MCA: The Sacrificial Lion
from the IQN Editor's Desk
IQ Newsletter -- Issue #9, January 2001

Original version available HERE (archived).

I confess, I'm a little flustered at the moment. I've been trying to get my head around a particular piece of computer hardware-- a 1997-vintage Compaq Presario 2100-- and around a particular notion of what "proprietary" technology really is. There's a whole bunch of different ideas about what "proprietary" means, and its close neighbor, "standard." I should have this down by now, right? After all, IACT is all about open standards, the free exchange of data, unhampered by technological constraint. Maybe that's why Compaq's corporate point of view, as expressed by the architecture of this Presario, is so bewildering to me. While this machine was billed as a typical "standard" PC, in reality it encompasses a vendor-specific combination of Windows software with proprietary hardware-- which is to say, not "standard" at all.

But the confusion didn't start with the Compaq PC and it doesn't end there either. For many years I've been running into "proprietary" and "standard" a lot, sometimes mixed together, as in "proprietary standard." Add to this the term "legacy" and we've got three horribly abused words in the Information Technology dictionary. These important words and their constant misuse have affected me in my purchasing decisions, in my decisions on what technology to use for a given problem, and in how I view the technology that I use. I can't escape them; they sneak in everywhere.

The misuse of those words by salesmen affected my roommate's decision to purchase this Compaq in the first place. My own hardware purchase decisions have been guided by different motives though. I prefer to use hardware which conforms to IBM's MicroChannel Architecture, which, although proprietary, offers several concrete advantages to me. MCA is generally thought of as a bus architecture, like ISA, PCI and USB. But unlike those three, it never has enjoyed popularity-- not for any technological reason, but because it has been maligned and misrepresented.

If you're already familiar with those bus architectures, you're probably thinking that I'm helplessly moored in the past to be bringing this up. After all, MCA is a "legacy" architecture, and the bus wars were supposed to have ended a long time ago, when the industry firmly settled on the new "standards" of PCI and USB. The point is that regardless of MCA's merits as an architecture, it was helplessly choked to death by Microsoft et al through deliberate misuse of the words "proprietary," "legacy" and "standard." To be sure that MCA remains a dead technology, we continue to call it "legacy" as if that were an accusation, and IBM would have to have their economic head examined if the company even wants to pursue the point any further. But for those of us who care about technological innovation, we might want to pay attention to MCA's true legacy-- a hardware architecture done in by the misuse of the English language, rumor, and outright fabrication.

The Myth of the 'Proprietary Standard'

The technical advantages of MicroChannel Architecture are superseded only by the vast array of myths that surround it. First came the accusation of being a "Proprietary Standard"-- MCA was crucified on these two words. But what is a "proprietary standard" supposed to be, anyway? That label is a self-contradiction and a misuse of the language. A standard cannot be proprietary. For instance, ATA/ATAPI is a standard. ANSI/EIA/TIA-232-E-91, better known as RS 232-C, is a standard. Neither is proprietary.

Proprietary, of course, means owned. It does not mean evil. Somewhere, proprietary got the reputation of being inherently "bad" about at the same time it became regularly coupled as the modifier to "standard." The irony today is if you think your computer is built right on "industry standards," you would be wrong. Even Industry Standard Architecture-- ISA-- is proprietary despite its name. MCA, PCI and USB likewise are not standards. PCI, although popular and bound by rules, is a proprietary local bus architecture controlled by the PCI Special Interest Group. Intel and their corporate partners, Compaq, HP, Lucent, Microsoft and NEC, also control USB. If your computer uses DMA, or Direct Memory Access-- and I can assure you that every machine made to be "PC-compatible" or greater uses DMA-- then your computer uses proprietary technology of IBM.

MicroChannel Architecture is proprietary. IBM developed it as a radical departure from the Industry Standard Architecture bus. It was also freely licensed by IBM to anyone, even an IBM competitor. If you wanted to create an adapter card for an MCA system, you would have had to pay royalties to IBM not exceeding 3% of the total sales derived from your adapter card, and often less than 3% as chances were that as a hardware manufacturer, you already paid IBM some other royalty for some other technology... like DMA. Due to the nature of MCA adapters, you also had to consult with IBM to verify your adapter would have a unique adapter ID so that it would be correctly configured in an MCA system.

Although MCA was a radical departure from ISA and in no way backward-compatible to it, it was a vastly superior technology in many ways. It offered faster data transfers through 'burst' mode, and allowed for automatic configuration of the machine, eliminating DIP-switch configuration of IRQ/DMA channels. It offered bus-mastering. Perhaps most importantly, it was designed throughout to be paranoid about your data. An MCA machine-- most often but not always an IBM PS/2-- will halt when memory corruption is sensed. An MCA machine will check for data parity during bus transmission of data. The machine is designed to fail so that under no circumstances would bad data be saved to a disk. In the design of MCA, data integrity and security are paramount. In other words, years before PCI existed, MCA anticipated most of the features for which PCI is known.

These are the facts. This is what MCA is, in brief. An IBM PS/2 will function like most other PCs from the hardware level up, and to most operating systems is not "special" in any way other than in having a superior hardware base from which to run. However, this is not the public perception of MCA. Thanks to the misrepresentation of MCA by Microsoft et al., the architecture has been dragged through the muck as a "Proprietary Standard" and an alleged failure of IBM Marketing.

The architecture entered the marketplace in 1987, predating Win95 by at least seven years. Microsoft, although willing to provide configuration support for certain inherently non-Plug'n'Play-able ISA and EISA devices, purposely avoided providing support for *common* MCA adapters. One of the most common MCA sound cards, for example, has no Windows support, and in order to be installed under the Win95 environment, requires substantial chicanery with the Windows Add New Hardware Wizzard. Among Microsoft's current offerings targeted towards PCs (Win2K and WinME), neither one has been tested by Microsoft for use with MCA machines even though common models, such as the 9595 series PS/2s, meet Microsoft's exorbitant resource-requirements-for-Windows.

These actual hardware complications, though not unsurmountable, would have been enough for MCA proponents to cope with. What could not be overcome was the increasing perception that somehow MCA-- and all IBM PS/2s by implication-- were "incompatible" with where Microsoft wanted you to go today.

MCA Bashing

In 1996, I took an Operating Systems class at a community college, a class designed for people who intended Information Technology to be their field. The instructor (who was actually one of the best instructors I'd ever had) distributed a document retrieved from the Internet for the purpose of explaining some DOS basics. Before delving into the technical details of the DOS boot sequence, the document, Surviving the Next Operating System by Howard Gilbert of Yale University, offered a type of prologue to put us in our historical context:

"22 Mar 1996

The modern world was formed on April 15, 1987. Today we are simply living out the consequences of that day....

IBM was about to enter the Twilight Zone. On April 15 they announced the PS/2 line of personal computers. Its Microchannel bus was too expensive and did not have enough adapter support, compared to the clone's continued use of the ISA bus.... The VGA standard was good, but at the high end a long string of technologies ending in the XGA would be overwhelmed by less expensive SVGA adapters....

It [IBM] made sure that OS/2 could only run on PS/2 systems, nominally increasing demand for its own hardware but really dooming the operating system....

Microsoft tells everyone exactly what they think. People don't believe it. This is a big mistake.

IBM tells everyone what to think. People don't understand it, usually because it doesn't make a lot of sense. IBM may have lost the ability to explain anything to anyone without getting lost in meaningless marketing buzzwords."

[see hyperlink below in the Endnotes section]

There were 30 people in this class. We all read this creative interpretation of "history" as if it were fact, given to us by a man we trusted. Some of us went on to receive MCSE credentials and would doubtless influence somebody's purchase decision and operating system choice.

I might be one of the few who didn't buy into Gilbert's myths about MCA. I certainly know better than to believe the old Sufficient Adapter Support Myth, for example. That's where Gilbert's generalization "Its Microchannel bus was too expensive and did not have enough adapter support..."is stated with no proof. True, the average PS/2 computer system was more expensive than an ISA clone of the same time. But it was also more advanced. Regarding 'enough adapter support,' this is a myth, persisting to this day, that only IBM created a few, select adapters for MCA systems. I have adapters that were manufactured by NEC, Kraft and Kingston-- a small selection of the numerous manufacturers who created MCA cards, including Adaptec, ATi, and Creative Labs. In 1995, towards the end of MCA's commercial existence, a PS/2 could be outfitted with a varied number of SCSI host cards, NICs, Memory expansion cards, sound cards. Of course that was not as great a variety as was possible with ISA, but you have to wonder-- with PCI only being suggested three years before and PnP BIOSes not yet widely available-- how many cards for PCI were there in 1995, at the beginning of Microsoft's Plug'n'Play revolution?

Take the next myth. "The VGA standard was good, but at the high end a long string of technologies ending in the XGA would be overwhelmed by less expensive SVGA adapters." Ah yes. Inexpensive SVGA. Which SVGA flavor would that be? There are so many....

Then there's the OS/2-PS/2 Myth. "It [IBM] made sure that OS/2 could only run on PS/2 sytems, nominally increasing demand for its own hardware but really dooming the operating system." This is a lie, in boldface. OS/2 runs on many different kinds of hardware. As far as the 'doomed' operating system goes, this was another boldfaced lie presented as fact to an easily-swayed audience-- because the document was printed *before* OS/2 Warp 4 was released and therefore before any realistic prediction could be made. Is it any wonder why MCSE-trained individuals think OS/2 is dead?

We can see Gilbert's true colors when he says, "Microsoft tells everyone exactly what they think. People don't believe it. This is a big mistake." while in the next breath he claims that "IBM tells everyone what to think [but] it doesn't make a lot of sense" In this last bit the author confirms any doubts about his prejudices. I'm beginning to wonder if he got jilted on some IBM stock and was wanting to strike out at the company. Of course, in his book I would be a "revisionist historian." Read those lines above again and correlate it to your own experience. Odd, isn't it? Isn't it Microsoft that has told us repeatedly *how* to think?

You will accept the Microsoft "defacto standard" even though Office2000 is a rare thing yet.
You will accept Win2K as your alternative for an operating system in the workplace.
You will accept Microsoft's proprietary file formats.
You will accept Microsoft's web browser with each of their operating environments.

And every step of the way, between blue-screens-of-death, Microsoft will pat you on the back, tell you you've purchased the "defacto Wintel standard" and what a good little consumer you are. Microsoft, not IBM, has been telling you how to think by very purposefully controlling how you use a Windows computer and how you move data out of it. IBM could never get away with this.

IBM can't win for the losing. In supporting MCA, they were deemed to be acting in a "proprietary" fashion, against the industry and against the consumer. Had they not supported MCA as much as they did, they would have been accused of abandoning the baby on the doorstep. Had they supported Microsoft more fully, every original idea produced by IBM would have been swallowed whole by Redmond and spit out in some slightly-incompatible format for the next Windows Patch. Buck Microsoft too much, and Microsoft will make you pay... by not including any support for your hardware. Then you'll have influential idiots like the honorable Mr. Gilbert insisting that IBM is both as inept and evil as the Soviet characters in Sylvester Stallone movies. IBM, it seems, can't please a person without offending 100 other people. "Proprietary" technology is fine-- as long as your initials aren't I, B, and M.

Of course, I will be accused of backing Big Bad Blue too much, of revising history, of pulling the Gilbert article out of thin air and using it to back a non-existent argument. But in Scott Mueller's Upgrading & Repairing PCs [see link below], a book published at the same time and highly regarded as a "classic" by academics and journalists, MicroChannel was described again:

"IBM not only wanted to replace the old ISA standard, but also to receive royalties on it; the company required vendors that licensed the new proprietary MCA bus to pay IBM royalties for using the ISA bus in all previous systems."
True enough, IBM wanted to be paid for its work. But somehow we collectively had gotten the idea that ISA was "ours" and in the public domain.

So while Mueller continues granting MCA's superiority to ISA "in every way," he also asserts MCA was the impetus for Extended Industry Standard Architecture-- a 32-bit version of ISA that vendors and manufacturers felt would remove the responsibility of paying IBM royalties on ISA. But EISA was proprietary as well, owned mostly by Compaq. EISA, if one is to use the same stick by which MCA is measured, failed horribly.

Mueller-- who again, gives MCA its technological dues-- later on describes the creation of the PCI bus. Look for the word "proprietary:"

"In early 1992, Intel spearheaded the creation of another industry group formed with the same goals as the VESA group in relation to the PC bus. Recognizing the need to overcome weaknesses in the ISA and EISA busses, the PCI Special Interest Group was formed."
The word isn't there, is it? PCI is a proprietary architecture, though. The PCI SIG controls and owns it. Who makes up the PCI SIG? Well, that would be Intel, Compaq, Adaptec, Phoenix, Texas Instruments, AMD, Hewlett-Packard... and IBM.

So IBM figured out how the game was played. They had to. After all, IBM is in business to make money, not just bus architectures.

Mueller's book, never identifying the PCI bus as proprietary, always describing it as promising, was read by at least thousands of people. The book was a requirement of the PC Hardware class I didn't take the following semester and is still regarded as one of the best books on PCs bar none. Mueller deserves that credit; it is a fine book. But the treatment of MCA as the proprietary albatross is wholly undeserved.

Mueller's and Gilbert's comments are just two examples from a specific year, but they were part of an overwhelming trend. Time and time again, this was how MCA was labelled. The public perception of MCA, if one exists at all in 2001, is that it was as Gilbert mentioned, "IBM's entry into the Twilight Zone." It was treated as a historical abberation on the path to true Nirvana-- oneness with the "Wintel standards."

So MCA was hung out to dry. IBM ceased making PS/2s at about the same time-- 1996. The other manufacturers that had tried marketing to MCA left. And now even RS/6000s lack MCA. It must have become a legacy.

The Legacy Myth

Haven't you heard the news? Legacy applications, legacy workstations-- they are everywhere, sucking the lifeblood from the true workhorse, the "Wintel Standard" box. An NT LAN has to cater to legacy systems as does Win95/98. They are the bane of Information Technology, with their slow processors and proprietary busses.

I might disagree with that. Not that legacy computer items aren't everywhere (they are), or even that, yes indeed, a lot of legacy hardware doesn't offer awe-inspiring performance (it doesn't.) However, I am going to question how we value it and how we determine what a legacy product is. Your legacy could well be my upgrade. For IACT readers who don't speak English as their first language, feel free to throw out your dictionary at any time-- this changing definition of "legacy" gets dirtier before it gets better.

One of the huge assumptions made about legacy-anything is that even with so much of it supposedly out there, exactly how much of it is *in current use* is hard to determine. As such faulty logic goes, we should assume legacy systems aren't being used at all. But how can any of us know? Quick, tell me-- by reading this document-- what kind of computer I'm using, or what kind of processor. You can't. You can't tell what the bus architecture is of each 'n' every server and router that carries this text to your computer. Before you go around claiming something is "dead," you might want to make sure it's not your mail server. MCA? It's supposed to be dead. It was supposed to have been dead all last year, too, and the year before that. It's supposed to be a legacy technology. Yet as recently as August 2000, my ISP retired their IBM 9595 dual processor-complex newsgroup server, a "legacy machine" (which is why a buddy of mine now has the most overpowered desktop). I know this, a few other people know this, and now *you* know this. Most of the people using my ISP never knew-- or cared-- how the newsgroup service came to them, just that it did. The "legacy" 9595 did its job admirably well, and was changed out with nary a hiccup from the end-user perspective.

The other assumption-- maybe *the* assumption behind legacy-anything-- is that because of its age, it has lost innate value. Hogwash. Value is something people give objects. If Civilization ended tomorrow, the value of legacy matches would far outweigh the value of 866Mhz personal computers.

As it is, the 866Mhz baby of today will be tomorrow's legacy product, unsupported and unloved by Microsoft. That doesn't mean you can't get useful work out of it. After you've owned the thing for a year or two, you might well have it tuned precisely to fit your needs, accounting for and compensating for, say, a slow disk drive system by using disk caches. You've got your applications and you're happy. The stuff works. Microsoft then will enter the picture and tell you that your life is no longer worth living, unless you take the "better alternative" they have in store for you. However, you won't be able to bring any Legacy Luggage with you. It's unsupported.

Before you ditch your legacy system, ask yourself what Microsoft is tempting you with. Does it really outweigh the advantages of your own "legacy" hardware? Both your legacy system and the new one give you e-mail, spreadsheet, word processor, web browser, perhaps some database software. Those are the four or five core tasks that most of us use PCs for-- the same core tasks any legacy system can do easily. What else does your legacy system offer you? It saves you from having to spend more money on Microsoft's continual program of "upgrades." You can use your legacy machine until it dies (and most computers are far tougher than we give them credit for). And you can spend that same amount of hard-earned cash building your own portfolio instead of Mr. Gates'. He has enough. You don't. Spend it on yourself.

Back to my hardware problem...

A long time ago... well, in the intro to this article, I said I was trying to get my head around a piece of computer hardware. I still am. I'm thinking of calling it a "legacy" computer so that its value plummets.

To make a long story short, it has given me an incredibly frustrating time. It merely needed its floppy drive replaced-- a simple operation, you would think, and one I've done many, many times before. It took me over an hour to tear the machine down, replace the drive, and build it back up again to test the surgery's outcome. Although not spanking new, this machine is far more accepted as "standard" than any MCA machine can ever hope to be. It uses EDO memory... and the PCI bus. It's not a PC that anybody would derisively call wierd.

This Compaq is the machine that deserved to die on the words "proprietary" and "standard." It is proprietary to a fault. For example, a generic 1.44MB floppy drive would not fit in the case and have its eject button line up with the outside of the case. The machine's PCI bus is trimmed down to a single socket with a modem stuffed in it and bolted to the machine's chassis (a non-expandable version of the PCI expansion bus?) Of course, the only reason I could enter the hollowed cabinet of this Compaq is because the warranty had expired. The warranty was evident by the existence of a seal on the outside of the case. Clearly, if you purchased this computer-and-warranty combo, you were forbidden from tampering with any of its proprietary components.

The solution to the floppy-eject problem solved itself by the addition of an ordinary pen cap, which when thrust through the Compaq proprietary oval hole on the front of the case, trips the lever & ejects the diskette. A legacy writing tool saved the day.

And I'm still wondering... as my PS/2 systems offer excellent protection of my data by their superior architecture, as a floppy drive in a PS/2 can be changed out within minutes, as IBM never believed me to be so stupid a consumer that I would require a seal on the outside of the machine... I'm still wondering how we allowed a worthy technology to get obliterated by Microsoft's phony definitions of "standard" and "proprietary." We allowed ourselves to believe Microsoft's twisted language. We allowed IBM to be labelled "greedy" and "stupid," merely for wanting to be paid for its own efforts. That's the MCA Legacy, folks. It serves as a warning to all you innovators out there: if even Big Bad Blue can get trampled, what makes you think you'll survive Microsoft?


on-site links

Edward R. Mortimer, IQN Issue #3
- The Amazing Metallic Techno-colored Standard
examples of battles over standards, in IQN Issue #6
- The Fight of the Year: The End of the X.500 Standard
and more on proprietary versus standard, in IQN Issue #8
- Time for a Standard Operating System

Diane Gartner, IQN Issue #6
- Windows Myths and Facts: A Smart Approach to User Choice
and in this same issue, IQN Issue #9
- The Orwellian Dictionary of Computer DoubleSpeak

IACT 24x7 Report on proprietary plan from MSFT, Intel and Compaq
- The Slippery Slope of PC 98

On-line and in-print sources

ANSI/EIA/TIA-232-E-91 (RS 232-C)


Gilbert, Howard
PC Lube & Tune website, 1996
- Surviving the Next Operating System
current on-line version (PC Operating Systems)

Heath, Chet & Rosch, Winn L.
Simon & Shuster, 1990
- The Micro Channel Architecture Handbook
(see Appendix A: Patent Licensing Tutorial/IBM Worldwide Patent Licensing Practices)

IBM's Deep Blue

Mueller, Scott
Que, 1995
- Upgrading & Repairing PCs, 5th ed.
- review, at Greg's Resource of the Week website

Louis Ohland's "9595 Ardent Tool of Capitalism"


Plug 'n' Play


Peter Wendt's "Microchannel Enthusiasts"

. . .

Copyright © 2001 MoonWolf Enterprises. All Rights Reserved. Unless otherwise noted, the entire contents are copyrighted as law permits by MoonWolf Enterprises, and may not be reproduced, stored in a retrieval system, or transmitted in any form or by any means without prior written consent of MoonWolf Enterprises.

An original IQN commentary first published at the IACT website, <> in January, 2001.

9595 Main Page