Free Advice for Unix Vendors: Fujitsu-Siemens

Published: May 11, 2006

by Timothy Prickett Morgan

One of the great things about my job is that I get to speak my mind, to think out loud, to think about the possibilities. It is through the sharing of ideas that ideas can take hold, and then, perhaps if you are lucky and enough of us start thinking the same thing, change the world. Over the next couple of weeks, I am going to pick on each of the major Unix platform vendors, in no particular order, and give them a little free advice on how to better serve their customers and, therefore, in the long run themselves. This week, I plan to start with the Fujitsu-Siemens partnership.

Don't get all sensitive about it, Fujitsu, and I know you won't say anything, Siemens, as you haven't for about six years now. I pick on Sun Microsystems, IBM, and Hewlett-Packard all the time.

As the number five server maker in the world, the combined efforts of Japanese server maker Fujitsu and German IT supplier Siemens have created a considerable and respectable force in the IT industry, just as the partnership was heralded a decade ago when it was formed. But sometimes, I get the feeling that Fujitsu-Siemens is a bit like retailing giant, Sears and Roebuck--what did they do with Roebuck, anyway?, Or like Fox News putting hard-core conservative Sean Hannity up against barely liberal Alan Colmes. Does Colmes really get to speak, or win points off Hannity? I don't think so. Not that I watch television any more, or ever watched that show for more than a few minutes, it is so laughably lopsided. I prefer to do my own thinking, and my own talking.

Which brings me to my first piece of advice for Fujitsu: Just buy the PC, workstation, server, and storage businesses from Siemens and get it over with. Keeping track of which part of the Fujitsu empire, or the Siemens empire, or their partnership is doing an announcement or selling a product to what set of customers is just too annoying. Fujitsu is always complaining that journos misidentify the units of the company that are doing things. In my case, I am doing it on purpose to remind everyone that Siemens even exists and contributes to the Unix wares I write about. Why should Fujitsu get all of the credit if it hasn't done all of the work? And if it has, then just buy what's left of the Siemens IT biz and get on with it. If Germany thinks it needs to have an indigenous IT supplier, that is sooooooooooo 1970s thinking.

The next thing Fujitsu--remember, we just got rid of Siemens--needs to do is start supporting Solaris across more of its server product lines. I asked the top brass in North America for a comment on this very idea, and was given the cold shoulder. Which is fine, because that is what I expected, really. If that means some customers who have Sparc64-based PrimePower machines will migrate down to Primergy machines using X64 processors, so be it. Eat your lunch or be someone else's lunch, as they say. Solaris is, in many ways, a much better platform than Windows and Linux, and not supporting Solaris on the X64 boxes is just silly, particularly if you want to have customers grow into PrimePower RISC boxes some day, which do support Solaris. Rather than say Solaris should be supported across the entire Fujitsu server span, I hedged a bit. While Sun had created a variant of Solaris 8 that ran on the Itanium processors, and then never let it see the light of day, I am not so sure that porting Solaris 10 to the Itanium-based PrimeQuest machines makes sense. But, what the heck. Let's be consistent. So, Fujitsu, you should join the OpenSolaris community and work with Hewlett-Packard, NEC, and Unisys to create a Solaris 10 variant for the Itanium processor. If you start, and get such help, I am sure the OpenSolaris community will be all for it, and then Sun itself will pitch in.

Opteron, Opteron, Opteron. While Fujitsu has workstations and blade servers that use AMD's Opteron variant of the X64 architecture, the Primergy line is all about Intel's Xeon processors. This made sense five years ago, but it makes no sense now. All server vendors need to embrace Opterons, if for no other reason than no chip maker can seem to keep to a schedule most of the time. Your customers want to be able to deploy a wide variety of technologies. To put this another way: If your Solaris customers on PrimePower are looking to save money, what do you think they will do first--port to Windows or Linux or go buy a Sun "Galaxy" server that is based on Opterons and can probably support their Solaris workloads (particularly if it is homegrown code)? We know the answer to that one. It would be far better to offer customers a wide variety of choices, and to engineer servers from the get-go with this approach in mind. Why swapping out a Xeon motherboard for an Opteron motherboard is such a big deal is baffling, as far as I am concerned.

The real problem probably has more to do with how much money Intel and Microsoft throw around to server vendors for their promotions and advertising. Call me cynical about true motives in the server racket. I think the dominant chip maker and the dominant operating system maker have far too much sway over what technologies do and do not get adopted in the market. We don't have a free market so much as a semi-captive one. Which is why Microsoft was sued by Sun, and Intel was sued by AMD. (Sun settled because it had little financial choice, but correctly said that Microsoft was using its monopoly on the desktop to finance the building of a monopoly in the data center. AMD has claimed, in essence, that Intel's muscle with the PC and server makers has made it very difficult for it to peddle its Athlon and Opteron chips, which for years were technically superior to Intel's products. But no one would use them.)

Finally, please explain to us what "TRIOLE" means. In plain English. Using real words. If you say anything close to "adaptive computing," "on demand," or "real-time infrastructure" in your explanation, I will hang up.

"The possibilities are infinite," your advertisements say. Well, no they are not. But there are some things you could do to make your products more broadly appealing, like the things I outlined above. There is even a good chance you could move up to the number four position in the server market, if you start pushing aggressively and adopting a broad spectrum of technologies. Maybe even number three. Because if I know one thing for sure, Dell isn't about to do any engineering or offer a truly broad hardware and software stack.