Solaris Enterprise System: What Sun is Learning from the AS/400

by Timothy Prickett Morgan

Maybe they should just start calling it "iSolaris" and get it over with? If imitation is the sincerest form of flattery, then IBM's Rochester iSeries division should indeed be flattered by some of the aggressive integration moves that rival Sun Microsystems has made to create an integrated system software stack based on its Solaris Unix variant. Sun isn't done yet, of course, and it remains to be seen how well or poorly its integration efforts will go. But I do think that the iSeries division should take a hard look at what Sun is doing, because it is a very real competitive threat to the iSeries.

IBM is about all about business--it is the company's middle name, after all--and you could argue that during the dot-com bubble, when Sun was making money hand-over-fist and was well on its way to becoming the dominant supplier of server iron in the world, Sun, too was all about making money and shooting its mouth off. Sun took it on the chin pretty hard from early 2000 through early 2005, and only this year has it seemed possible that Sun might not only survive, but possibly thrive in the new IT ecosystem where people want to spend less on servers and they increasingly want to use open technologies.

Sun's young top executives--some might even go so far as to say new-age hippie with a techno twist--have replaced the older, stodgier, and IBM-esque executives who were the front men for Sun during the boom years. Complete with the suits. Scott McNealy, Sun's founder, chairman, and CEO, got a haircut--both symbolically on Wall Street and literally at Supercuts--and hired a fresh new team of executives who have new ideas about what it means to be in the information technology business. Or rather, they rediscovered some very old, founding-principle ideas that were gathering dust from day's gone by. (Does this sound vaguely familiar to any of you?) And Sun is becoming a different kind of company, one that is as interested in building communities as it is in making money. To Sun's new way of thinking, you don't try to sell products. You build communities, offering inexpensive or free hardware and software, and then you tax enterprises with services when and if they need it. This is the open source model, of course, as championed by Linux.

Earlier this summer, Sun took its Solaris 10 Unix variant open source, establishing the OpenSolaris project that now has 10,000 community members. About 1,000 of those people are the original founder members from inside Sun itself, while the remainder are real people who mostly have day jobs who are helping to create the features and fix the bugs in the code "Nevada" development release that will eventually become Solaris 11. The way the OpenSolaris community and the related Community Development and Distribution License for Solaris works, anyone can join the community and can port Solaris to any computing device their little heart desires. If you want to run Solaris 10 in a logical partition on your iSeries, you can start a sub-project, get a few cases of Jolt, a bunch of pizzas, and your best buddies in the iSeries community together, and have a go. In fact, Sun is hoping that you will do this, and I fully expect that some bunch of wiseguys are going to start working on a port of Solaris to the Power architecture, and maybe the MIPS architecture (which is used in embedded devices even though Silicon Graphics has abandoned it for its supercomputers and servers). OpenSolaris started at the end of June, and it has 10,000 members. This is a very large number of people for any open source project.

Having done the operating system, Sun is moving up the stack and down into the hardware, too, with its open source initiatives. A little more than 18 months ago, Sun packaged up its entire middleware suite--Web application server, identity manager, high availability clustering software, and so forth--and called it the Java Enterprise System. And rather than sell it as a per-processor or per-server product, Sun priced it according to company employees, allowing customers to pay once a year for the right to use the software. Initially, it cost $100 per employee per year, but Sun forked the stack, creating sub-suites that cost $50 per employee per year and then added in development tools to the full suite and goosed the price to $140 per employee per year. While sales have been difficult at first, the company inked a huge deal with General Electric and now has nearly 1 million seats and an annual revenue run rate of about $100 million. This is, arguably, small potatoes for a $10 billion company, but you have to consider this: Sun wasn't making any money at all selling the old way, and the new way was easier and arguably more profitable. Two weeks ago, Sun went crazy and announced that it was not only going to open source the entire Java Enterprise System middleware stack--including its compilers--but that it was also going to create an integrated platform it now calls the Solaris Enterprise System. All of the components you need, from the operating system, app server, middleware, clustering, security, compilers, Java environment, Tarantella client technology--the works--all integrated and all open source. For both Sparc platforms and X86/X64 platforms. Free to use, just like Solaris 10 and free to tweak just like OpenSolaris. Taking this whole stack of software to an open source model will probably take years, but Sun is committed.

But wait, it gets crazier. At the announcement last week in New York of its new "Niagara" T1 Sparc processors, which have eight Sparc cores, each with four processor threads, running on a single chip with four integrated memory controllers and PCI Express peripheral electronics, McNealy dropped a bomb on the crowd and said that not only was Sun very, very proud of the servers using the new T1 chip--which Sun claims can do five times the work of a Xeon-based server using one-fifth the electricity and taking up one-quarter of the space--but that Sun was going to give away the design of the T1 chip. Yup, open source hardware.

Now, I don't usually toot my own horn, but I have been arguing for years that that power usage and open source hardware are exactly what the IT industry needs (among other things). Like Sun, my thinking on this has evolved from being disturbed by the power consumption of modern servers to actually thinking that vendors were not listening and the only way to really get a new technology going was for us to do it ourselves. (For example: Lean, Mean Green Machines from February 2004 talked about power and server design issues, and Open Source Servers from March 2005 said that we should do with hardware what the industry has done with software. And, for the record, I started arguing for something I call Open Source/400 back in August 2002.) Imagine my joy at hearing McNealy make the announcement of the OpenSparc project in New York last week. (You can read our full coverage on what Sun has been doing in the past editions of The Unix Guardian, our Unix counterpart to The Four Hundred.)

"This is going to bend a lot of brains in the marketplace," McNealy conceded at the announcement, with his characteristic wry smile. When asked what Sun would do when and if rival Intel decided to grab the T1 intellectual property that Sun is going to let loose and then make a clone chip, McNealy quickly countered, "Wouldn't that be cool?" The audience was not exactly with him yet, so he explained a little further. "If it works in software, why wouldn't it work for processors?" He went on to say that by open sourcing the T1 chips, Sun could envision other operating system suppliers--including the open source Linux and FreeBSDs as well as suppliers of embedded operating systems--having an easier time tuning their software for the chips. And McNealy said that open sourcing of the T1 specs would allow makers of other devices such as set top boxes and embedded controllers to build variants of the chips, or if someone wanted to make an even larger, more powerful chip, they could do that, too. "You don't know exactly where it is going to go, and that is the beauty of it," he said.

Said as only a company founder could say, perhaps. You see, McNealy has seen this movie before. Sun embraced the open source BSD variant of Unix to make SunOS, the predecessor of Solaris, created and open sourced the TCP/IP protocol we all use today, created and open sourced the Network File System software that is one of the underpinnings of Unix, and created a very large community of commercial Unix customers where before it had customers who might merely be interested in buying a high-performance Unix workstation. As Sun says again and again, it is getting back to its open source roots. It is letting go, and it is building new communities--as it did with Java--that it hopes to some day make some money off of.

In making the Solaris Enterprise System announcement two weeks ago, Sun's president and chief operating officer, Jonathan Schwartz explained why Sun was letting go of a substantial part of its intellectual property. "We are going to be driving for volume first and then figure out how to monetize that volume with services afterward. We have to create a developer opportunity if we ever hope to have a market opportunity." The fact that Sun does not make very much money on software today is the best reason why Sun can get away with this tactic. With Sun's sales already substantially off and minor losses, Wall Street not expecting much in the way of a revenue boost or an improvement in profits, so now is indeed the time for bold experimentation. "If you are not going to pay for software, we would rather that you use our software," said Schwartz. "And we do not have any enterprise customers who will use software and not pay for support. Free and open source software is not about eliminating revenue, but eliminating barriers to revenue." If Sun is right, the upside could be huge. You get a large installed base of open source developers, who are, in effect, your main sales people inside the enterprise. And when enterprises are ready to deploy software in production, you charge for services relating to that software, but not for the software itself.

I am not so sure that this is the IT sales model of the future, particularly since it does not leave a lot of room for profits. But the volumes are on Sun's side if it is right. In about 10 months, Sun has distributed over 3.5 million registered licenses to the Solaris 10 operating system. Now, let's be honest here. Most of those are not in production, and only a percentage of them are actually resulting in paid-for services back at Sun HQ. In fact, I have 13 of those 3.5 million licenses right here at Guild Companies HQ, and so far, not one is in production--yet. Sun's Solaris installation and technical support pricing is lower than what Novell charges for SUSE Linux, and it could turn out that IT Jungle adopts Solaris rather than Linux on our next-generation Web systems. We do not use OS/400 servers any more because of the prohibitive cost of computing capacity of the iSeries line. The iSeries is not an affordable box for Web publishers, and it was not designed to be. Although it could be if IBM woke up and smelled the Java. . . .

Let's say for argument's sake that Sun is right. That, in effect, all of the hard work and money that you pour into software and hardware will, in the near or distant future, not have value, but that the value will be in the creation of a large community of developers who mix and match technologies and who drive enterprise services and who work with vendors like Sun to create a more tightly integrated hardware-software solution. Why, that is the AS/400. IBM may charge for the hardware, but that is not where the price tag really belongs, and we all know it. The value of the System/38 back in the late 1970s--and which allowed IBM to charge an order of magnitude more for the machine, on a per-MIPS basis, compared to a mainframe--was in the fact that it had an integrated relational database management system and a relatively flexible and easy development language, RPG, to harness that database. I am not making this up. In 1980, after the System/38 had been shipping for a while, it cost approximately $3 million per mainframe MIPS for a high-end box, while mainframes of that era (and using much less MIPS-hungry flat-file databases) cost about $300,000 per MIPS. It wasn't until the advent of the AS/400, in 1988, when relational computing on the AS/400 cost the same per unit of power as flat-file processing on the mainframe--around $100,000 per MIPS. And the value of that AS/400 box was still in the integration.

The problem with this value assessment is that, under Sun's thinking, IBM is asking customers to pay it forward for the value of the iSeries. Cash on the barrel head. Sun is taking a bolder--and possibly suicidal--approach and saying that it will get the volume first and then see who will pay money for the extra support services later. These are diametrically opposed viewpoints, and the price/performance differences between an i5/OS stack of hardware and software and a Solaris stack of hardware and software will be jarring once this all happens. How do you compare "expensive" (the iSeries) to "free if you are crazy" (those who use a Sun offering with no service) to "could be less expensive" (those who use Sun products with services).

If Sun is right about its talk of "the participation age" and open sourcing of its technology, then all of IBM's divisions, not just the iSeries, have a big problem. Is IBM likely to follow suit? Can it create Open Source/400? And if so, can it build a community of users that is large enough to sustain continued investments? How would the iSeries channel be affected when there could quite possibly be several vendors of OS/400-compatible servers? What happens to the OS/400 governors and Enterprise Enablement and all of the tricks in the iSeries that cost money if there is an open source version of the software? What is to prevent someone from porting DB2/400 and RPG to another platform? This begs a lot of questions, and I surely don't have any answers.

What I do know is that IBM had better be watching what Sun does. And IBM had better be ready to react--and react quickly and decisively--if Sun's ideas get roots in the data center. There is no guarantee at all that these crazy ideas that Sun is trying will work, but I have a suspicion that Sun is on to something that is bigger than just making money quarter to quarter. Sun is on a mission, and it has zeal in its eye. What Sun is doing is every bit as radical as what IBM Rochester did in creating the System/38 so many years ago. And that should be enough to get Big Blue to pay attention. Watch your back, Rochester.