The three new Puppet products based on Distelli’s technology are Puppet Pipelines for Apps, which automates key application development and delivery tasks; Puppet Pipelines for Containers, which enables users to build Docker images from a repository and deploy them to Kubernetes clusters; and Puppet Container Registry, which gives developers a comprehensive view of their Docker images across all repositories.
When we’re talking with customers about the value that Puppet brings to them, invariably we talk about the future, and the future in their mind in some ways includes containers. There’s a lot experimentation going on. There’s a lot of Docker work being done and container work being done, Kubernetes work being done on their laptops. The conversations we have with them is how does Puppet help you bring it into production, into mission-critical production? How do you keep it secure? How do you operate it? All of those things that we know how to do and have done with various kinds of infrastructure, whether it was OpenStack, whether it was virtual machines, whether it’s just server configuration. For us, we take the same approach to containers and are evolving our road maps to make sure that customers have the same benefits they’ve have had over the years now with containers or other technology.
From an interview with Puppets CEO, Sanjay Mirchandani.
Growth, in terms of customers and revenue, appears to be significant. About 34,000 companies use Puppet’s software to automate data center operations, with 6,000 added last year, Kanies said. The company is approaching $100 million in annual sales and is expanding its headquarters in Portland, Ore. In January, it secured $22 million in financing from Silicon Valley Bank to fuel further expansion.
And then a stray container momentum/penetration number from Forrester:
Stroud sees promise in Puppet’s container support, despite his observation that only about 8% of companies are using containers for production workloads. “It gives current Puppetized work the ability to be containerized,” he said.
“Puppet alone provides the basis for the majority of OpenStack deployments, at 56 percent. Puppet plus deployment tools that utilize Puppet (e.g., Mirantis Fuel, Red Hat OpenStack Platform, PackStack) provide the basis for 72 percent of all production deployments.”
So we’ve built some first-generation integration between Chef and BladeLogic 8.5, which we’re demoing in our booth for the first time here at ChefConf. You can use BladeLogic to call Chef cookbooks and recipes on a push/scheduled basis, and you can reference BladeLogic compliance policies from inside your Chef cookbooks. It’s all very early and not production-ready, but we want to put this integration front and center with the people here at ChefConf and start a conversation about how they want to blend these two approaches to a stable, managed IT infrastructure.
BladeLogic plays an interesting role in the history of the Puppet/Chef/etc. automation world. As I recall, Puppet’s founder Luke Kanies worked on Blade for a short while and, you know, was interested in a better way, which eventually led to Puppet. Also, for those who like startup culture books, Blade was the chief rival of Opsware, where many of the stories in The Hard Thing About Hard Things come from.
Secondly, when evaluating new IT hardware and software assets for potential adoption, you need to institute a much stronger requirement for programmability and open APIs. Complete automation of your infrastructure requires programmatic access, and it’s simply insufficient to only have control via graphical interfaces. This isn’t just about provisioning and configuration support via such APIs, you also need to ensure that vendors are providing reliable APIs to get sufficiently detailed status. A core tenet of DevOps is the ability to measure the state of your infrastructure for future improvement and this really needs to be automated programmatically. Ideally these APIs for automation and measurement are simple, easy to adopt, and accessible to people who aren’t full-time software engineers. Thus, beware of complex, language-specific APIs, and strongly lean towards vendors using simple HTTP or REST APIs and standard, easily parsable data formats like JSON.
This is the private cloud in a box that Dell doesn’t call a “private cloud.” It’s converged infrastructure. Adding in Puppet is great and certainly makes it feel more cloudy.
Dell has spent a long time, with several acquistions, trying to get ActiveSystems just right. As TPM sums it up:
Active System Manager is largely based on a sophisticated set of infrastructure management tools that Dell took over and rebranded after it acquired Gale Technologies in November 2012 for an undisclosed sum. Dell had a mix of systems and cloud management tools, including those it got when it acquired Scalent and licensed from DynamicOps (which VMware bought in July 2012), for its infrastructure management.
To a limited extent, there’s all the assets in Quest and then adjacent is Enstratius.
The two questions for Dell on automation are always around Crowbar and Gale:
“We have been following Chef and Puppet in the industry for a long time,” Ganesh Padmanabhan, director of products for Dell’s converged infrastructure solutions, tells EnterpriseTech. “With Crowbar, the exercise at that time was to make a bet and see how far it goes. We are going to keep that activity alive, Crowbar is still massively important for our larger scale and Web 2.0 customers. But we are seeing more traction on the Puppet side, so this becomes a more strategic bet for us.”
And on Gale:
Going forward, new configuration and control templates for Dell machinery will be created in Puppet, and the vast trove of templates for third party gear will be available to customers using Active System Manager.
Overall, it sounds good. It does make Dell a Chef and Puppet house, which is always an architectural quandry to sort out internally.
All of these mega-vendors are going after the soup-to-nuts cloud transformation portfolio: they have to, even if it’s highly undifferentiated. Most recently, I took a stab at writing up HP’s approach and strategy (you can extract the general strategy if your pattern recognition reverse engineering skills are up to snuff), available for 451 subscribers or those who care to lead-gen themselves for a 30 day trial.
I’ve been in Chef Fundamentals training this week. It’s a good course. Chef is a lot more ruby than I’d realized. It’s pretty close to what a programmer would do if they were forced to manage machines. Of course, programmers also came up with Ant and Maven, so we have that. All this will be a good baseline for chomping through the successors of Puppet and Chef, the SaltStacks and Ansibles who seem to be taking an even more simplified approach.