How Puppet pulls the strings when it comes to technology automation

Developers and SysOps take heart, Puppet Labs and CEO Luke Kanies wants to take the tedium out of your lives so you have more time for the good stuff.
Written by Colin Barker, Contributor on
Puppet Labs' CEO Kanies: If IT isn't looking at automation it should be.
Photo: Puppet

As our world becomes ever more complex so do the systems that run so much of it. One way to deal with that is to automate, and to do that, you first need to break down complex components into simpler units.

While this might work for tasks like building cars, it is quite another to do it with tasks like writing, implementing, running and de-bugging computer software and systems.

One of the exponents of coding automation is Puppet Labs. Based in Portland, Oregon and with a European Office in London, the company offers a configuration management system. Using this you can first define the state of your IT infrastructure (in whole or in part) and then Puppet will automatically enforce that correct state.

Puppet's aim is to free-up the system administrators' time by doing the routine stuff for them, including updates and upgrades. It is an innovation that is proving very popular as the company is growing in leaps and bounds.

On Tuesday the company launched Puppet Enterprise 3.8 which, the company says, supports containers and cloud environments and there is also support for Docker.

Puppet has also introduced a module for Amazon Web Services (AWS) that is intended to maje it easier to provision, configure and manage AWS resources, including EC2, Elastic Load Balancing, Auto Scaling, Virtual Private Cloud, Security Groups and Route53.

Razor, Puppet Labs' bare-metal provisioning capability, has moved from Technical Preview into full release mode with the 3.8 release. "Razor automatically discovers bare-metal hardware and dynamically installs and configures Windows, Linux and VMware ESXi operating systems and hypervisors," the company says.

Puppet Code Manager is a new Puppet App designed to help users manage the code that defines their infrastructure. The first capabilities it will have are based on the r10k technology, which is intended to improve the reliability of code. It provides a consistent way to change, review, test and promote the Puppet code, the company says.

In the space of its ten year existence - it celebrates that anniversary this month - Puppet has grown from being a tiny company to one that now employs 360. Two years ago Business Insider listed it as one of the 25 enterprise startups you can bet your career on and the company, along with its charismatic CEO Luke Kanies, has so far justified that sort of confidence.

We caught up with Kanies to talk to him about his company and about its roll in the fast growing world of DevOps.

ZDNet: How did you get into the IT automation field?

I started the company about ten years ago and for the first three or four years the company was just me. Then I got our first round of funding in 2009 and since then we have grown from three people to now about 360 and it's been a good ride.

Now you look at the community size and we have about 25,000 companies who are using our products. We have about 3,000 models on our "porch" where people publish things that they are doing with our products.

We have been between 80 and 100 percent growth for what seems like forever.

Now I think what is most interesting is that when you look at the companies that existed when we were formed they were companies like OpsWare [a HP company now] but the penetration of the market for automation tools was, and still is, about 15 percent.

So, on the one hand we have a company that is doing really well and on the other there is so much up-side and so much interesting work available. To do that [work] I think of it like this: the last ten years was good, the next ten years is going to be amazing.

In terms of technology, there are a couple of things that we do differently. The first is that when they talk about us, companies tend to talk about "model-driven" so with Puppet you describe the state you want your infrastructure to be in and then Puppet's job is to get it into that state. We have our own custom language and that gives a number of benefits.

One is that you can do a level of analysis with that language that we believe is above and beyond what you can do with anybody else's. You can get much better security. You can get a much better ability to fix questions like: "Is this new version going to be compatible with this old version because I don't want to upgrade 4,000 versions before I confirm that".

You can also do the security with a very high confidence rating very quickly. Effectively you can get answers very quickly without taking higher risks.

Q: How do you see the DevOps market panning out for you?

I met with four or five companies yesterday and what I am seeing is that the percentage with operations that are working as efficiently as they could do is pretty small.

One vital point is that as an industry we have to get to a point where our ability to deliver and manage software in an organisation is seen as a key function. Doing it well should be seen as a key achievement of an organisation and doing it poorly significantly degrades the quality of an organisation.

DevOps is really all about getting organisations to recognise that they are a technology company. Yes, they will be other things as well. You may be a media company, or a retail company, but you have to be great at technology and in the course of becoming great at technology you have to switch to a different way of looking at the world.

So we are on the way to ensuring that the concepts behind DevOpps are as well understood by as wide an audience as possible. For our part we need to understand why our technology and our services are valuable to our customers.

Q: A sizeable chunk of this for any organisation has got to be, how do they handle people, doesn't it?

Yes it is true that the people part and the culture part is the area that is often most important but is also the area that is least understood. How do they handle it?

You know it is not as if you can ask a company, "Hey, do you do culture consulting?" When it comes to the technology and so on they know who to call on, but culture? And it's the critical piece.

You can get everything else in place but if your people don't work together, if they don't talk and if people don't believe that, "Hey, this a critical part of success for me", the rest of it just won't work.

But really it is the old definition of leadership. You have to have someone who cares deeply about the project. Someone who is willing to spend a ton of time making sure that it gets done right.

Then again, it is all about change and you have to drive through that change and we know that the only way to do that is that you have to turn up every day and drive that change through. And that is something most organizations aren't that great at doing.

Q: Doesn't that mean that DevOps isn't just a CIO issue, but a CEO issue?

Well, in my opinion, it had better be because the one thing a CEO can do is execute on it. You know it is one thing to say that it is just a CIO issue. Look, if you're a car company you can bet that the COO knows more about what cars are being made, what process are going into it and so on. But the CEO better know what the constraints on the business are and what the high level aspects are for the company.

So given that every company is becoming a technology company, then the CEO better have a handle on the technology, what parts of the technology are really an asset and are driving up the value of the company and what parts of the technology are holding you back.

You [have to] know that what may not be working so well now, is going to become a liability pretty quickly over the next couple of years.

And you see the CIO may not have what's needed. You know the CIO may have control of the development team but not the operations team. So if you are going to ask the CIO to fix all the problems in the IT organization you are never going to succeed because the only person qualified to fix things across an organisation is the CEO.

Q: Do you have any examples of companies that are doing DevOps well?

PayPal is a company that is close to this. They went from it taking six weeks to progress software from the stage where it is ready to go, to getting it into production. Now, they have that down to 30 minutes.

When you make that change, it is a pretty massive change because it is not just that you can do it faster. Because you can do it faster you can do it more often. As a result your entire organization changes the way it does things.

At the very top of the scale, look at the way that banks have changed. The mobile functionality you get now so that you can do all your banking on your phone.

If you think about it, that has really happened quickly with banks and the only way you can get that to happen is if you have really tight co-operation between all of the teams. If you are a bank now you are a technology organisation to.

Further Reading:

Editorial standards