Sun Microsystems president and COO Jonathan Schwartz is in Sydney to meet with government and corporate clients. His outspoken, maverick views frequently put him at odds with his competitors as Computerworld's Rodney Gedda discovered in an interview yesterday.
Does Sun have a strategy to deal with Sparc-to-Intel migrations?
Sure. We can talk about the definition of commodity in a moment, but we have the most complete line- up of 64-bit x86 systems in the market today. So this is not a new market opportunity. The x86 server market is about a $US25 billion a year market in which to date we have had little to no participation.
With the acquisition of a company called Kealia about a year ago, we really put both feet into the x86 server market and now have a 1-way, 2-way, 4-way, and discussions about the evolution of that roadmap.
But what's most important in the discussion is to understand that the increment of x86 server business to Sun is interesting. But what's more interesting is how many platforms Solaris, our core operating system, can now be run on because there are more than 250 such x86 servers that can run Solaris whether they are from Dell, HP or IBM. All of that is new market opportunity for us.
Now what's particularly interesting in that evolution is that [when] you look into x86 servers you can see that we have an operating system, Microsoft has an operating system, and largely Red Hat has an operating system and HP doesn't and nor does IBM. So we believe we are at the beginning of one of the single biggest competitive windows in the history of Sun where we can now begin pricing and evolving technology in a way that neither HP nor IBM can possibly keep up with, because they have foolishly taken a short cut with 'let's throw away our operating system' in the hope of pursuing a partnership with Red Hat which I think is great for Red Hat. I just wonder what it does for HP or IBM.
Any projections for Sun's bottom line?
The word commodity, to me is a fantastic word. And I think there are several people in the industry who shy away from the word commodity because they believe it means zero margin.
Well, let me tell you that the largest companies on earth are in the commodity businesses. They happen to be our core market - financial services, telecommunications, oil and gas, and I'd argue that the government is in a commodity business. But the largest businesses on earth are those built around companies that can create value in commodities. Now the reality is that a 4-way x86 server running a 64-bit operating system is anything but a commodity.
The real commodity in our industry is bandwidth. And by the way, every place you look, I don't care where you're going in the world you not only have access to bandwidth but everyone who lives where you're visiting is consuming bandwidth and wanting a lot more of it. So when I talk about a very large-scale business opportunity, consider whether you have taken the maximum number of digital pictures you'll ever take? Or listened to your life's allowance of digital music? The answer is no.
It's up to us to fully exploit that as a market opportunity, but we're one of the few companies that can because we have the technology that is fundamental to the growth of the Internet, so I feel great about that.
You said on your blog, "Sun is all about ideas and communities." Is that statement a reflection of Sun's business? How do you compare that to 12 months ago?
That blog was all about the centrality of intellectual property to our business. Even though people look at us and say 'you ship servers, you ship storage, and you ship CDs with software on them' the reality is we just produce ideas. In our business we refer them to a manufacturing company which takes our designs and builds a server out of those ideas.
So how does that compare with HP which theoretically invents and IBM which is the largest patent...
in the world and so on down the line?
Let me give you a couple of specific differences. If you look at HP they're trying to become Dell. They're, despite their tagline, moving away from invention, they're moving towards resale. And if you examine, for example, the choices they've been making they've moved away from microprocessors to support Itanium, they've moved away from HP-UX to support, I don't know what they support but it's not HP-UX, they've moved away from the evolution of their hardware systems to instead simply resell what an OEM builds for them. They resell flat-panel TVs, they resell ink. They are becoming a channel company and I believe that is in fact the case.
So when we talk about ideas we talk about - and let's compare and contrast the two scenarios - when it comes to x86 systems we are building everything but the microprocessor and the single biggest investment, by the way, is in an operating system which we are in the midst of open sourcing. Now compare that to HP. HP doesn't build the system and they don't build the operating system so that is the starkest contrast I can identify. That's why I believe we can achieve greater margins over time because in general R&D should equate to greater profitability because of the technical competitive advantages.
Now compare that to IBM. IBM is a very inventive company, no doubt about it, but they're also historically known for making incredible strategic gaffes. My favourite among them was they thought that PCs were very clever machines and you should source your operating system from a little company in the Pacific North West that they thought would do them the favour of providing an operating system. Well that company turned into Microsoft and while they were living through that first incarnation of the Stockholm Syndrome - when you fall in love with your captor - they fell in love with Microsoft and Microsoft is more valuable than IBM now.
Well, round two of that same opera was that they thought Linux was going to boil all the oceans and cure all the cancers. And it was a wonderful concept as a social movement but the pragmatic reality is that you have to buy a product from a company. And so while IBM was saying 'Linux solves all of our problems' they failed to notice that Red Hat was beginning to erode their relationship with their customers. And so now if you're going to talk about Linux, why on Earth would you talk to IBM? Because they have no skin in the game, they're just a box shifter. You talk to Red Hat. And so how does that now represent IBM's investment in R&D. It doesn't, it represents their investment in rhetoric and resale which I think is going to be a tough competitive posture for them to take when we go into their accounts and say we'd like to move you onto our full line-up of x86 Opteron systems running the single most secure scalable and performing operating system available, Solaris 10. And by the way, it's open source.
Where's the value in open sourcing Solaris for Sun, are you giving away the jewels here?
The only companies that have something to fear from open source are the companies that can't innovate. Because if you can't keep up with someone who's copying your code, that's appalling. In the history of Sun Microsystems we've spent more time and energy giving ideas away because that yields standards and a growing market. Why do we give them away? Because we have such express confidence in our ability to come up with another 50 good ideas tomorrow. So I'm not worried about being competed against on our history. The competition should worry about competing against us in the future because they appear to be grasping onto history when they are pursuing their next-generation strategy. HP pursuing a strategy that says 'let's just resell somebody else's product' reminds me of the strategy Unisys had back in the early nineties. And there is a history lesson in this for us. There was a project called the Chicago Project. And it was what was supposed to blow Sun out of the datacentre. It was Windows NT. The companies that adopted the Chicago Project, they're dead now. Why, because you can't honestly have a relationship with a customer if all you're doing is reselling products.
What are customer concerns?
There was definitely discussion of migration strategies off of the competition, and specifically HP. I think they are more vulnerable now than I've ever seen them. It was curious to us the extent to which they wanted us more engaged in the operations of their datacentre.
Traditionally we've said there is a fine line that we don't cross which is we deliver technology and innovation, but we're not an ASP. That's EDS, CSC, and IBM Global Services' world. And the most interesting thing about the discussion was what we raised mid-way through this, the fact that we don't have a retail business, that we've never tried to deliver something to the bank's customers or to Telstra's customers means that they can look to us and trust us because we're never going to figure out a way to get between them and their end customer. That's why they were interested in getting more help because basically they trusted us not to go after their customers.
Now that's not a relationship they have with Microsoft given that Microsoft has a retail business. And it's one I think is being stressed a bit with HP as HP tries to build its consumer business.
What happens when the industry ceases to deliver innovations which customers value? Because I think that's really where Dell has most effectively run its business. Which is 'why bother with innovation in a PC if all you want is a good price' and all Dell will give you is a good price. To me Dell runs into a couple of interesting competitors. One of them will be Wal-Mart. Because last I checked Wal-Mark had actually perfected the process of delivering products that have no innovative value, just a better price. The second is, I think, Dell needs to worry about the evolution of its operating system partners. Because as we move to, for example, a model that the telecommunications companies have all deployed, which is 'the handset is free if you subscribe to the service'.
When we move toward 'the hardware is free if you subscribe to the operating system' in that world of low-cost supply, how does Dell compete? Because I think there has been an aberration in time that has suggested that you buy all the components of your datacentre one by one. But what I've begun to see is customers would merely like to subscribe to most of the services if they were made available to focus on what is really core to their business. And so when we start delivering Solaris 10 on a subscription basis I don't see how Dell makes money. Moreover, I really don't see how Red Hat makes money. We're the only vendor that has the ability to offer a complete package because no one else owns their own operating system - they've all moved away in pursuit of a press release saying they have scaled all of the tall buildings and boiled all the oceans.
And unlike Red Hat, we would not only open source the operating system we will then give you a roadmap of the innovation that will show up in the next quarter rather than saying, 'the benefit is we put a brand on the box'. I think the biggest problem Red Hat is having in the marketplace is that customers are realizing that a thousand dollars a CPU doesn't feel like free any more. Solaris at $795 - all of a sudden we're cheaper than free. And for the customers that we've begun migrating away from Fedora and Red Hat Enterprise Linux we know that we have a technical competitive advantage. That advantage is, our product is just better. We've spent three years fundamentally re-architecting Solaris. We outperform Red Hat on every benchmark. I have no interest in competing with a social movement called Linux. Remember the irony of ironies is Sun was founded on an open source operating system called BSD. So as we return to our roots we have nothing to fear. And do I think another company can out-execute us on Solaris? I'm not worried.
Any thoughts on Oracle's moving its development platform to Linux, hence competing with Sun?
No, to Red Hat. Let's be clear. Which is curious as Red Hat now begins to compete with Oracle. Where does that leave Oracle?
Oracle is going to adopt the platform that gives it the best price-performance in the marketplace, period. And while we have historically been reticent to support Solaris on x86, now not on the support of Solaris on x86 but with the broad x86 product portfolio in 64-bit computing. I'm very comfortable we can go deliver exactly the [Intel] hardware savings on our own systems. And, by the way, will run the fundamental operating system with much better security and much better scalability than Red Hat. So I think you'll see our relationship with Oracle come back to where it's been historically. We're two companies that care a lot about innovation. And it's funny how MySQL has become so popular isn't it, speaking of commodities. And I've seen an awful lot of customers save a huge amount of money moving to alternative databases whether it be Sybase or MySQL.
In the long run customers always have exercised choice. So, do your best to lock them in, like IBM did with mainframes, [but] they find alternatives. Do your best to lock them in with open source, they'll find open standards.
The companies that are at risk are those that fail to recognize that history repeats. Just as I watched Unisys perish because it elected to pursue a strategy of 'let's go resell technology' rather than invent something, I think we're in the midst of watching HP go through exactly that demise as it moves away from innovating operating systems or microprocessors and tries to become a company that's all about reselling iPods.
The best parts of Solaris being put into Linux?
First of all you can't use Linux in the room with me. You have to identify the distro and the company that's delivering it. Linux is a wonderful concept [but] customers don't buy concepts, they buy products from companies.
There is a whole slew of advantages we can offer against Red Hat. Now will Red Hat be able to take our innovations and place them into Red Hat's products? I think that will be the subject of a licensing discussion. Just as if we were to put Red Hat into Sun's products we'd have to have a discussion with Red Hat. Now the licence we use is going to be OSI (Open Source Initiative)-approved and we'll be sure to make the code and technology available to as broad a constituency as possible.
We just release a new version of Java and when you look at who is contributing to that it's this incredible community - academics, researchers, companies, individuals, not-for-profit foundations - it's a huge community. It's not just the GPL community that delivers innovation. The Java community delivers innovation, the closed-source software world delivers innovation, and I think the one complaint I hear about Red Hat, consistently, is that Red Hat doesn't deliver any innovation. All it does is deliver a box that has a nice brand on it.
I was talking to a customer who said 'I have a problem with Red Hat and I've paid them $US1500 dollars for a two-way system I'm running it on so I call them up and say I'm an important customer I have a problem and I'm going to send it to you in an e-mail'. The next thing the customer sees is that question showing up as a posting in a news group. The exact question. And he said why on earth am I going to pay you US$1500 per year to have you take my questions and post them in a news group, I think I can do that on my own.
So I think the competition that's really going to be interesting in the next 12 months is going to be between Solaris 10 and Red Hat. Not between Sun and Linux. Sun is Linux. We're the single-largest contributor to open source in the world, second only to the University of California, Berkeley. That's not going to come down to social policy and rhetoric, it's going to come down to 'is Solaris 10 better than Red Hat Enterprise Linux'. Bring it on! I can't wait for that to happen.
Are you confident you won't get a few phone calls from Darl McBride of the SCO group saying you can't open source Solaris?
We're very confident we own the intellectual property necessary to pursue whatever business objective we identify. Very confident. And in an odd way a lawsuit from SCO would probably do more to establish our street cred in the open source community.
HP sent a letter to Sun asking you to stop commenting on their strategies publicly.
I haven't found a single customer that disagrees with me [that HP-UX is dying]. I think that was the single, biggest communications gaffe that that company has fallen into. To me, when it made those statements, it just felt to me like its management team is out of touch; you don't go complaining about what the competition is saying because ... what it does is ... it must have driven a hundred thousand people to my blog. So thank you for the traffic.
But HP has come back and said 200 customers have migrated from Sun to HP in the past 18 months.
Why don't you call a few of those customers and see which operating system they're running on. How about a product not shipped by HP? In our HP Away program we've been migrating HP-UX customers to Solaris. To me that's a migration. Migrating a legacy Sun customer to a new x86 platform is called a discount. That's not a migration. I think HP has got a bank in Bolivia that has moved some 15-year-old Sun server to a new Windows box and HP calls that claiming a Sun customer. So I'd love to see the detail. We'd be happy to provide you with the details of the 250 HP customers that we've migrated with the HP Away program.
Despite how it may appear, we are very careful about what we say. And we pick our battles and if you read through our discussion on HP it merely says basic data.
Your comments in the press earlier this year suggesting IBM should buy Novell were controversial. Why did you say that?
The vendors that have adopted Linux have in fact adopted other companies. You can't adopt a social movement called Linux, you have to adopt Red Hat. And specifically Red Hat. Now a lot of Red Hat customers have told me that their IBM reps have walked in and said you must move off of Red Hat or you'll lose your discounts, you can preserve your discounts if you run Novell. And these guys were angry because they have built their systems. And so what that felt like to me is IBM has discovered that they have a problem - a dependence on Red Hat. As they began to introduce an app server it was competing against IBM.
Unfortunately Oracle isn't qualified on Debian so you can't just move off Red Hat. Most of the ISVs are not qualified on SuSE yet so you can't move there either. This is why IBM is struggling because as Red Hat now increasingly raises its price and begins to compete, IBM has a dependence problem. As far as I'm concerned IBM is stuck. If you check to see which [Linux] runs on zOS, now it's just Novell. Why? Because it doesn't want Red Hat on those mainframes as it's too big a risk.
To say that IBM could pursue a strategy that ultimately doesn't yield the same dependence on an operating system is a bit naive. It just doesn't make any sense. No matter how great all the little distros are, they do not represent deployment targets for enterprises. The perception that there is complete competition in the Linux market could not be farther from the truth. To me, what's happening with Red Hat is that it's creating a lot of market opportunity for Sun and preserving it in Unix; I can now go back to all those Red Hat customers and say I can give you a lower-cost operating system.
Mark Jones and Sandra Rossi contributed to this interview.
This transcript has been edited for clarity