Interview with Dirk Hohndel – Intel’s Chief Linux and Open Source Technologist

In this interview, we talk with Intel's Dirk Hohndel about Moblin, the explosion
of devices, and how large companies do (or don't) really "get" open source.

Sean Campbell: Dirk, give us a little bit of background on you, your role at Intel, and your experiences with open source.

Dirk Hohndel: I’m the chief Linux and open source technologist at Intel. I work in the Open Source Technology Center, which is part of the Software and Services Group. I’ve been doing free software and open source for about 20 years.

I started while I was at university, when I had the incredible fortune of getting involved in Linux at the end of 1991, with version 0.11. That made me one of the very early people who started contributing and getting involved.

I have been around free software and open source ever since, through a number of jobs. I have run a few open source projects either myself or together with other people, and I’ve been involved in quite a few other open source projects over the years.

I was CTO of SuSE from 1999 to 2001, where I started and ran their professional services business. I co-authored the first ever published book on Linux, and I’ve written a bunch of articles.

At Intel, my role is very much about making sure that Intel internally really understands open source and acts in a way that is compatible with the community. My charge is to help the company be a positive member and contributor in this space. At the same time, I’m trying to make sure that the community understands what we’re doing and where we’re going.

Sean: What can you tell us about the current state of Moblin, since it has moved under the auspices of the Linux Foundation? Is the focus likely to be moving away from Intel’s original vision?

Dirk: Moblin was conceived inside Intel, as a project that would bring a different user experience, to use the term very broadly. We identified the issue that there were no OSs that were focused on being connected to the Internet and using the services that are out there, such as social networks and so on.

We saw the need for an OS that could achieve that in a very natural way, hiding all the complexities from you, without being too restrictive. We wanted a user experience that is intuitive, easy to use, and visually interesting, while at the same time retaining the freedom that a Linux-based OS gives you.

It’s actually been almost three years since we began this effort now. We’ve gone through a number of iterations of how it should look and be implemented, and we eventually came to the conclusion that we really had to roll our own OS from scratch.

We felt that a very large proportion of the underbelly and infrastructure needed to be optimized for this use case and type of device to be really successful. We recognized that we couldn’t just create it as a patch kit on top of an existing Linux flavor; we really needed to develop it from the bottom up to focus on this environment.

Sean: How did the transition from Intel to the Linux Foundation come about?

Dirk: One of the things that happened is that we ran into a lot of people who were hesitant to contribute to Moblin because they saw it as Intel’s project. Even though they understood that the project was very important to us, they had concerns about what would happen if Intel lost interest.

We realized that the solution to that perception issue was actually pretty simple. From our perspective, Moblin was always an open source project, and we had planned from the beginning not to make it an Intel thing. To communicate that to the community, it seemed wise to bring it under the auspices of a neutral party that is beyond any doubt in the Linux community.

The Linux Foundation was a great choice, since it has a wonderful track record as being neutral and having the best interest of the Linux and open source community in mind.

It’s odd that this move created some fear in the market that we would disinvest; it’s actually the opposite. Over the last six months, we’ve increased our investment and put even more resources toward targeting different sub markets. So we’re doing a lot of work in this space, and I think we’re pretty close to having the first product based on Moblin in the market.

Sean: What is the status of Moblin 2 at the moment?

Dirk: I’m sure you’ll understand that I can’t disclose any of those plans–the marketing people would not find that funny–but I think it’s fairly close. We are currently releasing weekly or bi-weekly updates to our latest beta on moblin.org, so people can play with it. We are very much in a code freeze space, so we are not taking any new features right now. We’re just fixing the bugs that we see.

In parallel with that, we have actually started development on the next release, because of course, software is never done. You always have the next 20 things that you absolutely want to integrate, but we are freezing and serializing Moblin 2 right now.

I’m very excited, and I’m running the latest beta build every day. I think it’s a very impressive user experience, and it certainly fulfills our goal of making it easy and intuitive to do what we perceived and what our early testers perceived as a typical task that you would want to do with a netbook.

I should point out that the current version of Moblin 2 is focused on netbooks. We would make different decisions if we were targeting full-fledged desktops or touch-based devices like tablets.

Sean: If you consider portable devices that range in capability from maybe a smart phone at the low end to a netbook at the higher end, with others in between, like maybe the Kindle, how does Linux serve that full range?

Does it scale up to laptops, or do you think it has a sweet spot on that scale that maybe is the best place for it?

Dirk: If I put on my aluminum foil hat for a moment, of course, it’s the best operating system for anything and everything you could ever dream of. [laughter]

Having said that, and taking off my aluminum foil hat again, I think the market has been very clear over the last few years, in terms of where the majority of end users do and do not prefer it over other OS’s.

This is my personal assessment, because I don’t think there is a conclusive study on this anywhere, but my personal feeling is that for a general purpose PC, where users assume they will have a full office suite, image processing, and integrated UI, Linux has a much harder time succeeding in the market. That is the space where Microsoft Windows is the market leader, and Mac OS is a strong contender.

But as soon as you specialize things more, and as soon as you start using the term “devices” instead of “computers,” you see Linux-based OSs being much more successful. The Kindle, for example, runs Linux. It’s an extremely specialized device. Anybody who has tried to use the embarrassing attempt at a browser on the Kindle knows that this is not, by any stretch of the imagination, a general-purpose device.

Sean: Oh yes. I used that once. [laughs] That will probably be the last time.

Dirk: Consider things that people often don’t think of as computers. Think of your wireless access point, your cable modem, your set-top box, and your DVR. Those are very lightweight connected devices, and they almost certainly run Linux. In the area of cell phones, Android is obviously a hot topic right now, and it is also Linux-based.

That means that a key question around the netbook concerns whether it is a general purpose computer. To some people it is, and so you see Microsoft Windows running on more and more netbooks. To others of us, the netbook is a special-purpose device designed with the goal of being very affordable yet very robust, as well as easy to carry around. It targets surfing the web, connecting with friends, and social media networks. You might use a netbook for playing some media content, but not for general computing, and that is the space that we envision Moblin for.

Moblin fulfills a fixed set of purposes, and it does so very well, with a great user experience. It is distinct from the full PC model, with the goals of being light, easy, visually appealing, and fast. That distinction is why I think Linux makes a lot of sense in this space, because of how configurable and powerful it is, and because of how much you’re able to tailor it for your purposes.

Sean: On my desk right now, I have a MacBook, a Windows 7 T400S, an iPhone, and a Kindle. Obviously, that’s different from the average person whose computing equipment may consist of a phone and something bigger than that. There are limits to how much a person wants to carry around.

Do you see that threshold as being at maybe three or four devices? Maybe people want, on average, a smart phone, a netbook or other specialized computing device, and then a higher powered machine for productivity apps, editing photos, or whatever. How do you see those limitations shaping the future of a project like Moblin?

Dirk: I do not speak for Intel at all when it comes to big picture vision questions like this, but I can give you my personal opinion.

I think there is a continuum between the tiny phone through the bigger phone, the smart phone, the really scary big smart phone, the MID, the tiny netbook, the slightly bigger more powerful netbook, then the consumer thin and light CULV kind of devices, the notebook, the desktop replacement, the desktop, the gamer machine, and the Mac Pro workstation with two Xeons and 12GB of memory.

Where people draw the line in terms of which and how many devices they pick on this continuum is very much dependent on who you’re talking to. I have that eight-core Mac Pro, a Core2 Duo-based laptop, an Intel Atom-based 10-inch netbook, and a Celeron-based seven-inch netbook. I have a couple of Android phones and I have a Motorola Razr. There are seven or eight devices on that continuum, and even though I hopefully never have more than three of them with me at a time, some people would find that lineup to be absurd.

That’s the point, though–people make different choices in this area. I think that, as a vendor, it is important to embrace the fact that people like choices and to provide them alternatives that let them find the devices they like best.

For example, when Asus first started talking about the Eee PC two years ago, a lot of people told them that no one wants a diminutive laptop. Guess what? People were wrong, so I’m very happy that we invested in this and thought it was a great idea to provide the platform to do it on.

I don’t think it is our job to tell the end user which device is for what and how to use it. It turns out that they usually have their own ideas, and our job is to facilitate that.

Sean: Where do you see data synchronization playing into that vision? Sugar Sync now supports Android, and some additional solutions like Ubuntu One are trying to build out a sync story. But if a user really wants to live in a world where they can sync all their devices, they have some obstacles.

For example, a user might have a Windows or a Mac box, a mid-tier device like a netbook, and a phone. If they want even something as simple as a picture to show up on all of those devices, and to have an edit made on one show up on all of them, they are probably going to be frustrated.

What are either Intel’s or your personal thoughts on that, or what do you think the community is doing that’s interesting around that issue?

Dirk: This is a really hot topic, although I don’t think synchronization today is so much bound to your OS, even though there are a couple of providers who have a very jail-focused mentality in this space. They basically refuse to interoperate with devices that haven’t been made by them, which I find to be a very weird approach. I think that long term, that’s not going to be successful, but those companies are in the minority.

We do see that there are a number of services out there that are by design cross-OS, and by design simply a way to store and share data. I think these are the services that will define how we interoperate between our devices.

For example, and without any endorsement in this context, the way I store my personal contacts is that I have a central server that I connect to all of my devices. Whether it’s the big Mac, the Linux-based laptop, or my cell phone, they all get the contact information from one single place, and for me right now, that’s Google Contacts.

There are other services that would allow me to do the same thing, like you can use Funambol and create your own server. I think we’re going to see a number of different offerings tailored toward different user groups that provide the ability to access data in all of its forms across devices.

One consideration is that you have to have the ability to do this online and offline. If you don’t have the ability to do local work with your local data, you’re going to fail. We still have a terrifyingly bad existing wireless data infrastructure, by and large.

Here in the US, we pay among the highest rates for all wireless data service, and we have disappointingly spotty coverage in most places. If you don’t have the good fortune to live in Portland where you have WIMAX fast wireless wherever you go, you turn a corner and you’re back to GPRS, or to nothing as soon as you leave the metropolitan area.

Your application and data need to live on the device, and they need to sync with a central entity if and when they’re connected.

Sean: I agree that there are different solutions for those of us who are technically inclined, but would you also agree that right now, no one is fully solving that problem because there is a somewhat jailed model where some of the sync platforms exist?

In other words, if you have a Linux-based mid-range computing device, a higher end Mac device, and a phone from another provider, there is no proverbial grandma solution out there.

Dirk: I think my friends at Google will disagree with you, but I understand your concern. There isn’t a solution that is totally and seamlessly integrated out of the box with all of the different OSs, and market forces make that a pretty difficult problem.

It certainly appears that there are some tensions between the big players. Without commenting on any of these situations, anybody who has read the press in the last couple of months knows that there are some disagreements between, say, Microsoft, Google, and Apple over who controls what in this world.

Scott Swigart: I’d like to touch back on something from earlier in the conversation, which had to do with integrating open source thinking into a large company’s way of doing business. Of course, this isn’t at all unique to Intel.

Microsoft is trying to raise its open source IQ, Sun Microsystems worked hard at it for a number of years, and IBM has figured out to some degree where open source fits into its operations. Oracle is going to have a bit to digest around that, because they’ve brought on a lot of open source assets to a company that traditionally hasn’t had that.

In general, what have you found to be some of the steps along the path of a software company successfully embracing open source?

Dirk: I think there is a common problem that over the years, large companies get into patterns of behavior where assumptions are made about how problems can be solved, how interactions with the outside should happen, and so on.

All of that dictates cultures that range from an open approach where you publish your program manuals and you invite developers to join you, all the way to the secretive approach where people need to sign NDAs even to be allowed to see the contract that they have to sign, in order to be allowed to develop software using another NDA that is only available under NDA.

There are different approaches, and most software companies come to open source as a result of the value they can get from having open source components as part of their product or to use in conjunction with their product.

Very often, what happens is that the people who make the decision to go with the open source components look at this and say, “So what does it mean for us? How does this change the way we act?” The organization below them, meanwhile, tends to be very resistant to change.

Scott: How does a company address that successfully?

Dirk: The challenge is that if you want to gain a commercial advantage using open source components, just as with everything else that you do, you have to go beyond just following the copyrights, the licenses, the laws, and the rules. You have to understand them not only on a legal level, but also on a conceptual level that can inform you about how to successfully interact with the people involved.

Say you want to interact with Microsoft. That means you have to follow the license that Microsoft gives you and the rules that they put around their SDK. If you want to have a successful relationship with Microsoft, you most likely need to act in a way that Microsoft doesn’t find offensive.

The very same thing happens with the open source community. You need to follow the letter of the license and the copyright, but you also need to act in a way that is compatible with the spirit and the intentions of the community. This is a learning process that tends to take a while, because it’s not only a couple of decision makers who need to understand it.

It also needs to become the ingrained behavior of the company as a whole. A lot of companies come from a very proprietary-focused world, where everything is under NDA and based on company negotiations rather than developers freely talking to each other. Therefore, it very often happens that while the company as a whole, says, “Oh, yes, this is the way we want to go,” a lot of people in the trenches don’t understand it and have essentially a different approach to solving the problem.

Scott: How does that scenario play out at Intel?

Dirk: Over the past eight years that I’ve been with Intel, I think the company has changed very dramatically in the way it interacts with the open source community, and I think we have done very well. We are now one of the most respected vendors in the space, and we actually received an award earlier this year at the CEBIT Fair in Germany, as being the best hardware firm to work with.

We have a lot of very highly respected developers from a number of open source projects who work at Intel. I think most open source projects perceive us as being beneficial and a good force to engage with, which may have not been the case 10 years ago.

This is a process that takes a while, and that takes commitment from an executive level down, but it is something that I think each and every company can be successful with. I’m very excited to say that Microsoft certainly has taken some great steps in this space. Their cooperation with the Samba team over the last couple of years has been very successful and has had very good results in improving interoperability, which helps everyone. Their contributions to the Linux kernel a few weeks ago helped Linux be a better guest in the HyperV environment.

There are a lot of areas where Microsoft has their own open source projects, and different companies have different degrees in which they progress on this path from being completely proprietary to understanding the heterogeneity of the world and successfully engaging with the open source development community.

Scott: Particularly if you go back three or four years ago, it seems that many areas of the open source community had a very suspicious and skeptical view of commercial companies. When a company decided to embrace open source, they were at great risk of having any misstep they made seen as a “gotcha” moment.

There were always people waiting to dismiss them as big, evil companies that weren’t really serious about open source. Do you feel that enough companies have now changed their direction and stance toward open source that the degree of suspicion has lessened? To what extent do companies still need to be cautious about that issue as they start down the open source path?

Dirk: Let me give you a slightly oblique answer and then bring it back on point. I think the Daily Show’s Jon Stewart is actually one of the very few people on TV who asks the hard questions of people who need it, and something he said a few years ago has stuck with me. He said that there are very few people with truly extreme positions in the world, proportionally, but they seem to be the ones dominating too many conversations.

I think it’s often true that, whether you’re talking about the public debate on health care right now, software development models, or the actions of companies, the people that the press reports about and who get noticed tend to be the people on the extremes.

The pragmatic moderates that just want to get work done are too busy to spend hours and hours screaming and yelling and telling the world how terrible it is. Whether it’s politics, economics, or technology, there is always a very vocal minority with extreme opinions that get a lot of attention even though they do not represent the consensus in the community.

I think that the majority of the people who do the actual work of open source, as opposed to the people who have all day to blog and comment on Slashdot, are extremely positive to each and every company that comes around and contributes. They try to help them around the pitfalls, and they try to make sure things are done right. By and large, they just try to embrace anybody who wants to join the open source community.

I’ll give you an example. Gregg K.H. at Novell is doing a tremendous job of embracing companies that want to have Linux drivers but just don’t know how. If you have a driver that doesn’t quite follow Linux standards, that is kind of awkward, ugly, or broken, he’s willing to put it into the Linux staging tree as a starting point. He has a bunch of developers who will take documentation, under NDA if necessary, and work to create the drivers and bring more companies into the fold.

The same is happening in many other projects. There is a very strong outreach there, of trying to get more and more people involved. There are a devout few who will complain and whine and be unhappy with what’s happening, but the reality is that the majority of the people actually doing the work are very excited to get the companies engaged. They’re very tolerant of the fact that there is a steep learning curve, and they’re willing to go out of their way to get as many companies involved as possible.

Scott: We’re running a bit low on time, but that’s a great place to end.

Dirk: Thanks–this was a fun conversation.

Bookmark this: