An open (source) discussion with Hal Stern

As the new chief technology officer of Sun Software, Hal Stern plays a vital role in developing and executing Sun's strategies for open source projects. In this interview, he explains his shift from CTO of Sun Services, the growing role of open source at Sun, and what open source has in common with Cinderella.

Inner Circle (IC): You recently changed positions at Sun from chief technology officer of Sun Services to chief technology officer of Sun Software. Why the big shift?

Hal Stern (HS): I don't think there's a huge technological difference between services and software. For the last two years, services and software have been working very closely on everything from the customer network systems architecture (how we tether customer systems to Sun) to software diagnostics (how we gather information on a piece of software to determine what may have failed and what should be done).

There's also overlap between services and software when it comes to software quality, overall systems quality, and service level management. In fact, most of the newer programs I was trying to drive as services CTO — in terms of diagnostics, support for custom engagements and the overall customer experience with Sun — were largely governed by software.

IC: Where does open source fit in at Sun in your view?

HS: I think it's really important to dispel this myth that open source is a product. When people talk about us, they say: "You're going to open source Solaris or you open sourced what was access manager." Those are products available as commercial distributions from Sun, and they're now also open source projects. This allows other people to go build projects around them — to reincorporate the source code into something else.

IC: Point taken. So, what is Sun's strategy for open source projects?

HS: The reason that open source is a vital part of our strategy is that we've reached a juncture in the economics of the software business worldwide. The critical change is that it's no longer the case that software distribution and software compensation happen at the same time. Historically, you went down to your local Fry's or Circuit City and you bought a piece of software — whether it was a game, a CD to slip into your PC, a music CD, or a DVD. Regardless, when you bought the software you acquired the bytes and you paid for them at the same time. That model worked very well when there was a unique and differentiated value in those bytes.

IC: What has changed?

HS: Today, there is a tremendous amount of work being done at what I call the platform layers — the operating system layers, the virtualization layers, and the Web services layers. Distribution and compensation are becoming separated. Distribution is something that is being fueled by open source. Whether it's a Linux distribution, an OpenSolaris distribution, or the distribution of an application server, these distributions create large and vibrant communities. People are adding to the creative commons around the software. The distribution, in fact, provides the basic tools for the creative commons.

IC: OK. I see how distribution has changed, but what about compensation?

HS: The compensation part is about creating value on top of those distributions — and then charging money for the enhancements. For example, some of our Java Enterprise System suites are differentiated — they're unique to Sun, so we charge money for them. When you look at things like information lifecycle manager or federated identity, those are products that are built on top of a common platform. We can charge money for those things that are differentiated, but at the same time, we also want to create and harness the large amount of intellectual power and innovation that is going on at the platform layer.

IC: How does the new organizational alignment at Sun help create added value for open source projects?

HS: What we're trying to do here is be very practical about where the value is created, where there is unique value that someone is looking for, and where Sun will benefit the community and the community will benefit from Sun. I think that's a very big change. Open source is important to us because we are on the leading edge of this change. It's not a new thing for us: We started looking at the changing economics a few years ago when we switched the pricing model for the Java Enterprise System — from CPU, per user, or some other difficult-to-measure metric to per employee pricing.

IC: How does open source affect Sun's competitors? Do you think they are being persuaded to contribute to the mission?

HS: Each of our competitors has, to some extent, been participating in a number of open source projects. This allows them, as it does with Sun, to be very clear about where they offer value. What are they doing above the platform level? And what are they doing to be innovative on the platform level? Personally, I hope we'll see far fewer proprietary extensions to platform level. I hope those are the sorts of projects that will be done in open source communities.

IC: What do you think of Sun's portrayal of the future of open source?

HS:I believe that open source has been painted — at least from Sun's perspective — as and either/or kind of thing. It's either Solaris or Linux. It's either Red Hat or Solaris. It's either renewing your public license or CDDL. That is not how open source works. Open source is very much a plurality of communities. Open source can be a both/and situation. This is a case where we support Linux, we support OpenSolaris, and we support commercial Solaris.

If someone has a great idea for an operating system innovation, they could add it to OpenSolaris, they could add it Linux, and they could add it to another appropriate vehicle. We do not believe that any one of the licensing models or project collections is wrong or incorrect.

IC: What influences at Sun does your open source office draw on?

HS: The open source office is part of a larger effort to drive a greater sense of community within software engineering at Sun. Today we have some absolutely world-class governance processes. If you look at the way we design software, review design, and review the way the software is manufactured — the proof is in the pudding.

For instance, look at the Solaris releases and their stability over time: This reduces the amount of work that our ISVs go through compared to having to do a full port or having deal with changing interfaces. Not to trivialize what our ISVs do, but look at how relatively easy Sun makes things for them to upgrade to each release of Solaris. That is a direct result of our governance processes, which govern how to make our products durable and sustainable.

IC: How do those influences inform the work of the open source office?

HS: The challenges are: How do we make our products into systems, and how do our products interact as part of a system? Open source is a system. It is a system of developer communities, and a system of consuming communities. One of the goals for the open source office is to strive for consistency in the mechanisms that we use. I'm not saying that there has to be one single mechanism, one single way of open sourcing a project, or one single licensing scheme, but we need to have consistency in the way we make those decisions, as well as consistency in our goals.

The Open Source Office will be helping the people involved in Sun's open source projects employ best practices — one of those will be targeting license proliferation and using a limited pool of licenses, such as Apache, GPL and CDDL for example.

IC: How do you measure the value of open source?

HS: There are many ways to measure the effect and depth of an open source community. You can look at what people are writing about you, where you show up in techno blogs, how many Google searches you show up in, and how many people are blogging about you. Or you can examine how many active members of your community you have and how many projects have resulted in contributed code. To me, one of the fascinating things about OpenSolaris is within a month of the Solaris source code becoming public, we had a project that was put back into the commercial Solaris source room.

IC: What was that project?

HS: It was great. It was a request for an enhancement, something that had been bugging a particular developer for a long time, and he just went and fixed it. To me that is a great statement. That's how open source is supposed to work: If something in the code bothers you, if there's something that you want to see fixed and it's obvious that it will benefit everybody, then you can make the fix, and the bytes are back in the distribution.

Then, that OpenSolaris fix cascades down to commercial Solaris, and the fix becomes part of the main base. That is another measurement you can take to show that there is value in creating distributions and there is value in working closely with the open source community.

IC: It sounds like looking for new ways to measure value is also a core component to the success of the open source office?

HS: Part of the open source office is ensuring that we're measuring the right things. In fact, we want to continue to invest in and evangelize those things that don't have clear revenue streams, because, in some cases, it is more important to find promising projects than to have clear revenue streams. If you don't have good distributions and you don't have good pervasiveness, it is a lot harder to go sell enhancements that represent value on top of your software platform. It's much easier to sell a support contract to someone who is running your software than someone who has no idea whether they want to run Solaris. Driving the prevalence of Solaris is going to help our other businesses.

IC: As software CTO, do you decide which software Sun will open source? How are these decisions made?

HS: This is a great question because, again, I think there's some concern about statements that we're going to open source everything. Personally, I think there is a word missing from that statement: We are going to open source everything we have today. The missing piece of the statement is that software engineering is not static. We are continually innovating up the stack. Two years ago there was no Java Enterprise System. Before that there was no system for identity management.

You look at the problems we're thinking about today: security, federated identity, and information lifecycle management. Or, how we continue to build world-class communication suites. Or, what software liability and software availability looks like down the road. Those are all areas where we're innovating. Those are all areas where there could be value added on top of the software that we may or may not open source.

IC: How is the decision to open source something made?

HS: The question of how do you decide to open source something comes down to a number of questions: Is there value to the community? Will there be a community that will embrace it and want to work on it? Is there a very clear demarcation, between where there is added value and the platform itself? If you look at open SSO (single sign-on), for example, it's very clear that single sign-on is something that is becoming accepted as a core part of almost everybody's product.

If our goal is to drive interoperability, to make it easier for people to add extensions to other sources of user information, the best way to do that is to open the source code and let people add to it. Just as device garbage got written in the Linux world because people added to that base, we should drive interoperability by opening source code for the single sign-on layer. That is a clear example of where it makes sense to open a layer, because we're going to get the right things back from the community.

IC: Are there any other factors in the decision to open source something?

HS: Yes. We look at what constitutes part of the platform, what constitutes real unique value added by Sun, and where we are going to go add value to the whole stack. Just because something is open sourced doesn't mean we stop innovating. We can continue to work on the things like reliability and quality. And those enhancements can drive other parts of our business, like services and developer opportunities. I think there is still a tremendous amount of internalization that will happen.

Part of the larger question about the open source model is where is initial innovation going to happen? Where is the next really great idea going to come from? And, conversely, where are we going to have incremental innovation? I believe what you'll see is that companies with large software R&D efforts — Sun, Microsoft, Oracle, and IBM, as well as major universities — are going continue to have initial innovation. I wrote a blog entry on initial versus incremental innovation.

IC: It is your view that initial innovation is going to continue to happen at places that can afford to do R&D?

HS: There's going to be tremendous amounts of initial innovation going on, and I would say that the bulk, not all, but most is going to happen at places with funded R&D groups. At the same time there is going to be tremendous amount of incremental innovation going on — people adding to a base, incrementally evolving a base, fixing it, extending it and so on. That is going to happen through the open source communities.

The challenge is determining what are we hoping to get from our R&D dollars. What are the things we think are going to create an advantage for Sun? And, therefore, what do we do as part of an open source project, and what do we do as part of a software product that we innovate and sell? Plus, there is a lifecycle, so just because we start off selling something doesn't mean that over time it might not turn into something that we end up open sourcing.

IC: Are there any other open source misconceptions that you want to clear up?

HS: Yes. Just because something entered the creative commons, or has entered the public domain, that doesn't mean that there is no more value that can be added to it. The best example I can give you comes from fairy tales. Whether they are Grimm's fairy tales, Aesop's fables, or Hans Christian Andersen's stories, they are in the public domain. They have survived long after the author's death.

Everyone says: "OK, great, can you make money from those works today?" Whether it's Cinderella, Three Little Bears, Beauty and the Beast, or Snow White and the Seven Dwarfs — those are all tales that were in the public domain. They are as close to open source software as you can get from an artistic perspective. So if you're Walt Disney and you color in the story, add the characters, make it accessible, and create a narrative around it, all of a sudden you open up that story to another generation of people.

IC: So, adding value is part of a virtuous cycle?

HS: Today, there is a lot of open source software, it's freely available, but the value doesn't stop there. It's what you do with it, what opportunities you create from it, and what other kinds of services you build on top of it. From my point of view, you don't say: "Well, it's freely available, so there's no value in it." Open source is a starting point. What you do with it is the difference between a fairy tale about Cinderella and a DVD that you want to watch with your kids.