Panel: How to Enable Large-Scale Collaboration
Companies are increasingly turning to collaborative software development to build their products and services and speed innovation, keynote presenters at Collaboration Summit told us this morning. But how does this process actually happen? Open source directors from Intel, Citrix and the OpenDaylight Foundation shared some of their secrets of collaborative development in an afternoon panel discussion, moderated by Linux Foundation Executive Director Jim Zemlin. Below is an edited version of the conversation, which covers the rise of open source foundations, how to attract top engineering talent, how to manage open source developers, and more.
Jim Zemlin: How would you describe what you do?
Neela Jacques, OpenDaylight Executive Director : I try to get an industry that's bitterly competitive to work in a way that's completely different from the last 20 years. To set aside some resources and work together on something without a direct ROI. I also work closely with developers to tackle the biggest problems in the industry and release great code.
Mark Hinkle, Director of Open Source Solutions at Citrix : I run Citrix's open source business office. We develop ecosytems around open source projects that benefit our business interests, including Xen Project, OpenDaylight and Apache Cloudstack. We also do user enablement and developer relations, and educate partners on how to work with open source communities.
Mauri Whalen, Core System Software Director at Intel: I manage a team within Intel that's part of the Open Source Technology Center. We make sure Linux-based operating systems run best on Intel architecture.
Zemlin: How do you manage developers who are influential in their communities? How do you decide how they spend their time and define the boundary between company and the project?
Mauri Whalen: It's the business first. I look at Intel's road map 12-18 months out and make sure Linux runs best on it. Past that you need to make sure open source developers have a lot of creative freedom. So we do have some flexible time for people to work on their own projects. It also involves a lot of open communication.
Neela Jacques: It doesn't work to be a boss and tell other people what to do. It's about being able to enroll people into a vision – listen to the community and try to tease out the common views. You don't manage these folks – you show the challenges that exist, the possible solutions and facilitate people working together.
Mark Hinkle: Understanding your end game is critically important. Sometimes it may be to facilitate shared collaboration across companies. In that case governance and bylaws are importantso it's vendor netural and they can collaborate. Others are aimed at user adoption. In that case you want to allow people with a single voice to make a difference, like plugins in a Firefox or Chrome browser, you have to allow all the discreet elements to collectively make the browser really powerful.
Zemlin: Some people have been critical of the recent rise of these open source foundations. Is open source a groundswell that happens from the bottom up, or is it a grand scale collaboration between large scale organizations?
Mark Hinkle: I do think there's a rise of the vanity foundation and the reason it happens is it helps to have name recognition. The Linux Foundation has a strong success rate in Linux, but when you get to OpenDaylight, it's hard to see the connection between Linux and this open source collaboration. That is a problem for awareness. If a developer drops code in the world and nobody's there to see it, does it make a difference? Other than that singular reason I don't think it's necessary.
Mauri Whalen: You have to take advantage of when developers have good ideas to keep creativity and innovation going and brew that. You have to figure out if you want to try to make a business out of them and put Intel's name behind them. Having foundations and specifications and standard APIs has it's place and time, but allowing the creative minds to come out is still worthwhile.
Neela Jacques: The real question is : Can you boy-band an open source project? No, I don't think you can. At the heart of every open source project is a singer songwriter. What you can do as a foundation is to get the songwriters together. There's a support system to get it recognized. A set of smart executives can see a trend in the industry but they recognize that for something like OpenDaylight to work it can't be controlled, it can be only be supported. You have to build frameworks and structure to enable grassroots efforts to thrive.
Mark Hinkle: It depends on the situation. Sometimes you're looking to facilitate collaboration across partners, sometimes it's to facilitate lower operation and deleopment costs, sometimes because you want to drive massive adoption and create an opporunity to support them in a commercial way.
Zemlin: What's the most important element in enabling mass collaboration? Is it the bylaws, the tech development structure?
Mark Hinkle: It's all about the ability for distributed contribution to the project. If you look at the Linux kernel, the actual kernel development happens among a small percentage of the users. Core development requires a high degree of coordination that doesn't scale for mass collaboration. Mass collaboration allows collective intelligence to reign.
Mauri Whalen : There are different levels of documentation and bylaws with each project. Take Tizen for example, we're primarily collaborating with Suamsung. They're both a competitor and a partner so we need a neutral space to do these things. But it's like most projects internally, you need a high degree of transparnecy and openness.
Neela Jacques: It's three things: governance, culture and sweat equity. Governance helps answer questions with competitors. Culture is the trust that builds up between individuals where you don't even ask the questions even more. And with sweat equity people have to be willing to help.
Zemlin: In every key sector of tech, whether it's the Internet of Things or SDN or the cloud, big foundations are coming together and investing resources in building a huge new technology. There is a boy band factor to that in that the project is manufactured to answer a problem. The question is, who will be our Linus Torvalds? Which model works better - the benevolent dictator or the open board structure?
Mark Hinkle: It needs to be someone who is loving and touchy feely and soft spoken like Linus Torvalds. No seriously, it has to be someone who is solving a problem that's very personal to them and has massive impact in their life. It has to be someone with passion in it. There are boy band open source projects – that' happening prominently these days. It has to be driven by people who have a personal skin in the game, with a reason to do this who can be as tenacious as a rabbid penguin to do this.
Zemlin: You would argue Linus is a unique guy. What if you can't find these people or to get started you have a technical steering committee?
Mark Hinkle: You can still have that, but maybe there's a group of people who have a passion around that. Brett Salisbury is one of the reasons I think OpenDaylight will be successful, even though there is a foundation. You don't need the foundation but you need those people.
Zemlin: How do you find these people who can be passionate and technically smart and successful?
Mauri Whalen: We go to a lot of conferences and talk to people. It's also keeping things challenging and interesting to people and not putting people in a box. You need strong visionaries in the wild west. Once you start building that type of culture people gravitate toward it. They like to work in that environment.
Zemlin: How is it going for lack of Linus Torvalds in OpenDaylight? You have a technical steering committee.
Neela Jacques: It works just like many organizations. Every once in a while you ave a visionary leader that's perfect for the task. But when you don't have that you get multiple individuals who step up. It works because it's a group of folks who believe in the vision that we can change how networking happens.
Zemlin: Since we can't name who the next Linus Torvalds will be, can we name what the next big mass collaboration will be?
Mark Hinkle : I think it's the class of projects : 3D printing and prosthetics. The idea that you can democratize a low-risk medical device that has a huge return on quality of life is what the next non-IT related collaboration will be.
Mauri Whalen: I go back and forth because the wearables and internet of things is big. But going back to the cloud and datacenter I think there's still tons of room for innovation.
Neela Jacques: I'm bullish on OVS and Docker, but I'm really intrigued by Cloud Foundry. Developers need a good PaaS that saves them time.