Sun's chief open source officer was one of three keynote speakers at the recent Australian national Linux conference.
He spoke to iTWire soon after he had given his keynote. Edited extracts:
iTWire: Do you think that Sun made a small mistake in delaying the open sourcing of Java?
Simon Phipps: I would have personally preferred to see it happen sooner. But there were really good reasons why it took as long as it did. And where we're at now, I think I'm very pleased with the results we've had. Java is available now in all the main Linux distributions. It's available in a ready-to-use format that works with the package management system on those platforms - so it's there in Synaptic on Ubuntu. It's... integrated into Fedora. So despite probably a year or so delay, I think we've actually reached a really good place with it.
iTWire: Do you think it has cost you anything in terms of adoption vis-a-vis .NET?
SP: Not really because .NET isn't really an aspect of the Linux platform despite the very best efforts of people to get it there.
iTWire: No, I'm asking about the broader community.
SP: No, I don't think that Java and open source is relevant to the space where it's competing with .NET.
iTWire: Sun's other main open source effort in recent years is OpenSolaris. How is that going at the moment?
SP: OpenSolaris is actually going reasonably well. The way they look at OpenSolaris is to see it as a transition from the classic data centre Solaris to a Solaris of the Web 2.0 and afterwards. That transitional process is something that most of us thought would take maybe as much as five years to get through. I'd say that we're going reasonably well on that timeline. There's always things to be proud of and things to regret in any community activity, but taken as a meta-observation I'd say that we're going pretty much on track.
iTWire: Is OpenSolaris primarily for developers who want a graphical desktop or is it also for ordinary users?
SP: I'd say that OpenSolaris is aimed squarely at developers who want to have a graphical desktop. Those are the developers who, in the future, will produce the next generation of both client- and server-based OpenSolaris activities. That's not to say that that's what OpenSolaris will be about in three years' time. But where we are at the moment, I'd definitely say it's for web developers, and for the next generation of Unix developers.
iTWire: Basically tech-savvy people?
SP: Pretty much for tech-savvy people. Tech-savvy people - you know, the comment that I've made during the keynote here at LCA was that, well, I have a new laptop that I bought myself the other day, and I very much would like to have OpenSolaris on it, but I don't want it to be a hobby. I want it to be a computer. And even tech-savvy people want to have a computer that's a computer, and not a hobby.
iTWire: Have you set some kind of target - by this date we're going to have an OpenSolaris which can be used by someone who's been using computers for so many years, but who cannot actually be called tech-savvy?
SP: I don't think we're thinking in those sorts of terms, no. If you look at the current OpenSolaris, the OpenSolaris 2008-11, it is already remarkably approachable. You can get a live CD and run it up on your PC right now with no technical savvy at all, and it will work great. From that live CD, you can install it with minimum fuss.
The real thing that is a question is whether the respository that it's using - the software repository from which you do your installations, whether that's populated with the software you need or not. So it's not so much about your technical savvy it's about your area of application use. And for 2008-11 that repository is well-populated but in need of growth. Then the next release that's coming up - 2009-04 - will have a much better populated repository because the community is energetically migrating packages into it.
iTWire: What kind of impact will the current loss of jobs at Sun have on your open source strategy?
SP: The Sun open source strategy is an adjunct to our product strategy. In the restructuring that's going on at the moment, there are undoubtedly going to be some projects that are cut, and the open source activity that results from those projects will be impacted, I'm sure. But it's worth commenting on though, that what's going on at the moment - this is not a classic head count reduction activity. We already did that - that happened last year. This is a full-scale restructuring that's going on here.
We're taking all the building blocks that are on that table, and we're building a new company on this table. The people being laid off are not a percentage of the workforce - rather, they are the parts that are not being incorporated into the new restructured company. Some (parts) of the restructured company actually has open vacancies. They're actually hiring new people. Some of the people who are over there will actually find new jobs in the restructured Sun.
iTWire: Is this something that has been forced on you by the economic conditions?
SP: I'd say this is something that was a necessary next step in the growth of Sun. But I would say that it's been precipitated now because of the current economic conditions. We had the plan in track in the middle of last year. We heard from other companies such as Intel and Microsoft just today (January 23) that they were laying off.
What's happening at Sun is the working through of an announcement we made at the end of last year that was itself part of a process that started earlier on. Obviously it is the economic conditions that are involved, but it's part of a more thought-through restructuring that we're doing.
iTWire: There have been several comments from senior Linux people, including kernel developers, that the one thing that has held up adoption of OpenSolaris - which has some very good features - by people like them is the licence. Does Sun have any plans to make any changes to this (CDDL) licence?
SP: I disagree with that. One of my comments in the keynote, was how you pick licences based on the community you want to affiliate with. One of the essential parts of OpenSolaris, all along, has been to say that the OpenSolaris community is the OpenSolaris community. The Open Solaris community is not exiles from the Linux community. The Open Solaris community is not defined by some other community. It is a community in its own right.
And that community is actually very attached to CDDL. Now, if in the future, there were to be any change in the licence, it would be in order to affiliate that community with some other community, rather than because there is some misfit in the licence.
At the moment, if you look at the OpenSolaris community, it's grown substantially over the last year. There are a six-figure number of participants. There are something like a hundred user groups around the world. There are substantial daily pings from computers that are using OpenSolaris, checking our respository for new builds. I'd say that it's working fine, as a community in its own right. It's not in any way hampered by CDDL.
What is hampered by CDDL, is the relationship of that community to a community that's using GPL. I quite recognise that. When the time comes for the OpenSolaris community to have a new kind of relationship with a GPL-using community, then probably there'll be some sort of change in the licence. But that time isn't yet.
iTWire: So you think that there's scope for a whole new band of users who appreciate the strengths of OpenSolaris to come in, without poaching people from the Linux community?
SP: Absolutely. And that's demonstrably happening. There's very little flow that I'm seeing in any way demonstrated, either from the OpenSolaris community to some other operating system, or from some other operating system to the OpenSolaris community. The OpenSolaris community has got a lot of students in it. It has a lot of software developers from enterprises who are prototyping it, rather than deploying it. And I think what we're seeing here is a net growth in the open source community, rather than a migration around the open source community.
iTWire: What future plans does Sun have vis-a-vis open source? What other projects do you see in the pipeline?
SP: Well, we've got pretty much our entire portfolio in a state where it's open source now, and the one or two projects that aren't open source, like the composite application server, Java caps, are unlikely to go open source for licensing reasons. I think what we'll see now is a focus on community growth and on anchoring the product strategies around each of those open source communities. I'm not anticipating that we'll see substantial new open source projects coming from Sun in the near future. I think what we're now going to see is a focus on community consolidation.
iTWire: There were initially some complaints when community members tried to contribute to the OpenSolaris code base - they said there was too much of bureaucracy. What steps have been taken to ease this because you say the community has grown to such an extent?
SP: I think that was 2008-05.The difficulty that we had was that when Solaris was being developed inside Sun, it was being managed in a version control system that was proprietary. And one of the steps we had to take in making it open source was to move all of the source code into a publicly accessible open source version control system. So we moved all the code out, and into one called Mercurial.
Moving a substantial code base to a version control system with a completely new architecture actually is a really big job. It wasn't complete at the stage that you're talking about. That is complete, I think for the vast majority of OpenSolaris now. So there's now publicly accessible version control systems where developers wanting to make commits can download the code and make commits back in. There's still a few places where that is a work in progress, but the vast majority of the areas in which people are interested in contributing to are now on the public version control system.
iTWire: The timeline you have for OpenSolaris is working out more or less?
SP: I'm close enough to the project that I can find plenty of things to be unhappy about.
iTWire: What are you unhappy about?
SP: They are all things that are of really no consequence if you're not looking at it with your nose up against the glass. In the big scheme of things, I'd say that OpenSolaris is going pretty much as well as I would have hoped it would go. The release that comes out in 2009-04 looks like it's going to be pretty well based with a good repository, with the earliest alpha of Sparc support in it; and its looking like it's heading towards being data centre ready by the end of the 2009-11 release. That's pretty much what everyone said ought to happen over a year ago. So we're pretty much on the right timeline.
iTWire: I'm asking all these questions because sometime back you had a system called JDS. There was an equally huge amount of enthusiasm around it. Suddenly this kind of disappeared...
SP: JDS became the GNOME desktop in OpenSolaris. So it didn't actually go away, but what did happen was that instead of the assumption being that we would have a Linux kernel underneath it, the assumption was that we would have a Solaris kernel.
iTWire: I know what happened to it but...
SP: I think that that reorientation worked, and at the moment, I'm looking at OpenSolaris feeling that as a community project, it's heading in the right sort of direction. In business terms, you'd need to talk to the business part of Sun, because that's not me.
iTWire: When I speak to people in the technology arena who have been there for some time, they say that no matter what you're selling right now, if it's proprietary you have to adapt. You have to go open source at least to some extent or you'll die. Do you agree?
SP: In the keynote that I gave this morning, I suggested that the emerging way of software entering a business is using what I called a adoption-led approach, where, instead of going through a long procurement process, software gets into the business by being prototyped and iterated, until you've got something that kind of works. Then what you go out and buy for things that are mission critical, is the wherewithal to take the thing that kind of works into production. That adoption-led market, where you adopt and then procure is what I think is going to replace what I call the procurement-driven approach, where you go through an RFP process and then a long scrutiny process before you put any bytes to disk, so to speak.
Now, in that world of an adoption-led market, I think that proprietary software can play. You can give away your code up to a point for people to adopt and take into production. I think, however, that the benefits that you get, in terms of diverse sources of service and support from open source, means that it will be very difficult for proprietary solutions to widely compete in about four to five years. In the interim period, there will be plenty of people who see IT as a cost centre, and see proprietary systems as the most predictable way to spend that money.
There'll be plenty of companies that see IT as a competitive weapon, and see open source as the smartest way to have a tuneable, scaleable competitive weapon for their business. I think both of those situations are going to be with us for a really long time. The ratios of how many companies are going to do each is going to change over the coming years. I would say to any proprietary software company that they need to look not so much at open source, as the changing reality of how software is going to get into businesses and be ready for it. The smart way, in my view, to respond to that, is to consider open sourcing your products. But the market is a diverse market that will take many solutions for many years to come.