OS/2 eZine - http://www.os2ezine.com
16 January 2002
Douglas Clark is a program management consultant who first started using OS/2 version 1.3. He's married, with 2 girls, and is old enough to remember when 4 color mainframe terminals were a big thing.

If you have a comment about the content of this article, please feel free to vent in the OS/2 eZine discussion forums.

There is also a Printer Friendly version of this page.

Previous Article
Next Article

OS/2 Netlabs.  Open source software for OS/2.

Why Not Linux?

By now most of the readers of the OS/2 eZine are aware of IBM's strategic direction for OS/2. The most reported aspect of the document is the official IBM position of not supporting OS/2 in convenience packs or public fix packs after December 2002, to quote:

"Following the expiration of Program or limited warranty defect support, IBM plans to offer fee-based defect support. We anticipate that over time the price of defect support will increase and might be charged on a per license basis."

However the most interesting part of the document, and the main thrust of the document is the advice from IBM to its customers to stop using OS/2. To quote again:

"Briefly, IBM recommends using Internet technologies on both internal and external networks with server-centric business logic delivered by thin-client applications. Customers should exploit OS/2 e-business enhancements and deploy new e-business technology applications concurrently with existing OS/2 applications until platform neutrality has been achieved, and then change the operating system." (Emphasis is mine.)

What IBM means by internet technologies is (again quoting):

  • Java - for program portability.
  • XML - for data portability.
  • Internet protocols - for data transmission and communication control.
  • Browser - for user interface.
  • HTTP Server - for an HTTP Server with proxy and caching.
  • Domino and WebSphere - for application serving.

The internet technologies part is interesting because it reflects one of IBM's corporate directions, namely that all applications should be written in Java so they are platform neutral, or platform independent. That corporate direction was stated early in 2001 as the reason why IBM was getting out of the C/C++ compiler business for Windows and OS/2. It is also very manifest in the articles that are published in IBM Systems Journal, in the emphasis on the Developer's Connection (or whatever the current name de jour is,) and in the software that IBM is now pushing. This platform neutrality direction is driven by the internet craze, where the thinking was that the internet was going to be the sole method of commerce between business and customers, that it was going to grow at explosive rates forever, and that therefore all other types of business computing needs were irrelevant and needed to be converted to "eBusiness". Unfortunately for us IBM customers (and it appears IBM,) that bubble popped a little over a year ago. It should now be clear to everyone, if it wasn't clear before, that the internet is not going to bring world peace or solve world hunger, that it won't grow at astounding rates forever, and that putting up a web site for your business will not automatically mean that sheets of money start flowing out of your printer. Anyone that is still not convinced of that needs only to look at Lucent's stock, which fell from $80 a share to $6 a share (along with lots of other companies,) or to read Qwest's announcement that it won't spend any money this year on adding DSL capability to its network, or to realize that Amazon.com still has not ever turned a profit. The internet certainly isn't dead, but it also isn't pure magic anymore.

The problem for IBM, and IBM's customers, is that by the time IBM figures this out many of IBM's customers may well be long down the path that IBM is recommending, but in a manner that IBM doesn't expect. The issue is platform neutrality. IBM's recommendation to its customers to convert all applications to a "technology" that is operating system neutral is puzzling coming from a vendor that currently makes and sells 8 different operating systems: z/OS, z/VM, VSE/ESA VM/ESA AIX, OS/400, PC/DOS, and OS/2. Each of those operating systems has unique characteristics that fit the hardware the operating system runs on, or that match special processing needs and goals, e.g. 3270 terminals and front end controllers are designed to allow a relatively small mainframe to handle large numbers of users, the OS/400 operating system running on AS/400 hardware is designed to be an application programming environment. Are we now to believe that those characteristics are meaningless in an eBusiness world (assuming that it still is an eBusiness world?) Is IBM recommending that all its AIX customers convert all their applications to "operating system neutral" technology? Is IBM suggesting that all its mainframe customers throw away their 3270 terminals and replace them with PC's that can run a browser? Does IBM really expect all the users of DOS to somehow figure out how to get a browser and Java Virtual Machine to run under DOS? Are we to take seriously IBM's suggestion that the banking industry convert all the ATM programs that currently run OS/2 to run some browser based program so the ATM machines can be transitioned off of OS/2?

IBM bet big bucks on Java, thinking and hoping that Java would weaken the strangle-hold that Microsoft has on the desktop market. Java became the linchpin of IBM's corporate strategy, a strategy of making applications operating system neutral which would therefore make Windows irrelevant: if people didn't need Windows to run their favorite applications, then Windows wouldn't have the leverage it needs in order to determine the direction of the industry. And the internet fit in very nicely with that goal. Remember when the internet was going to be the computer and all people would need was enough hardware to run a Java virtual machine and a browser? Anyone remember the issue of Byte Magazine that featured on the cover desktops that were run completely inside of browsers? The problem is that by the time Java became mature enough to handle real-world programs the desktop war was over - hence the need for Java went away. With more than 90% of the desktops running some version of Windows there is no need for an application to have platform independence; what's the point? Plus Java turned out not to be the answer to the platform independence. Java is not standardized, Sun can and does make changes to the language any time it wants; each vendor's implementation of Java is different - witness IBM's Java 1.1.8 and Microsoft's extensions to the Windows version of Java, which are now going away; each version of "100% Standard Java" has different capabilities, so it is still impossible to distribute a Java application without also controlling the version of JVM and class libraries used on the clients; the security features of Java make it difficult to use for building corporate applications, and finally Java, in concert with a centralized server based networking model, is unacceptable to the majority of home users who are still bound by dial-up connections.

However while Java is not the answer to the "platform independence" question, Linux is. With IBM merrily porting Linux to all its hardware platforms IBM has accomplished for Linux what no other operating system has ever achieved - hardware independence. Linux now runs on virtually everything, from embedded systems all the way up to IBM's largest mainframe. To be sure Linux cannot run by itself on the mainframe, it still takes VM to run Linux as a guest operating system. But that doesn't make any difference to the applications running on Linux, they don't change. Linux, with IBM's help, has turned platform neutrality on its head. With Linux, platform neutrality becomes independence from hardware rather than independence from an operating system. It is the ultimate platform neutral enabler.

It would seem that IBM has cut its own throat. What reason does anyone have to buy IBM operating systems (with the possible exception of VM) when they can buy, or download for free, Linux and have a fully scalable, fully portable implementation that runs on all types and sizes of hardware, from palm devices up to mainframes? The problem for IBM is the next logical step is that the hardware becomes a commodity. Why buy an RS6000 or AS/400 when any hardware will run Linux? Simply buy the cheapest box you can find and when that breaks or runs out of oomph, buy a different one. It won't matter, they all run Linux. Couple that with the fact that it will probably always be very difficult to sell commercial software in the Linux world and it would seem that IBM is skipping down a path that will hurt not only their operating system sales, but also their application sales and ultimately their hardware sales as well.

So much for IBM, but what about us? Given the logic of platform neutrality and IBM's positions on supporting their own products, the question for us remains, "why not Linux?" That becomes the crucial question if you have bought into the platform neutrality mentality. If you believe that platform neutrality is the Holy Grail of the computing world then there is no answer to the question, because Linux is the most obvious choice.

But platform neutrality is a worthy goal only for those that think they are going to, voluntarily or involuntarily, switch platforms, or those that still believe the internet/intranet is the computer and all end user computing needs can be handled by a browser. And platform neutrality is not free, it comes at a huge cost. What is given up in order to achieve platform neutrality is everything specific to an operating system. All the special features and functions provided by an operating system, the very reasons why the operating system was purchased or acquired in the first place, are sacrificed on the alter of neutrality. To buy into platform neutrality you would have to believe that the world has so dramatically changed that those reasons do not matter anymore.

I think platform neutrality is a bankrupt concept. To accept applications that are browser based is a giant step backwards in computing capability; to build complex applications that successfully run in a browser can be much more difficult than building those applications to run on multiple operating systems. But the main reason I am not, and will not, embrace platform neutrality is because I already have an operating system that provides me significant benefits and features that are not available on a "neutral" platform, and in fact most of the features aren't available on any other platform. It is these features that persuaded me to purchase OS/2 in the first place, and it is these features the persuade me to continue to use OS/2 now. When some other operating system provides these features, or when the time comes that these features are no longer important or provide benefits to me I will switch. Until then I will continue to use OS/2.

It behooves us all to remember the reasons why we started using OS/2, and why we continue to use OS/2. I have identified seven features and characteristics of OS/2 that compel me to use OS/2:

1. Workplace Shell
2. SOM
3. Rexx
4. Documentation
5. Release to Release Consistency
6. DOS and WIN16 Support
7. IBM

The Future of OS/2

Does OS/2 have a future? I think it does. People and the press have been predicting the demise of OS/2 since 1995, just three years after version 2 was released. To be sure there is a much smaller OS/2 community now than there was 6 years ago, but the future of OS/2 really depends on whether we, the OS/2 community, continue to use OS/2.

What role will IBM have in the future of OS/2? No one really knows, certainly not IBM. IBM may wake up, get a clue, and decide to continue OS/2 maintenance and enhancements. It certainly wouldn't be the first time that IBM has tried to drop a product only to do a turn-around some years later and reverse course. At one time Rexx was going away to be replaced by Visual Basic, IBM's new corporate scripting language; IBM's basic is dead and Rexx still survives.

However even if IBM removes themselves from direct OS/2 support it may not matter much. We already have Serenity Systems, which has released a new version of OS/2 (albeit with a funny name) and is in a position to support OS/2. IBM removing themselves from the OS/2 picture may even help the situation because it eliminates the confusion of who is supporting OS/2 and who is deciding the direction of OS/2. Right now you have the convenience pack from IBM competing with the upgrade protection being sold by Serenity which appear to be two different "upgrade" directions. In the future all upgrades will come through Serenity; in order to upgrade, users of previous versions of OS/2 will have to purchase a copy of eComStation - just like the rest of the commercial software world.

Does OS/2 have a future? If DOS is still sold and supported by IBM, it is hard to imagine that OS/2 will go away.

Why not Linux?

Because OS/2 is here, OS/2 works, and OS/2 is the best desktop operating system around.

Previous Article
Next Article

Copyright (C) 2002. All Rights Reserved.