The Coming Demise of Sun Microsystems

I started working on this essay almost a year ago, in February of 2002. I've hacked at it off and on. Added a few web references now and then. An increasing number of these references include speculation on the demise of Sun Microsystems. I've also gotten some experience recently with the RedHat Linux release. So, in the spirit of piling on and kicking 'em when they're down, I figured I better finish this essay before Sun disappears, rendering this essay meaningless.

The Past: The best development platform available

I have been using Sun Microsystems workstations to develop software since they were based on the Motorola 68000 processor. This was a time before the first Sun SPARC chip, when RISC architecture was a topic of research papers. Back in those days Sun Workstations were the best development platform available and software engineers lusted after the latest Sun hardware.

Sun Workstations ran a version of UNIX based on Berkeley UNIX, which was the best version of UNIX available. The Sun Workstation had a big monitor that could display lots of windows and it worked well on a network using TCP/IP. Sun systems supported the Network File System, making file sharing easy. Sun even had a graphical debugger. Compiles were fairly fast, since you had the whole system to yourself, rather than time sharing a mini-computer.

If you did not have the good fortune to have a Sun workstation on your desk, the alternatives were not attractive. They included VAX mini-computers like the VAX 11/780 or, a few years later, the Microvax workstations. You could run Berkeley UNIX on a VAX 11/780 and DEC finally got the UNIX message and released their own version of UNIX (Ultrix) on the Microvax. But DEC systems were never as good as Sun, and competition from Sun was one of the contributing factors to the demise of DEC.

IBM came out with a UNIX based workstation based on an early RISC processor, but it was a sad system, with very poor performance and appeared to lack any commitment on IBM's part. In fact back then IBM looked like a faded giant in decline. In contrast, Sun was a young vigorous company.

In the days of the early Sun SPARC systems (the Sun 4), the Intel based IBM PC architecture computers were considerably slower than workstations. These systems ran Microsoft DOS, an operating system that was beneath contempt or an early version of Windows, which was buggy, did not have enough memory and was barely adequate for developing simple graphics applications.

The only computer systems which were close to being as cool (or maybe even cooler) as a Sun workstation were the workstations made by Silicon Graphics (now called SGI). Silicon Graphics systems had (for the time) blazing graphics performance and supported 3D graphics. But these systems were too expensive for engineering applications.

Intel based systems become powerful computing platforms

In the terms of the computer systems that existed during Sun's golden years, you can now order a desk top supercomputer from Dell. The performance that is delivered by Intel processors is so high that most people who are not doing heavy numerical simulation work have more than enough processing power.

The cost of developing microprocessors is so high that only Intel and IBM seem to be able to afford leading edge silicon foundaries (note that Advanced Micro Devices, AMD, has partnered with IBM). Along with the cost of fabrication facilities, the cost of design has gone up, as the device density and complexity of processors has increased. All these factors mean that Sun will not be able to compete in the long run with Intel architecture microprocessors. This will drive Sun to adopt the Intel architecture, which will, in turn, lead to direct competition with Dell and other low cost manufacturers.

Java, a marketing campaign that took over Sun Microsystems

Sun is probably most widely known for Java. This was the result of a huge marketing campaign behind Java during the "dot-com" boom. Java and the marketing campaign put Sun and Sun's CEO Scott McNealy "on the map". Sun even ran adds in publications like Vanity Fair.

Before all the Java hype, if Scott McNealy was trying to help push a big sale by calling the CEO of a company, the response would have probably been "Sun Microsystems? Scott McNealy? Never heard of them." After the Java hype, people returned McNealy's phone calls.

At the same time that the Java hype was fueling Sun Microsystems name recognition, companies were starting to develop a web presence and some were developing e-commerace or business to business (B2B) systems. Sun sold large servers to many of these companies.

One of the results of the Java assisted boom in hardware sales was that the Java hype took over Sun. Sun marketing executives and flacks ran around making statements like "Java changes everthing". There were claims that Java would replace the Microsoft platform. Vast amounts of money was thrown at Java projects within Sun. Some of these projects resulted in innovative and exciting technology (e.g., JINI and Java Spaces). But the software that these projects produced was more like research software written by graduate students than high quality product grade systems.

The sad state of Sun's developer tool set

Sun got its start with software developers. With the Java boom inside of Sun, the Sun compilers and debuggers, which were already lagging badly behind Microsoft, were starved for resources. The Sun C++ compilers were buggy and the debuggers compared poorly to any high quality integrated development environment. Sun neglected its core user base. The users that would develop new applications for the Sun platform. Even in the case of Java, Sun's development tools and runtime environment lagged badly behind IBM and HP.

Linux as a UNIX standard

For various reasons I was not inclined to be a fan of Linux. This includes a strong bias toward the BSD and freeBSD operating systems. However, at the end of 2002, I installed Linux on a system at work as part of a project. I was really impressed. The install was very easy. And the Linux environment is very nice. Much better than I expected. In fact, much better than the Sun Solaris system I have used for many years.

IBM had made Linux a current platform and has stated that it will replace IBM's own version of UNIX, A/IX. The strong support by IBM and the adoption of Linux as a platform for web servers means that Linux is an increasing threat to Sun.

Sunset: Sun Microsystems follows the path of Digital Equipment Corporation

Digital Equipment Corporation struggled against fate for two or three years, but the company was eventually pulled under by the rip-tides of technological change. The same process seems to be taking place with Sun Microsystems. The combination of inexpensive high performance Intel architecture systems running Linux look like they will destroy Sun as a hardware company. Companies like Dell have become experts at operating on thin profit margins. Margins that are too thin for Sun.

The Apple Model: What could have been done differently?

Sun cannot base its future on a hardware strategy that is based on competing head-to-head with companies like Dell. Sun can protect their market by providing excellent software which is not available on other systems. This is the strategy that Apple has successfully followed. Even if Sun chose to follow such a strategy, it appears that it is too late. Sun software reminds me of software developed by graduate students. Innovation has largely disappeared from Sun's software and the quality has been poor in many cases.

Postscript

Scott McNealy has these chipmonk teeth and a rather substantial ego (while the teeth are non-standard accessories for a CEO, the ego is definitly a standard CEO feature). I have a few comedy rifs aimed at McNealy that I subject my long suffering wife to. She asked me the other day why I pick on Scotty. One reason is that there is something there to satirize. Bill Gates can be seen as a sort of "grey man. Gates has few publicly known traits (other than the famous rocking) which you can pick on. Gates sits in his 30,000 square foot palace on the shores of Lake Washington quietly plotting world domination, while McNealy is up there on stage making noise.

The other reason that I pick on McNealy is that I'm disappointed. I remember how great Sun Microsystems once was and how exciting their products were. I don't welcome the demise of Sun. In fact the best sales guy I've ever worked with (Mr. Bill, you know who you are) worked at Sun last I heard. But they seem to have taken too many wrong turns. Sun could have invested heavily in their core software (e.g., not just Java). But they didn't. Sun could have realized how important a leading edge development environment was, but they did not do this either. So at this point, things look grim. It looks like Sun will fall to the combination of Linux and Dell. As many, including McNealy, have pointed out, Sun Microsystems has emerged from tough times before. I hope that they pull it off this time too. But I would not invest in their stock betting on this.

References

Ian Kaplan, February 2002
Revised: October 2003