USENIX Update

Now that the LISA’12 Call for Papers is officially open, it’s time to start getting officially excited about our 2012 event, which will be held in San Diego from December 9-14.

In this interview, I asked LISA’12 program chair Carolyn Rowland to give us a super early look at what’s in store for LISA attendees this year. She and I have been talking about ways to help students and first-time attendees make it to LISA’12, so if you have ideas, please let us know. We met quite a few new people in Boston at LISA’11, and every one I spoke with was really enjoying their experience. Still, we want to make sure to pull up the welcome wagon and roll out the red carpet for newcomers. And, of course, we’re always happy to see familiar faces each year at LISA, too!

Rikki: Last year LISA had a DevOps focus. Any particular focus this year?

Carolyn: We want to continue with some DevOps content. LISA ’11 was our official recognition that DevOps is an important cultural shift for us as sys admins (or “Ops”). I think it is important to continue to include DevOps as a topic in the conference for people who don’t attend LISA every year and because it is still maturing as a movement, so I think there will be more to say about it.

As for other topics, the LISA audience is made up of a variety of specializations. The magic elixir for a successful technical conference is to build enough variety in the tracks to apply to a broad range of attendees. Pair these sessions with a few powerful keynotes and we have a great conference. So I would like to see a good mix of technical topics, which could include security, HPC, DevOps, VM management & cloud, networking, integration with Mac/Windows, automation & infrastructure management.

I would also like to see a sprinkling of professional topics — sometimes called soft topics, but that term implies that these are nice-to-haves instead of core skills for a professional systems person — that raise the level of the sys admin from just a techie to someone who understands how IT fits into the big picture for his/her organization.

Do I have a particular theme? No; however, I think there is something that separates LISA from the end-user conferences. A LISA attendee should walk away with a better understanding of the profession, a bigger circle of professional contacts, a greater awareness of how the building blocks of sys admin fit into the bigger picture of IT as it is viewed from outside our profession. I believe this separates the technical tinkerer or trades[wo]man from the career professional.

Rikki: USENIX events have a reputation for being particularly friendly and inviting to women and first-time attendees, and I met several first-timers last year. What would you say to someone who is considering attending LISA for the first time? What should they expect to get out of the event?

Carolyn: It is a firehose of information and there are a lot of people you don’t know. Get to know three new people. You’ll be amazed at how friendly we are and that three will turn into more because you’ll be introduced to greater professional circles from those three introductions. If you’re alone, make a point to go to lunch or dinner with someone you didn’t know before the conference. I used to introduce myself to the people sitting next to me in class and we’d pick up a conversation, which turned into lunch and dinner and more introductions in the hallway track. Don’t be intimidated if you see someone [in]famous from the profession. They are people, too, and they are interested in meeting new attendees.

For women, I know there is sometimes an additional challenge because there are fewer women at a LISA conference. I used to feel intimidated by this imbalance, but if I knew then what I know now… . Introduce yourself to people. If you feel more comfortable with other women, then start with a woman in the hallway track. I love to meet new people, so feel free to stop me in the hallway and say “Hey, I’m new to LISA — what do I need to know?”

Rikki: Are you planning anything new or different for LISA this year? What are your goals?

Carolyn: The submission process for the papers and experience reports track should be much more transparent than in previous years. The program committee is actively recruiting for submissions now. If you have an idea, contact me or anyone on the program committee — we’re all listed on the CFP.

The two things we’re doing differently:

1. Encouraging early submissions. This allows us to see your idea sooner so we can have some back-and-forth about what you can do to strengthen your submission or idea before the May 17th submission deadline.
2. We’ve introduced an author’s review and response period between the program committee reviews and the selection meeting. There is a week when authors can view the reviews for their submissions and respond to any misconceptions or questions from the program committee.

Goals? I’d love to sustain some of the energy of LISA ’12 after the conference has ended and everyone has returned to their homes. There’s a challenge to connect the LISA conferences together so each one isn’t a static event, but a continuation of previous conversations and discussions. I believe this can contribute to us moving forward instead of treading water technically and professionally. We don’t need to rehash the same issues and topics, but answer the questions we asked last year so we can generate new questions for the future.

Workshops are one of the ways we are successfully doing this. In the workshops, the conversation continues from year to year. Attendees bring their issues and experience with a specific topic and they work together to develop a path or a plan forward. It’s also great fun and you get to know a room full of people who share a similar interest.

Rikki: Anything you’d like to add?

Carolyn: Bring a friend! I’m still surprised at the number of professionals who have never been to LISA or never heard of LISA. We are obviously keeping this a well-guarded secret. The more the merrier. Share the reasons you go to LISA with your peers, officemates, and friends who don’t go. Maybe it’s the break from work to hang out with peers who “get it,” or the immersion into some new tech topics or the chance to hear what’s new in the industry — or all of the above. Let’s make this a grassroots effort to get more professionals because that will make the hallway track richer and the content will reach a wider audience.

Rikki: Thanks, Carolyn!

To see what you missed last year, be sure to read the blog archives or watch our videos on YouTube, including:

Surveillance or Security? The Risks Posed by New Wiretapping Technologies – by Susan Landau, Visiting Scholar, Department of Computer Science, Harvard University

Deployinator: Being Stupid to Be Smart – by Erik Kastner and John Goulah, Etsy, Inc.

An Introduction to Zenoss Core 4 – by Simon Jakesch

Fork Yeah! The Rise and Development of illumos – by Bryan M. Cantrill, Joyent

In 1993, Mark Burgess wrote the first version of the open source IT automation software CFEngine, and in 2008 he released CFEngine 3 and founded CFEngine AS, an Oslo-based software and services company. Last year, the company completed a Series A investment round and opened its first US office in Palo Alto, California. In this interview, he looks back at a productive 2011, and plans ahead for CFEngine in 2012.

Rikki: Your business really took off in 2011. Tell me about how the year started and what your big accomplishments were for the year.

Mark: Our year started with a Series A investment from a Norwegian fund, Ferd Capital and Haavard Nord, former CEO of Trolltech, the creator of Qt. That led to an incredible year where we grew from just a handful of full-time employees to 30 people with offices in Oslo, Norway, and Palo Alto, California. Our team has some incredibly smart and experienced people who are all bent on pushing the limits of innovation. We won a few major customers, including some highly complex environments on Wall Street, where the levels of scale and complexity have proven to be too much for the other technologies out there.

We released an upgrade to CFEngine 3 Nova, the commercial version of our CFEngine 3 Open Source technology — with a features like mission critical availability, virtualization management, and super lightweight performance.

In fact, to demonstrate the lightweight aspect, we brought demos with us to LISA ’11 showing the regular, out-of-the-box CFEngine agent running on an Android phone, and on an ARM Embedded device, typical of the kind of processor board you find in storage units. This opens exciting possibilities for CFEngine, as it means users can use one agile framework to manage and maintain systems as small as a phone, but as large as a mainframe or a cloud.

R: What’s new in 2012 with CFEngine? What are your goals for your business and for CFEngine?

M: As a company, we’re a bit spoiled for choice. The list is long, but we have an interesting project around the cloud that’s coming, and I feel as though I have a lot to do explaining to the industry why CFEngine’s modern approach to configuration is really the way to invest for the decade, given the transformations in IT infrastructure that are just around the corner. Taking care of individuality needs should go without saying today, not just mass producing everything as identical clones as we did in the past.

There are many new things in CFEngine 3, both in the open source core and in the commercial product. In the core, we have a simplified interface to services, a bunch of new functions and of course lots of new standard library methods including STIG compliance packs and cloud service packs, etc.

On the commercial side of things, we’ve focused a lot on designing a simple user interaction with the CFEngine `dashboard,’which we call the Mission Portal. We’ve also been pushing the limits of scalability both in simple star-net architectures, and laying plans for more advanced self-healing architectures that can scale to hundreds of thousands of hosts. Project Constellation is a way of connecting star clusters into a unified global picture. We put a lot of effort into making scalable reporting and monitoring that will either stand alone, or tie into the usual ecosystem of system software.

We also think in terms of long-lasting solutions to scale and complexity, and our “focus for the decade” is Knowledge Management, which I believe is at the heart of managing infrastructure in the modern world.

R: At the end of the year, what do you hope to have accomplished?

M: I don’t want to say too much about that, but we are working on the next-generation software that will set the industry on the right path for massive hybrid infrastructures, supporting any necessary level of complexity, without things getting out of control. We are so lucky to have some of the smartest minds working on our future plans (wink). Something exciting is coming.

Of course, now that we have a trans-Atlantic team, I hope to really make good on the promise of CFEngine. We want to help sys admins transform their traditional role from being reactive to agile and proactive — to become valued infrastructure engineers!

R: Why is CFEngine the right choice for someone over other options, such as Puppet and Chef?

M: I think you choose CFEngine if you want a solution that will last you the decade, and will not hold you back once you’ve gone beyond the initial installation state. You have to be willing to invest a little in training yourself to use the tool. I think a good analogy might be that Chef and Puppet make it easy to get started for a certain segment of users in the cloud — a bit like a microwave dinner. Just push the button and go. But the moment you want to do some great cooking, or scale up to a large meal, you need to be able to manipulate the raw ingredients beyond the level of packages. CFEngine allows users to mainpulate the raw ingredients with arbitrary precision. That will be the difference for businesses who need agile customization to remain competitive.

R: Which events will you be at this year? And what about training at events — what will you offer?

M: I will personally be giving the introduction to our CFEngine 3 training course in the Bay Area at the end of January. We are also going to be at SCALE 10x in Los Angeles, OSDC in Germany, FOSDEM in Belgium, Velocity, and of course we always attend USENIX LISA.

We hope to attend a lot more local sysadmin user groups and meetups, too, and we’ll be presenting a lot more webinars, training, and whitepapers on helping people to migrate from clunky CFEngine 2 to super-slick CFEngine 3.

R: What else should our esteemed readers know about CFEngine?

M: A lot of users aren’t aware just how much of an upgrade from the old CFEngine 2 our new CFEngine 3 is. I spent five years researching how to really solve the issues of configuration management. It was a lot of work and planning to get the model just right, and I really believe it paid off.

There are other products based on CFEngine 3 now, too. Rudder is a nice front-end for CFEngine 3 made by Normation, a French company, which has taken the approach of making an elegant and easy wrapper to help people get started. Once you are started, you can grow and use the full power of either the open source or the commercial versions.

And the fact that we can run on just about anything makes management of systems very homogeneous without having to make the hosts themselves homogeneous.

Watch Mark’s LISA ’11 talk on YouTube: 3 Myths and 3 Challenges to Bring System Administration out of the Dark Ages

Also, be sure to stop by the USENIX booth at SCALE10x later this month!

One of the options during each of the technical sessions is called “The Guru Is In.” These are informal sessions with an expert in a particular topic relevant to sysadmins. Guru sessions are a great opportunity to discuss ongoing issues that you’re actually experiencing. The best part is that there’s not just one guru — the other attendees are gurus as well. Those who attend to ask questions often find themselves answering the questions of others. While training sessions and the other technical sessions provide great information, the guru sessions are uniquely able to address your own specific bothers.

On Thursday, I had the pleasure of attending two excellent guru sessions. The first, starring Oracle’s Janice Gelb, focused on documentation. Her opening remarks included comments about the weaknesses of wikis as documentation repositories. This reinforces comments made by Mark Burgess in his Tuesday morning training: “wikis are where knowledge goes to die.” Janice suggests that wikis be used for interactive work, but that more authoritative documentation be maintained.

In order to keep documentation up-to-date, a culture of documentation must be fostered. This requires scheduling regular updates and maintenance of documentation. Management must also be convinced of the value of maintaining documentation.

In the afternoon, Tom Limoncelli held a session on time management. Tom, of course, is a leading expert in the field, having published a book and presented several trainings at LISA conferences. Sysadmins, being chronically short of time, packed the room to learn and discuss.

Tom began with five keys to managing time:

• Create a mutual interruption shield
• Turn chaos into routines
• Record all requests, don’t rely on your brain
• Make 365 to-do lists a year
• Document the procedures you hate

Unlike the earlier guru, Tom advocated wikis to document tasks and procedures. Checklist-type procedures and infrequent, error-prone tasks are especially suitable to be documented. This task list can serve as a list of tasks to delegate to junior admins and can be a guide for writing job descriptions.

Tom advocates making a to-do list for every day of the year. Any unfinished items should be carried forward to the next day. A large number of applications exist to help with this, although a simple text editor or even pen-and-paper are well-suited for to-do list management.

Managers can make it easier for their staff to by setting three policies:

• How to get help (hey, file a ticket!)
• Written definition of what constitutes an emergency.
• Scope of support

Kris Buytaert finished the day Wednesday on the DevOps track with his presentation: “Devops, the future is here, it’s just not evenly distributed yet.” (slides). It was a great day with several awesome presentations on various DevOps related topics. In this presentation, Kris went over the short history of the devops movement and pointed some of the key moments, starting with the first public event where he was involved with Patrick Debois and some other like minded people in organizing the first DevOpsDays event in Gent, Belgium in October 2009, until today when DevOps is the theme of LISA11 the biggest and most important system administrator related conference. So many things have changed during this time and right now DevOps is considered the one thing in system administration that has the most energy and most passion.

In the first part of the presentation, Kris explained how we got here in the first place and why this movement is so important: breaking the silos and bringing together the developers and operations on the same team, working together for a unified goal, and achieving real value for our company.

Devops is about CAMS:

• Culture: people and processes first
• Automation: tools for configuration management, orchestration, etc.
• Measurement: if you can’t measure it you can’t improve it.
• Sharing: create a culture for people to share ideas.

Some of the practical advices from his presentation included topics like:

• use version control
• continuous integration
• automated deployments
• use FPM for packaging
• configuration management
• orchestration
• high availability, scalability
• monitoring

Overall this was a great presentation about the history of devops movement, the current state and the challenges that we are still having in the community. You can find more from Kris on his blog: Everything Is a Freaking DNS Problem.