Linux Today: Linux News On Internet Time.

More on LinuxToday

The Retreat: The Loss of the Portability Battle

Aug 27, 1999, 13:41 (23 Talkback[s])
(Other stories by Paul Ferris)


Desktop-as-a-Service Designed for Any Cloud ? Nutanix Frame

[ The opinions expressed by authors on Linux Today are their own. They speak only for themselves and not for Linux Today. ]

By Paul Ferris, Staff Writer

It's been an interesting month. First, you have SGI dropping their support of Windows NT and then next you have Compaq dropping their support of Windows NT on Alpha. SGI scores points here for being the first in the trend, and I think Compaq scores points for making the biggest shock wave. The two events are related in terms of what they spell for the future of NT. SGI was at the outset, hoping to create hardware enhancements that would plug into NT the way they had with IRIX, their Unix based operating system. DEC was the last architecture that would support NT on a non-Intel chip.

Whether or not Microsoft will admit it or not, it's a retreat in the battle for the future computing platform. By losing the support of two major computer vendors, Microsoft has to do something. I say that that something, given the eyes watching them and the long-delayed release of Windows 2000, is spin.

Windows 2000 may be on time. Windows 2000 may be Microsoft's next generation operating system. It may or may not be feature rich. It may be their last hope. It may be a lot of things, but at the moment, one of them it is definitely not: Hardware Portable.

Hmmm, I wonder what operating system provides a stark contrast to that limited choice and lack of portability? Do I need to say? Linux's support of hardware architectures is nothing short of record setting(*). I'm not even going to start listing hardware architectures, my articles are long enough as it is.

If you are in the business world and at this point are pondering a new server operating system for the future I'd recommend that you ask yourself a few questions:

  1. Am I willing to limit my hardware buying choices to one architecture alone?
  2. Am I willing to limit my software buying choices to one operating system, that only runs on one architecture?
  3. Am I willing to limit my new choices to those provided by a single company that is taking ever longer to develop changes in a world where change is accelerating? Beyond that, change and adaptation to change are tantamount maintaining the status quo.

Now for the easy part. The answers: No, No and No. Microsoft promised a lot with the advent of Windows NT. A lot of the promises boiled down to: "We are going do everything that Unix does."

Their success compared to many of the key strengths of Unix and Linux is debatable. A lot of the issues get compared because Unix was supposed to be NT's competitor at the outset. One of the issues, however, they cannot debate. They cannot promise this aspect of enterprise computing with Open Systems or Linux - no way, no how. That one item helps to provide a healthy, competitive market that fosters innovation and growth. In the absence of such a market, choices go away.

In this case, Microsoft is not just asking people to forgo choices in software, it's now asking for people to limit their hardware choices as well.

By promising to kill off Unix, their perceived competitor at the time, Microsoft was really saying that they were going to limit choice at the high end, like they had been doing on the desktop. In order to come through totally on all of these points, they had to promise hardware portability.

Let's cover the retreat, in total. NT was going to support DEC Alpha, SGI MIPS, Motorola PowerPC, Sun Sparc (Yes, at one time, even that) and of course Intel.

This was because at the time, Intel x86 hardware pretty much covered the low end of the computing spectrum, as it pertained to raw power. For the moment, this gap appears to have closed somewhat. But it's a gamble to think that it will stay that way.

NT had to be able to run on new processor generations without much work. In other words, it had to be able to do what Unix clones and variations had been doing for years. Portability is essential when it comes to switching to better computing platforms and embracing new exotic hardware.

If you've studied the work of Alvin Toffler (Future Shock), you may recall that a good portion of his writings deal with change and the rate of change. He was right about the fact that we are on an exponential curve of change.

The Internet is accelerating what was already a blistering pace. Change is happening faster now than it has ever happened before, and of course even that rate is ever increasing.

New processor architectures and computing ideas are spawning at a high rate of speed, and the next big thing, in terms of execution speed or whatever very likely will not be Intel. It might be some other company. Who knows? But, lets suppose for a second that two changes happen at one time.

Let's suppose that we suddenly have change in the software market, say a new operating system that provides a quality, portable base of software. Let's say for the sake of argument that this new O/S also turns out to have qualities that make it shine on the Internet. Let's suppose that suddenly a new and/or faster computing platform arrives as well.

Let's suppose further. Let's call that operating system Linux. As for the processor, maybe it's an IBM chip. Maybe it's Compaq's Alpha design. Maybe it's Intel's Merced. Maybe it's HP's PA-RISC chip. Maybe it's an unknown, that's been holding back until now. There are a lot of possibilities.

I know, I know, a lot of maybes there. Maybe too many maybes. But it's not a very big stretch of the imagination in this day and age is it? Just one more maybe, and I'll move on. Should Microsoft not figure out how to get their Merced port working, they will lose support on the final platform that they support today: Intel. Today, they cling to this platform like a cliff hanger with one pinky on the ledge. And Linux already runs on Merced. No maybes there.

Should any of the above maybes come to fruition, you suddenly have Microsoft chasing software and hardware tail lights.

Maybe Microsoft feels that they didn't need to win the hardware portability battle. I can't imagine that, what with all the noise about any place and any device, but maybe they think that everyone was just going to buy Intel, so maybe they got that base covered.


But one thing is for sure. They've lost the hardware portability battle. That victory belongs to Linux.

(*)The hardware portability record is held by NetBSD, according to sources that I must admit didn't seriously flame me for being wrong. Thanks guys!