Puppet Labs’ fourth annual conference kicked off with high energy yesterday morning. With thousands of people attending the conference in person at the Marriott Marquis in San Francisco and on the livestream, there was plenty of lively conversation in the venue hallways and lounges, and in the Twittersphere.
Keynotes and talks covered topics that IT ops people and developers care strongly about: workplace culture, continuous delivery, and of course, the nitty-gritty of implementing Puppet. (You can read about a selection of Day 2 talks here. And you can see any or all of 87 PuppetConf 2014 talks and presentations by signing up here.)
Puppet Labs founder and CEO Luke Kanies kicked off with a keynote offering insight into the work we’re doing to make sure Puppet continues to serve its users — and delight them — for the next 10 years.
The plight of sysadmins — being snowed under with repetitive tasks, and too busy putting out fires to spend time on solutions that would un-bury them — was a huge motivation for Luke to build a configuration management and start a company. Our core purpose is still to give sysadmins time back so they can learn more, change more quickly and help their companies do the same. Luke brought Puppet Labs employees onto the stage to demonstrate the new Puppet Server and Node Manager, two of our new technologies for helping sysadmins do their jobs better.
Luke also discussed new Puppet Approved modules, which pass rigorous Puppet Labs standards for composition, design and maintenance, and that follow semantic versioning.
SHIT. My module for Jenkins is puppte aproved? SHIT.— R. Tyler Croy (@agentdero) September 22, 2014
You can read more of what Luke had to say about Puppet’s development in the blog post he published yesterday morning.
Puppet’s rapid adoption and development depends in large part on the community’s contributions to Puppet itself, to the upstream open source projects, and to great Forge modules that anyone can freely use for automating their infrastructure. Luke introduced some stars of the community — Tim Sharpe, Felix Frank, Daniele Sluijters and Erik Dalén — highlighting their contributions and their friendly helpfulness to others, whether on the email lists, in IRC or at live user gatherings.
DevOps is an important element of sysadmin and Puppet culture, and Gene Kim, whose research into what makes IT work well intersects with DevOps, talked about the powerful effects that DevOps practices can have on business results in both his keynote and his talk later with his fellow DevOps researcher, Dr. Nicole Forsgren Velasquez, a professor at Utah State University.
High performers have 30x more deployments and 8000x faster lead time, 2x the change success rate and 12x faster recovery #PuppetConf— Actinide (@uidactinide) September 23, 2014
The focus on culture took another turn as Kate Matsudaira’s keynote, Trust Me, incited a burst of appreciative tweets and enthusiastic hallway conversations. Kate’s frank, practical advice on raising your profile in your workplace started with a given: Most people in tech work hard and sincerely, but don’t always understand that it's important to make others aware of what they’re doing and what they’re planning to do.
The first part of building trust: contribution. How much impact does your work make? Look for gaps to fill. #PuppetConf— Actinide (@uidactinide) September 23, 2014
"If you use your 1-on-1 to talk about status, you're wasting time. Get to know your boss, solicit feedback on your performance." #puppetconf— Randi Harper (@freebsdgirl) September 23, 2014
Communication with one’s manager and team is critical, of course, but Kate also advises taking stock of one’s own reputation and relationships by looking at others on your team you think are doing great work and are reliably helpful. Match yourself against them: Does the quality of your work and relationships stack up? It’s all about raising your head from the work at hand to look at the big picture — something we aren’t always good at, especially when we’re immersed in the technical and operational details.
Kate’s slide deck, featuring Lego figures, was also much admired.
Ship smaller changes more often, and get used to shipping more often. That’s the message of people who practice continuous delivery, and a frequent goal of DevOps initiatives. The value of continuous delivery was reinforced yesterday (and will be again today) by a number of PuppetConf speakers.
Sam Kottler of Digital Ocean, a cloud-hosting service for developers, talked about how his company thinks about managing a large environment and delivering continuous, incremental improvements into that environment, which is managed with Puppet. Three things matter:
- Deliverability. You must be able to deliver services and code as needed.
- The ability to change. “If you aren’t constantly changing things, it becomes painful to change, and you’re fearful of what will happen,” Sam said.
- Mutability. Instead of changing a system or container, why not just blow it away and create new ones as needed? Sam pointed out there’s a false dichotomy in many people’s minds between state and change — one does not actually exclude or prevent the other.
“At the end of the day,” Sam said, “we don’t really care about systems — we care about what’s running on them, what the systems can do.”
R. Tyler Croy is a developer who works at mobile security company Lookout by day and on the open source continuous integration project Jenkins in the rest of his life. His talk centered on what he’s learned about continuous delivery of infrastructure at Jenkins over several years, using first open source Puppet and now Puppet Enterprise, including:
- the benefits of roles and profiles
- rspec and testing, which let Jenkins improve testing
- replacing many manifests with just a few Puppet modules.
Tyler also called out for help with issues in Jenkins, pointing people to GitHub — and got immediate help.
Using Puppet: Tips, tricks and war stories
One of the big benefits of attending PuppetConf is learning about how other admins and developers use it in their environments. Mike Stahnke, director of engineering services at Puppet Labs, talked about getting started with Puppet. Some high points:
- Automate to get things consistent. One of the most-tweeted quotes from Mike.
- Automate first what you already know how to do without automation. If you automate away a number of small tasks, they’ll add up, and you’ll deliver systems to your team much faster.
- Learn how something works first, but don’t learn two things at once.
- Ask yourself whom you’re automating for. The customers who pay you money? Internal customers? This will figure out what’s most important to automate, because you shouldn’t automate for no reason.
- Don’t automate your automation system as your first project. It’s too complex and not a good candidate for learning how to do automation in your environment.
Mike’s talk, which was, as always, frank and engaging, was much appreciated by attendees, both on site between sessions and on Twitter.
Kris Buytaert, a consultant with Inuits in Belgium, outlined what can go wrong with Puppet in his provocatively headlined talk, “7 Puppet Horror Stories in 7 Years.” Stating that “we’re all devs now,” Kris told stories that underlined some important lessons he and his team at Inuit have learned in working with their clients, practical points that were very much appreciated by those who attended Kris’ talk.
Lots of learning, plus food and fun
We spend the entire year before PuppetConf planning it, just to make sure it’s really awesome, useful and fun. So I’ll round up with a few tweets from attendees that make me smile. Thanks to all of you for coming, and so glad you’re getting a lot out of it!
Wow. Lots of great help from puppet employees for really weird problems I've seen. They are probably tired of me now. :) #puppetconf— Randi Harper (@freebsdgirl) September 24, 2014
Aliza Earnshaw is managing editor at Puppet Labs.