We grilled Jonathan Schwartz, Sun Microsystems' president and chief operating officer, about open source and Solaris 10, which the company formally launch on Monday.
With the possible exception of some small parts of the code such as third-party drivers, all of Solaris 10's code will be open source. Although most interest in open-source Solaris is likely to come from developers, corporate users may also be interested in having access to the code for performance and optimisation improvements.
In your weblog, you call Solaris 10 the single biggest improvement Sun has ever delivered in a commercial operating system. Why?
Well, there are probably 100 things that make it so. First and foremost, customers get the ability to run the same safe, protected and scalable operating system on our existing Unix offerings on Sparc, as well as the first truly vendor-neutral operating system that runs across Dell, Hewlett-Packard, IBM and Sun's x86, 64-bit systems. We're not only giving investment protection to our customers, but also extending the reach of those customers into a new commodity purchasing opportunity.
Second, we've really gone back and revisited the fundamental issues customers had with their systems. We all know the old chestnut that "The hardware I was running was too expensive, and not only that, I was getting only 15% utilisation so I built out a big grid of little x86 systems."
I was with one customer last week who told me they had bought a few thousand IBM BladeCenter xSeries systems, and their server utilisation was about 6%. Just because you chop up the utilisation problem into smaller units doesn't mean it has gone away. That's why we added logical partitioning into Solaris with containers. Containers allow a one-way system to be treated as if it were a 1,000-way system, each with its own IP address, each with its own root password and each almost instantaneously rebootable.
Those two alone are probably the biggest improvements to Solaris. And an important distinction with our partitioning technology is you only need one Solaris licence and you can run 1,000 instances, as opposed to the proprietary IBM release, which only runs on IBM's Power products and requires a separate licence for each partition you build. The old school of trying to get server utilisation just doesn't make sense any more.
Third, we have made an extraordinary investment in ensuring we bring an open-source Solaris to the marketplace, a truly vendor-neutral Solaris. As an open-source product, it will also be indemnified by Sun in the same way as a closed-source Sun product. Some of the marketing veneer from companies like Red Hat claiming they have a great product but fail to indemnify it strikes me as pretty cynical - it's trying to get customers to run something they can't vouch for.
But also, those in the proprietary software community saying "open-source software can't be safe" may be looking at an old-world definition of open source. Open source can be safe, open-source can be indemnified, and it can bring extraordinary economic and technical benefits and just raw innovation back in the industry.
Many of your customers are mid-size enterprises. What advantage does open source give them?
You have to understand that because Sun is a technology company serving a broad diversity of customers, the vast majority are not large enterprises. By orders of magnitude, the vast majority of our customers are developers. What developers build are ultimately deployed in very large-scale enterprises, and certainly enterprises have far more dollars to spend than your average developer does. Just like when NBC puts up a network, it can't only have programming for one audience. It has to make sure it's delivering a breadth of offerings for all the constituents it serves. That means we've got to be able to deliver innovation to developers.
By the way, some of those very large-scale enterprises want to be able to take a look at the source code, understand how it's designed and do a better job of taking advantage of the performance or optimisation insights they gain. They also want to be able turn around a bug in the event of finding one. They don't want to wait six months for an OS update.
What do you think independent software suppliers are going to do with open-source Solaris?
I'm not necessarily convinced it's going to yield a new set of new applications as much as a new way of interacting with Sun. So ISVs that want to be able to make enhancements or modifications to their products or the base operating system will have the freedom to do so - the freedom to innovate.
How will the open-source Solaris development community work? Will you create something similar to Openoffice.org?
Absolutely. It's critically important for us that we cultivate a very high-integrity relationship with the open-source community, which we have had historically. Remember, Sun was founded with an open-source operating system called BSD. We are returning to our roots in some sense. To establish a high-integrity relationship with a broad and participative community is really the principal objective of bringing Solaris into the open-source world.
What's the time frame for releasing an open-source version of Solaris?
We have already begun interacting with members of the open-source community. We've obviously begun consulting with folks like OSI and just the free software movement in general to make sure we use a software licence palatable to them, and that really gives them faith in the integrity of the ultimate delivery model. We will have the licence announced by the end of this calendar year and the code fully available in the first quarter of next year.
Is there anything preventing you from making all of Solaris open source?
Nothing at all. And let me repeat that: nothing at all.
There is no third-party software in Solaris that can't be open-sourced?
There may be a few little binary plugs here or there, things like drivers that maybe their owners weren't interested in open-sourcing, but the vast majority of features, functionality and breadth of technology represented in Solaris will be licensed under a common model and delivered to the open-source community.
You have called Linux a social movement. What are you saying about Linux in that regard, and do you want Solaris to become something similar?
Linux is a social movement only in the sense that it really builds on the foundation of the open-source community. And that to me is an unstoppable social movement in the sense that you can't bottle up creativity and say it's only going to come out in one form, under one licence with one product. So anyone who views themselves as criticising Linux is really missing some common sense, because what you're doing is trying to criticise creativity.
Bringing Solaris to the open-source community is really a way of returning to the open-source movement on which we were built. We participate broadly in that open-source movement already, whether it's through the work we do with Mozilla, Gnome and OpenOffice - all things that are representative of the relationship we have had with the open-source community. And to the extent that we lost our way - maybe chasing a lot of revenue when there was tens of billions of dollars to be had - I think it's high time we kind of got back to our roots and rediscovered what made that innovation really propagate across the world, as well as give us a different relationship with some core customers.
Is Solaris in competition with Linux?
No. That's like asking if Solaris is in competition with the open-source movement. Solaris is in competition with Red Hat. Solaris will be as much the open-source movement as anything else. The competition ultimately is going to be between companies that have competitive offerings.
One year ago this month, Sun announced a strategic alliance with AMD to deliver Opteron-based Sun Fire systems. What has this alliance accomplished in the past year and what are the plans for the upcoming year?
On the one hand, it's given us a springboard into the $20bn-plus x86 server market, with products that have distinct competitive advantages over similar systems from, say, Dell. I question the wisdom of anyone who continues to buy 32-bit x86 systems when you can buy x86 64-bit systems that cost less and run two or three times as fast. I think that has given us a lot of credibility in the x86 marketplace.
In addition, there is obviously a very close partnership between the AMD team and the Solaris kernel team, and we are looking to co-evolve the systems. It's just a very, very productive partnership, one that we are using to gain back some customers we had wandered away from.
When Microsoft and Sun announced the interoperability agreement in April, it was described as a five- to 10-year process. Five to 10 years is so far away that some might consider it irrelevant. It seems to set low expectations. Any reason to think otherwise?
There is absolutely reason to think otherwise. Software, unlike hardware, doesn't disappear overnight. You can swap a hardware platform, you can move out an Intel 32-bit system and move in an AMD system and do it over a weekend. Software tends to live for decades. Giving consumers the confidence that this isn't going to be some fly-by-night relationship is absolutely critical for beginning to rebuild the foundation of interoperability between the two companies.
Second, there is just some evolution that we need to ensure happens in concert with our customers. If you look at the evolution of Liberty and the Liberty Alliance - how long has that now been in existence? A few years. And we're now beginning to see very large deployments of Liberty Alliance standards-based products, as well as beginning to sketch interactions between the identity offerings from Sun and the identity offerings from Microsoft. These aren't things that are accomplished overnight; these are things where you need to bring customers and whole industries along. The longer we can ensure we are painting a picture for the marketplace, the more comfortable they are going to be.
How can the Liberty Alliance achieve its goals and Microsoft and Sun work together toward interoperability if Microsoft isn't a member of the Liberty Alliance?
You might want to ask Microsoft that question. But at the end of the day, interoperating with a standard doesn't necessarily mean that you've got to join the standards organisation.
Patrick Thibodeau writes for Computerworld
This was first published in November 2004