Day 2 of PuppetConf 2016 kicked off with Jeffrey Snover, a technical fellow at Microsoft, talking about Azure and Nano Server, and how these two offerings are designed to help customers use the cloud for greater business agility.
Jeffrey described the evolution from Windows Server to Nano Server, and how Microsoft took learnings from how customers use cloud to create Windows Server 2016.
Nano Server goes a step further: It's a much smaller-footprint, faster server designed to work within the larger Azure ecosystem.
Jeffrey talked about more new capabilities, including Hyper-V, which will run containers in a very strong security layer.
Apart from the technical side of his talk, Jeffrey asserted strongly that cloud is the new business model, and that it's the people who understand cloud who will replace today's C-level business leaders. He reminded the keynote audience that by staying current and on top of technology shifts, they can be those leaders.
Last but not least, Jeffrey reminded everyone how pervasive self-doubt can be, even for people who have made it.
How Salesforce went from proof of concept to managing 30,000-plus hosts with Puppet
Today, most people interested in DevOps know that Salesforce.com, the world's biggest SaaS company, relies on DevOps methods and tools to deliver its CRM offering efficiently. But it wasn't always that way. Salesforce, like most other companies, started out with limited automation. It's just a few years ago that the company was still treating servers like pets, according to software engineer Petersen Allen.
Peter talked about how DevOps has evolved at Salesforce.com, and the part Puppet has played in that transformation. One important attribute of Puppet is its declarative language, which allows Salesforce admins to define the state of infrastructure, rather than writing procedural scripts that don't scale very well, and that some team members will have a hard time understanding.
#puppetconf Petersen Allen telling awesome Salesforce DevOps jrny; "mean time to grok" — how long reqd to contrib as new employee or at 3am— Gene Kim (@RealGeneKim) October 21, 2016
Being able to define infrastructure with code also meant that Salesforce could start applying its well-established agile software practices to infrastructure, too.
Operations like patching have become much faster, saving hundreds of hours of ops engineers' time, and giving Salesforce consistent state across all its hosts. Patching is now a simple pushbutton operation.
#puppetconf Allen: "patching used to be hard; we had Patch Parties, so engrs cld focus on patching; now it's so easy, we can actually party— Gene Kim (@RealGeneKim) October 21, 2016
Petersen reminded everyone that if you want to do DevOps, Puppet has to be used by everyone in the organization, not owned by just one team that can then become a bottleneck.
"you wouldn't create a python team, and funnel every script change through that team." #PuppetConf— igor @ ✈ (@hirojin) October 21, 2016
Last but not least, Petersen also urged the audience to learn from Salesforce's experience: Don't start with a mega project, but with a minimally viable product — another DevOps principle.
Docker and Puppet, together
Scott Coulton, author of Puppet for Containerization, describes himself as formerly "a boxes and cables guy." Now he's an authority on Docker, AWS and Puppet — and the use of them together.
In his keynote, Scott addressed what a former employer's culture looked like before and after its adoption of Puppet, Docker and AWS. It was the classic Dev/Ops silo situation: devs would complete code and dump it in the company repo, not caring too much how deployable it really was, and ops would get frustrated, knowing they were the bottleneck but unable to get developers involved with deployment.
Containers became a vehicle for getting devs to deploy code themselves. The operations team took the lead on adopting containers — a somewhat unusual turn, as it's often developers who initiate use of containers. Puppet allowed the ops team to build and manage containers repeatably, keep them in the defined state, and then scale the whole operation up.
Once the ops team had a proven method, they shared it with one application development team. Ultimately, the pattern of using Docker with Puppet to deploy to AWS expanded. And with it, the culture changed: now dev and ops could communicate in the same language.
Scott has since moved on to a small startup, where he's doing all the things. There, too, Docker and Puppet are proving a great combination for shipping applications.
What your CIO cares about, and how to talk to her or him
The CIO's role is changing. No longer is it enough to keep IT stable and dependable. Now IT is being asked to help transform the business with digital innovation.
So what is your CIO thinking about, and how can you work well with your CIO? Puppet's own CIO, Nigel Kersten, led a panel discussion with Sanjay Mirchandani, Puppet CEO and a former CIO; Ronald Muijrers, CTO of the corporate markets division at KPN, the largest telco in The Netherlands; and Peter Richards, who's been the CTO at Goldman Sachs, Bear Stearns, JP Morgan Markets and Bank of America Markets.
The panelists offered their perspectives on what they think the CIO's role will look like in a few years; how to be strategic about apportioning budget to key IT projects; what CIOs expect from their IT teams; and what IT practitioners should expect of their CIOs.
"No one of us is smarter than all of us." - Peter Richards #PuppetConf— Bryan Jen (@WowSuchBryan) October 21, 2016
"The more senior in the org, the less you can direct. More about telling the story, sharing the vision." Peter Richards #PuppetConf— Ian Buchanan (@devpartisan) October 20, 2016
"70-90% of IT budgets focus on keeping what exists running. Turn innovation on that to free more budget for new." Peter Richards #PuppetConf— Ian Buchanan (@devpartisan) October 20, 2016
Last item of the morning was the Puppet community award. Luke Kanies, founder and board member of Puppet, presented the MVP award to Rob Nelson, a much-loved an extremely helpful member of the Puppet commmunity.
Thanks to everyone who came to PuppetConf for your enthusiastic participation. Thanks also to our livestream attendees — it's great to know you're participating from wherever you are. And for those of you who missed the live action, stay tuned — we'll start posting PuppetConf 2016 videos soon.
Aliza Earnshaw is the editorial director at Puppet.
- There's been a lot of talk, demo action and sharing here at PuppetConf around what's new in Puppet Enterprise 2016.4. If you haven't tried Puppet Enterprise yet, you can download the Learning VM for free, and start exploring.
- See Day 1 keynotes on YouTube: https://www.youtube.com/watch?v=7z03bZZqAyE