OS/2 and the Real World

Sponsored by


In the eyes of the public at large, there is really only one operating system for IBM-compatible personal computers: Windows. The general opinion is that there may be several varieties of Windows, but it's all really the same thing. In other words, "operating system" and "Windows" are synonymous. We might liken this situation to hamburgers at McDonald's: no matter how hard McD's has tried to cultivate the idea that there is a radical difference between "Arch Deluxe" and a 45-cent hamburger, everybody knows that it's all made of the same stuff. It's all mediocre. It's all cheap, fast food for people on the go. People are generally to busy to care what is the relative content of beef, soybean, lettuce, cheese, and other ingredients. They just grab their burgers and drive away.

However, there are times and situations when people want something better. They want to sit down and think carefully about the contents of their meals. They want to dine on choice food, carefully prepared, and to enjoy and savor it. They are not in a hurry; they don't mind spending a few dollars extra to get higher quality. Similarly, there are occasions when people want and need something better as an operating system in their PC. Under what conditions is this selection process likely to occur? When and why do people prefer OS/2 Warp -- and actually take the extra time and make the extra effort to upgrade to a world-class OS such as OS/2?


Several years ago, it became fashionable to perceive OS/2 users as some sort of anti-Microsoft zealots. The "conventional wisdom" was that OS/2 users were merely an "ABM" (Anyone But Microsoft) crowd. In other words, the belief was that OS/2 was not being chosen on the basis of any inherent technological superiority, but rather as an overt act of defiance against Microsoft. It was supposedly because they were "jealous" of Mr. Gates' wealth and influence, or because they were some kind of "conspiracy theorists" who feared an imaginary "Big Brother" controlling their computers. This explanation was a handy crutch for computer-industry pundits whose amateurish lack of insight into the technologies they wrote about was in danger of being exposed. If you don't understand the reasons why OS/2 is a superior technology to Windows, it's easier to just spout some psychobabble and do a little name-calling. It makes for a good story. And for a small but vocal minority in the PC media, the truth never gets in the way of a good story.

In reality, that small branch of the family of OS/2 users which consists of vocal anti-Microsoft activists has decreased in size over the past few years; many of them have switched to Linux. For the real "ABM" types, it really doesn't much matter what OS they are using, as long as it's not Windows. Of course, other OS/2 users have switched to Linux for technical reasons such as availability of a particular application, or access to the source code. And some have just gotten fed up with IBM's lack of commitment to the individual and microbusiness users of OS/2. But the clear majority of OS/2 users have stayed with the product because of its inherent excellence, its simplicity, its flexibility, its reliability. The typical OS/2 user has always been motivated by technical and economic issues over emotional ones. Even those remaining OS/2 fans who speak out against Microsoft do so mainly to complain about the shoddiness and wimpiness of Microsoft's products, or about the bullying tactics that Microsoft continues to use to prevent superior competing products from becoming profit centers for the more innovative, quality-oriented companies out there.


There are several reasons why OS/2 has always been a technological giant compared to Windows. Here are a few of the more obvious ones:

1. Reliability. Most people readily admit that Windows is cranky, cantankerous, and unpredictable. Yet that is only from the user's perspective, based on regular software crashes. There are deeper issues involved. For example, the Windows9X systems are DOS-based; they must switch from 16-bit DOS real mode to 32-bit protected mode dozens of times per second. This switching process puts a lot of wear and tear on the CPU and other system components. In addition, Microsoft has disabled the standard CPU clock cycle mechanism recommended by Intel: 3 cycles on, one cycle off for cooling purposes. By removing the cooling cycle from Windows, Microsoft was able to speed up the apparent performance of these OS's by 33% (1 extra clock cycle available for every 3 that other OS's would use). Yet this did not make Windows faster than other OS's; it merely reduced the performance gap that would otherwise have made Windows a laughingstock. Windows is still a slow operating system, but not as slow as it otherwise would be.

The disadvantage of this trickery is that CPUs running Windows operating systems tend to operate at 10 degrees to 20 degrees hotter than CPUs running "normal" operating systems that don't cheat against the Intel specification. This extra heat not only wears out the CPU faster, but also spreads throughout the inside of the PC case, threatening to reduce the life expectancy of other heat-sensitive components such as RAM, motherboards, VGA cards, and power supplies. OS/2 proves to be far more reliable in terms of its treatment of the supporting hardware, because OS/2 doesn't use sleight-of-hand; it is always running in 32-bit protected mode and uses emulations to provide its backwards compatibility with DOS and Windows 3.X applications. Therefore, OS/2 systems are destined to provide longer life and lower TCO (Total Cost of Ownership) over Windows systems.

Additionally, the inherent software architecture of Windows is risky, convoluted, and unreliable. This factor not only contributes to the poor reliability of Windows, but also to the complexity of operation and maintenance. These issues will be addressed below.

2. Modularity. While Microsoft has been intent on blending "everything but the kitchen sink" into the Windows operating system for the purpose of leveraging and extending its monopoly position, IBM has taken the superior approach of providing modular subsystems inside OS/2 that can be changed or replaced at will, without threatening the integrity or the reliability of the operating system. When OS/2 version 3.0 became the first consumer OS for the Intel platform to provide a Web browser bundled with the OS, it was indeed "bundled" instead of "blended" or "merged". People could load and operate OS/2 with or without the browser (IBM Web Explorer) with equal success. There was no performance penalty for installing the browser, unlike the penalty that Microsoft operating systems suffer from due to the interlocking, shared elements of MS Internet Explorer and the underlying Windows OS's.

Similarly, the Windows9X OS's employ a method of long filename support called VFAT (Virtual File Allocation Table). Because Windows will not run without VFAT, the file system is locked in. It becomes risky to add support for multiple file systems on Windows. However, OS/2 provides the IFS (Installable File System) approach so that a wide variety of file systems and devices can be supported. Any new file system that is invented can be added to OS/2 without threatening the integrity or the reliability of the system. Even Microsoft's proprietary FAT32 system has become available by simply loading a device driver IFS in the OS/2 config.sys file. Meanwhile, Microsoft's narrow, closed approach has continued to limit the options of Windows users. Windows2000's proprietary file system is not backwards-compatible and is not removable once it is installed. This lack of flexibility makes support costs and migration costs much higher for Windows platforms than for OS/2.

3. User Interface. The user inferface provided by OS/2 is powerful, reliable, and simple to learn. It was a joy to walk into my boss's bank (AmSouth) last week and watch the PC's in the VIP offices running OS/2. While corporate training centers number into the thousands for Windows systems, there does not seem to be a need for training people on how to use OS/2. There are no dancing icons or other distractions to baffle and confuse people. OS/2 doesn't lose track of files and leave empty icons the way Windows does.

Best of all, OS/2's modularity allows a wide variety of user interfaces for the same system. Freeware and commercial enhancements abound. Because the OS/2 interface is almost infinitely customizable, OS/2 can be all things to all users. And unlike Microsoft, IBM does not restrict the PC companies from providing alternative GUIs if that is what their customers want.

4. Performance. Not only does OS/2 consistently provide Java performance at or near the top of the charts, but applications ported from legacy Windows systems tend to running noticeably faster on OS/2 machines. OS/2's multithreading beats mere multitasking by allowing each program to access and interact with multiple system resources simultaneously. Processing time and overall system speed will almost always be faster on an OS/2 PC versus a Windows PC when both are outfitted wtih identical hardware. Beware of bogus comparisons in which a Windows machine has far more RAM, a faster processor, or a brand-new hard disk. If you compare apples to apples, OS/2 is faster.

But the OS/2 advantage gets better over time. Due to the goofy architecture of Windows and its Registry, the more applications are loaded on a Windows PC the slower it becomes!! Yes, Windows PCs gradually lose performance over time. The FAT-based file systems are also notorious for file fragmentation, which gradually gets worse as a system is used month after month. Windows does include a file defragger, but this program is very slow compared to the elegant OS/2 solution: OS/2's HPFS (High Performance File System) reduces or prevents fragmentation in real-time. When PC's were much slower 5 or 10 years ago, this overhead was noticeable. But in todays' high-speed systems the HPFS overhead is negligible. Since OS/2 does not rely on a complex Registry and an outdated FAT file system, OS/2's superior performance does not degrade over time.

5. Backwards compatibility. Unlike Microsoft's treadmill approach to forcing changes, IBM believes in maximizing the value of previous investments. This means that older DOS and Windows 3.X applications run just fine on OS/2. Microsoft is intent on "burying DOS", they have said, and they also fail to support older Windows versions. Yet the Windows 3.1 module of OS/2 is even fixed for Y2K compliance, according to IBM. As a result, people do not have to get on a constant treadmill of buying new software every few years. Since older software does not wear out, this means that training costs are also reduced. Many companies fail to realize this benefit because they do not account for the life-cycle costs of higher TCO caused by Microsoft's manipulative approach to software marketing. TCO is a bogus statistic if it does not factor in the mandatory, Microsoft-dictated shortness of product life cycles.

Furthermore, IBM's modular approach improves the compatibility of multiple applications on the same system. Microsoft's risky "merged" approach leads to DLL files (Dynamic Link Libraries) that overlap among multiple applications from different vendors, and with DLLs for the Windows OS's themselves. Since version numbers and version dates are different for these DLLs, many Windows installations can be damaged simply by loading a new Windows application onto the system!! Often the only way to make a particular set of Windows applications operate together is to carefully install them in a certain order. This process must be repeated if just one of them becomes corrupted and must be reinstalled. This is an extreme source of frustration and wasted time, as well as reducing system reliability and compatibility.

There are a dozen other reasons why OS/2 is technologically superior to Windows. However, the above explanations make the choice clear: OS/2 is a competitive advantage for the businesses who use it, and a time-saver and money-saver for the individuals who use it.


Who would reasonably oppose such a superior system as OS/2? Of course, ruthless competitors such as Microsoft have no regard for issues of reliability and compatibility with competing products. There are also splinter elements within IBM whose rivalry toward OS/2 supporters encourages them to downplay or ignore OS/2. But these are not the main sources of apathy and opposition toward OS/2. Instead, the resistance mainly comes from those who have a need to preserve their reputations as all-knowing gurus of the computer world. This class includes certain members of the computer industry media, but more commonly it includes PC consultants and dealers themselves.

Yes, there have been a number of blowhards and windbags who have used their soapbox positions to rave against OS/2, or to purposefully ignore it. There is nothing more frustrating to a technical columnist than to be constantly corrected by his or her own readers. Since OS/2 users are for the most part motivated by their knowledge of OS/2's technological superiority and lower costs, they are likely to be more knowledgeable about OS/2 than those who would be writing about it. Instead of taking the time and effort to become informed and educated about OS/2 and OS/2 products, the mouthy media miscreants have decided to marginalize OS/2 in hopes that they will never have to wise up and learn. One such crass remark was made by PC Computing's Ed Bott, who wrote that OS/2 users should "break out the grape Kool-Aid" -- an obvious reference to mass-suicide cultist Jim Jones. Others have been more subtle but just as determined.

However, even the PC media itself pales in comparison to the inertial response of the PC dealers and consultants. It is these front-line people who have betrayed the trust of their clients, their associates, and the public by stubbornly sticking their heads in the sand and providing Windows products instead of the superior OS/2 solutions. These are by and large the same people who fear Linux and hope that it never becomes mainstream. What is their motivation in continuing to support Microsoft and its bug-riddled, third-rate products?

For the most part, the lure of a steady stream of $100-per-hour software fixes is too big a temptation for many of them to resist. Windows provides plenty of opportunities to charge for formatting a hard drive, reloading an operating system, tweaking some device drivers, and loading application software. All of these processes have been somewhat automated by Microsoft, to the point that most of the time is spent simply waiting for the activity to finish. Consultants do not distinguish between a $100-per-hour charge for formatting a hard drive, and a $100-per-hour charge for a complex networking setup. PC repair shops such as CompUSA charge consumers $90 per hour for this simple, repetitive task. Lost icon pointers, missing files that must be replaced from the Windows CD, and thermally-related component failures all generate a steady stream of repair dollars.

Additionally, there is a relatively large base of Windows applications written for vertical markets such as doctors, lawyers, accountants, and contractors. This means a relatively easy "capper" sale on top of the regular sale of a PC and some installation and networking charges. There is no need to learn programming, to design a customized solution, or to provide any sort of creativity to the process of providing an off-the-shelf solution. Therefore, people of relatively modest intellectual backgrounds can masquerade as wise men and deep thinkers to clients and consumers who don't understand these marketplace dynamics. The entire process squeezes out the customization, the creativity, and the excellence that would be provided by a company focused on quality instead of mediocrity.

Therefore, the majority of PC consultants view Windows as a "sure thing" and a lot of easy money. They view anything more reliable than Windows -- which will cut deeply into their income from repair, maintenance, and replacement -- as a threat to their comfortably numb condition. OS/2 has the software equivalent of the "Maytag Repairman" problem: it works so good that it's possible to put yourself out of business by selling too much of it. I know of one concrete mixer company that needs OS/2 service only once every six months. My mom has had her OS/2 PC for three years of almost daily use, and it has never crashed once. Given a choice between a system that requires learning but provides almost no recurring maintenance income (OS/2), versus a system that is already in place and requires regular repairs (Windows), most consultants choose the path of mediocrity and just keep doling out Windows-based systems.


At Warpstock97, IBM's Jim Koerner provided the IBM viewpoint on just who benefits from OS/2. The IBM model is based on the fundamental unit of commerce, the transaction. IBM sees the number of transactions per second (tps) as the deciding factor as to which technology a buyer will use. (See figure 1 below.) Since the cost per transaction must become extremely low as the tps rises in order to maintain overall profit margins, downtime at the high end of the curve must be minimized. The ultimate example of a low-overhead, low-margin, high-volume enterprise is a banking system. There may be tens of thousands of transactions per second on a banking network, and OS/2 is ideally suited to providing high-volume transaction processing with minimal downtime and also lower MTTR (Mean Time To Repair).

FIGURE 1. Transactions-per-second (tps) versus Cost Per Transaction

Since the Internet and e-commerce are providing heavier and heavier transaction loads to networks and servers, OS/2 is ideally positioned to provide the reliable, high-performance processing and high throughput that the banking and financial industries will require. The global consolidation of financial, insurance, banking, and investment corporations is further enhancing the OS/2 advantage.


While IBM is greatly interested in climbing up the left side of the curve -- to ever-higher transaction rates, with ever-higher profits -- the average consumer is located on the right side of the curve, at approximately one transaction every one to ten seconds. Writers, artists, and other providers of time-consuming creative works are often located at the farthest reaches of the rightmost edge of the graph. They may be in thoughtful contemplation for minutes or even hours before entering the next bit of data. Most small businesses involved with instrumentation, process control, or other machinery will be found somewhere near the middle of the curve.

Not coincidentally, there is a close correspondence between the type of computer equipment used by each type of client, and where they are located on the transaction curve. On the far right are the MacIntosh users; Macs have historically been very user-oriented but with poor multitasking. An artist or a writer does not need the high rate of simultaneous processing that a bank or a factory controller requires, but they do need a user interface that provides them with a fully-implemented set of tools. A little farther to the left is where Microsoft has planted its flag, hoping to spread to the right and engulf the Mac world, as well as growing to the left and someday being able to provide enterprise-class transaction processing. The world of industrial control equipment has typically involved Unix machines, which have need for high reliability but little demands for user interface "fluff". At the very highest reaches are the mainframes and the OS/2-based servers that IBM uses in providing extremely fast and reliable transaction processing, but with little need for user-oriented features.

From this graph we see that historically Mac and Windows users have had to settle for lower processing capabilities in return for the user-oriented factors of their environments. Unix users have had powerful machines but without the fancy screens of the Windows and Mac worlds. OS/2 is therefore in a most interesting position; it can provide the "bells and whistles" if the user needs that, but it also provides the mainframe-class transaction support that power users crave. The goal of productivity growth for individual users is to ride up the curve toward the left, increasing the transaction rate and therefore providing more output in less time. As processing needs increase, this is the effect that IBM CEO Lou Gerstner was talking about when he said, "Everything's moving back our way now" a few years ago. The real battle is therefore between transaction-heavy systems trying to move down the curve and take over the right side of the curve -- where margins are much higher, versus user-oriented systems attempting to slide up the curve to the left and thereby increase volume while attempting to maintain higher profit margins.

Therefore, most computer users who work outside of the high-transaction commercial enterprises are missing out on the benefits of the OS/2 platform. That is, Microsoft's monopoly and Apple's decision to employ a less-robust internal architecture have historically placed non-enterprise PC users at a performance and/or reliability disadvantage.


On the surface it would appear that the transaction-oriented systems would have a serious advantage; it is easier to make a reliable system user-friendly than it is to make an unreliable system or a low-volume system into a powerhouse. However, the factors involved in the decisionmaking process for the purchase of computing solutions are also very dependent on the type of person and/or business inhabiting each section of the curve. In other words, just because OS/2 appears to have all of the desired qualities to fit anywhere on the tps curve, this does not necessarily mean that OS/2 will have a large share everywhere on the curve.

For example, look at the Mac end of things. Transaction rates are simply not a consideration for the core constituency of the Mac user base. Yet even for those Mac shops or Mac power users in need of more "juice", Apple now offers the powerful G4-based PowerMacs. Along with a new OS based on the NextStep kernel, Mac users are sure to get the power and the capabilities they have long been denied. So it makes little sense to argue in behalf of OS/2 in the Mac part of the curve. Mac people have always been known to "think different", as the slogan goes, so in all likelihood they will "stay different" now that Apple is once again on a forward track. Decisionmakers on the right end of the curve will choose the "most different" solution they can find, and that is the Apple product line.

Decisionmakers at the high end of the curve are going to pay whatever it takes to get the transaction processing they need. This means they will get OS/2 or Unix or some mainframe solution. These people will make their decision based on banking culture, not mainstream culture. How "popular" OS/2 is, or is not, is simply not an issue with them. This is IBM's bread-and-butter world. How OS/2 succeeds on this end of the curve is almost totally up to IBM. OS/2 will continue to provide the highest level of performance, reliability, and capabilities on a per-dollar basis of any of the enterprise-proven OS's. While it is possible that Linux may continue to improve to match the enterprise level of performance, it is doubtful that banking culture will accept it as readily as people on other parts of the curve will.

The general-purpose PC market is strongly dominated by Windows. This market includes home users, SOHO (small office/home office) users, and small workgroups inside major companies. These are the people who are most suffering from the lack of exposure to OS/2's superior enterprise-class attributes. It is also this market whose cultural decisionmaking is based mostly on the relative popularity of a particular product. Since Windows occupied this space first, it became the default decision for these people. It is quite difficult to dislodge a product which monopolizes this sector, since the decision to purchase is most accurately termed "think same", which is the opposite of the Mac mantra. "Think conformity", or "think mainstream", are the words that readily come to mind to describe the mindset of a market that has almost been beaten into submission by the Microsoft machine and its media lackeys. To enter this market successfully, OS/2 must succeed elsewhere, in some visible, respected way. (Running banking networks is certainly respectable, but it is not visible.) The cost of dislodging the default OS (no matter how miserable its technological condition) is simply too high for a direct confrontation. Here is where a low-entry-cost option like Linux may be effective in opening people's minds to the superiority of MS-free alternatives, because the perceived risk of a "free" OS is very low.

The remaining segment of the curve is the business automation/manufacturing area of the marketplace. It is in this class of small-to-medium business (SMB) that OS/2 can and should have the greatest short-term success. However, IBM is not focused on this sector because of the need to avoid cannibalizing its AIX and other Unix-based solutions. If a well-financed group of third-party consultants were to attack this market with gusto, it is quite possible that OS/2 could succeed and prosper here, and that its success would in turn enable a successful move into the general-purpose PC market. Knowing this, Microsoft has attempted to position its decrepit WindowsNT line of products here in advance. The key factor for OS/2 success in this market segment would be to recognize and appeal to the cultural decisionmaking process of this segment's buyers. This is probably where more research and dialogue needs to focus in the OS/2 community.


Technologically, OS/2 has the finest set of capabilities available in today's PC OS marketplace. Culturally, however, OS/2 is at a disadvantage with all but the biggest of the high-transaction users. It is not "different enough" for Mac users, and not "mainstream enough" for most PC users. However, if a serious, well-supported business case can be made in the SMB market, OS/2 can and should be able to move down the curve from the mainframe world, as well as up the curve from its niche SOHO market, and later begin extending its reach toward the mainstream markets. The key will be to identify and relate to the cultural and technological issues affecting buying decisions within the SMB segment.

The ironic situation of the OS/2 dealer, consultant, or developer is that OS/2 does not "break" or cause a constant treadmill of service income the way Windows does. As a result, it is difficult to attain the firm financial foundation necessary for successful, profitable consulting with OS/2. We OS/2 supporters will need to watch and learn from the Linux community, to see if they can find the "traction" necessary to make Linux consulting a self-supporting enterprise. OS/2 has all the technological features necessary to successfully occupy any part of the tps curve. What is now needed is the element of cultural and commercial support.


[ Top | Home Page | Faq ]

Copyright © 2000, Tom Nadeau.
All Rights Reserved.

E-MAIL: os2headquarters@mindspring.com