way. This new breed of companies have firmly entrenched DevOps methodologies, with ultra-rapid build, test, and release cycles, increasingly driven by packaged, repeatable scripts. It primarily takes place on cloud-based open-source software stacks. As a result, there’s not quite as much focus on resource and commercial constraints as we find in big enterprises.
But here’s the crux: methods like this will surely not remain solely the preserve of the startups. As business get deeper into digital, there’s increasing pressure on the CIO from the CEO and CMO, to deliver more stuff, more quickly. As the frontline business pushes its demands onto IT, long development and deployment cycles simply won’t be acceptable. And, with equal pressure on IT’s costs and margins, these technologies and methods will become increasingly attractive.
Rapid innovation and deployment is becoming essential to business success: PuppetLabs 2014 State of Devops study recently claimed that strongly performing organizations are deploying code 30 times more often, with 50 times fewer failures. Okay, those numbers come from squarely within the Devops camp, but they are credibly reinforcing past analysis such as MIT Sloan’s 2007 classic study on “the alignment trap”. IT can’t just be a ponderous back-end function if a company wants success.
That’s not to say that this conference is providing the answers. I’d argue that Configuration Management Camp is the “bottom up” conference to an ITSM conference’s “top down”.
Some of the toolsets on display here are very granular indeed. Many of the presentations were slides, or live demos, full of slightly arcane command-line-interface text input. We watched individual clusters get provisioned, quickly, impressively, but devoid of context.
However, there was also a sense of an increasing imperative to start connecting those dots: to define the applications and their inter-dependencies, and ultimately, the services. We’ve seen talks today with titles like “Configuring Services, not Systems”. Dianne Mueller of Red Hat OpenShift described new tools which focus on the deployment of applications, rather than points of infrastructure.
I spoke with more than one person today who described DevOps starting to meet “reality” — that is, the day-to-day expectations of an enterprise environment. There is a feeling of “cutting edge” here, probably justified, but the counterpoint to that might be that this community tends to see “traditional” IT is seen as slow and clunky. PuppetLabs founder and CEO, Luke Kaines dismissed this: “The enterprise space doesn’t move slowly because they’re stupid or they hate technology. It’s because they have… what’s the word… users”.
One thing that was clear today was that these technologies are only in their infancy. Gartner recently identified Software Defined Infrastructure as one of its key trends for 2015. Forrester have declared that Docker, the rapidly-emerging containerisation technology, will “live up to the hype and rule the cloud”.
And that’s why IT Service Management needs to take note.
We can’t understand the services we provide our customers, if we don’t have a grasp on the underlying infrastructure. We can’t formalise change control without getting the buy-in of the technical experts for whom rapid infrastructure shifts are a command-line prompt away. We can’t help prevent uncontrolled software spend, or inadvertent license breach, if we don’t proactively map our contracts onto the new infrastructure. With change cycles moving from weeks to seconds (it was claimed today, in one session, that Amazon deploys code to production on a sub-one-second interval), established ITSM practices will need to adapt significantly.
So, if it feels like ITSM’s “top-down” and infrastructure configuration management’s “bottom-up” are striving to find a connection point, it also feels like that join has not yet been made. It’s up to IT as a whole to make that connection, because if we don’t do it, we’ll end up repeating the lessons of the past. But faster.
It’s going to be a fun challenge to face. This is exciting stuff.
*ITSM: defender of the CMDB. Infrastructure: server deploying and tweaking wizard. Right?
Looking back at this now, I feel that my thoughts at the time have held up pretty well. I remain concerned that much of the change being driven by this community is poorly understood by the service management world. The challenges I talked about for ITSM have certainly materialised. But maybe I didn’t predict the other side of the story: in the years since, I’ve also seen the challenges faced by the agile infrastructure community as their importance grows.
Most of the people I spoke with in 2015 worked for small companies or startups, and a few more worked for rapidly growing enterprises like Spotify. More recently, at this conference, and related events like DevOpsDays, I tend to meet more and more people who are trying to work in these new ways within the bounds of long established large enterprises.
I’m speaking at this year’s Configuration Management Camp about this challenge, and discussing the opportunity for the Service Management world actually to help with the transformation DevOps teams are trying to drive.
I owe this conference a lot. It connected me with great people, and set me on a path of learning and engagement with the DevOps community which has been valuable and enthralling. I have presented at numerous DevOpsDays conferences, been involved in many important conversations around the future intersection of IT service and DevOps, and last year hit a career goal in being asked by Gene Kim and his associates to present at the DevOps Enterprise Summit in Las Vegas. It all started on that chilly Belgian campus in 2015. I’m delighted to be contributing this year.
And Ghent, by the way, is wonderful. Go there.