- - - - -
"You're a Luddite!" is a lot more succinct than saying, "You are a dinosaur! You need to get with the times and trade in your passbook for an ATM card."
- - - - -

In 1811, the followers of Ned Ludd represented a culture where the materials needed for clothing, bedding and sails were crafted by artisans. These skilled workers used hand looms even when weaving large, simple items such as bed sheets. As industrialization lead to factories that could mass produce cloth, the value of cloth (until then so static that it was sometimes used as currency) and the earning potential of the weavers fell. As these changes were underway, the Luddites saw the threat progress would mean to their way of life, They then began burning and breaking machines essential to that progress. The Luddites, as such revolted not against to technology in general, but to technology that threatened their way of life.

It is useful to note that the Luddites would have been the best resources to help design and build the new factories, and many other artisans did assist. But Luddites felt that the factories were dehumanizing and devoid of artistry. Thus those who could have been instrumental in the advance of technology became its opponents.

Flash Forward - The New Computer Professionals

Until the late 1970's, computers were primarily purchased by the government, academic institutions and large companies. Hobbyists were still building their own personal computers. Students were still learning BASIC on the Apple 2. When the IBM PC reached the market, it sent a message to the entire world ... affordable business computers have arrived.

As the computing industry took-off in the 1980's, there were not enough computer professionals to fill open positions. Many colleges (including the author's alma mater) would not graduate their first Computer Science majors until the second half of the decade. Thus new computer support positions were often not filled by trained computer professionals, but by workers from other industries. These workers had either learned about opportunities in the industry or touched a personal computer and found they liked it. The newly minted computer professionals began an ongoing effort of self training as new hardware, software and network tools continuously became available.

Companies started depending upon this new class of technologists for direction. Their first recommendations were to reduce or avoid investment in expensive, vendor controlled technologies. This occurred even though these were the only reliable solutions available. The belief that, "No one was ever fired for buying IBM," still held. But the new professionals promoted alternative, lower-cost solutions developed based upon PC technologies. Unlike the mainframe, with its monolithic COBOL and batch processes, the new PCs (servers, as a class of equipment distinct from PCs would come later) were the tools of an artist. There was no locked data center or rules that restricted your activity. If you were talented to write a new program by noon and reconfigure your network by five, the computer was your oyster.

As the computer artisans began to find success, they also began a subtle attack on the mainframe. More clever than the Luddites' choice of fire and axes, was the new computer professionals' use of words. They sent a single message, high and low, one as obvious as it was short, "The mainframe is a slow and expensive dinosaur."

The Luddites' Veer off the Course and the Computer Professionals Declare Victory

The Luddites had two big disadvantages in their battle:

1. The textile factories used more efficient production to reduce the cost of clothing. So while this may have made textile workers unhappy, it delighted everyone else.
2. The government strongly opposed the Luddites. They burned things, they scared people, they didn't seek a permit first. What else is a government to do but oppose such action.

The computer professionals were more fortunate:

1. The mainframe manufacturers pretended as if Moore's Law didn't apply to them. They plowed nearly all the benefits of decreasing technology costs into corporate overhead and increased profits. Thus, as the price of computing halved every eighteen months or so, the cost of a MIP, the mainframe's measure of computing power, dropped only slightly. This delighted no one.
2. The government supported the self taught technologists, or at least indirectly. A government anti-trust suit against IBM distracted the large and traditionally smart company as it contemplated break-up scenarios for half a decade.

In time, even romanticism came to the side of the self taught technologists. As incorrect as it may have been, the technology industry started to think that pretty much anyone could monitor a mainframe (handling ten thousand credit card transactions per second) ... but it required real skill to simultaneously run Lotus 123, ccMail and WordPerfect on an MS/DOS 5.0 PC connected to Oracle on a Novell network. Indeed the romantigeeks were right. An attempt to process credit card transactions on PC servers would require far more skill than that possessed by the typical mainframe operator. Unfortunately it also required more skill than anyone in the PC Server industry.

Similarities - Luddites and Computer Professionals

Let's compare the 1811-1813 Luddite revolt with the battle between the mainframe and X86 servers.
First the Luddite Revolt:

1. It was based upon a fear of loss of income and al loss of identity (as artisans)
2. It opposed technology that would make standard and artless what had otherwise been a fruitful and engaging enterprise.
3. It opposed a technology that was clearly superior for many purposes. No artisan would want to make sheet or sail by hand if they had a machine.
4. It failed because the government intervened to protect the interests of the large cotton and wool mill businesses.

Second, the mainframe X86 battle:

1. Fear. Competitors began to fear IBM in the 1970's. Customers began to fear IBM in the late 1970's and early 1980's. You could invest a million dollars in a mainframe and find after a little growth that you must spend significantly more to reap additional "utility" from your system. If you thought to try a competitor, you would find the few mini-IBMs were all very similar.
2. IBM Set Standards. It took lots of capital to design and integrate operating systems software, database software, transaction monitors, processing units, storage, cooling, clustering, networking and power. Standards enabled less investment by the computer maker, as 3rd party peripheral and software makers could help create what otherwise would have been left up to IBM. Standards also meant less opportunities for innovation as companies spent more time supporting standards and less time creating new ones. For workers in the industry, standards meant bureaucratic processes to control, program and manage systems. For those new to the industry and enchanted by the limitless possibilities of the personal computer, an IBM mainframe was a very unattractive option.
3. X86 Servers were and Continue To Be a Lesser Technology. Consider the technical challenges in the WinTel server space ... processor virtualization, storage virtualization, system-level high availability, improved security, reduced power consumption, the ability to update a process without downtime, the ability to add capacity on demand. These are all features available today on a Z Series mainframe, and most were available dating back to 1997, 1987 and even 1977. A mainframe performs these functions and others including sustaining a transaction processing routine for years without interruption.
4. One Big Difference. There is one major difference between the Luddite revolt of the 1810s and the X86 server revolution of the 1980's through current day ... the role of the government. After the Luddite revolt, the British government interceded vigorously to oppose the Luddites, enacting laws on behalf of the textile manufacturers making "machine breaking" a capital offense. Many Luddites, including the movement's leader, were put to death.

The Two Technology Markets - 1) Open and 2) Reliable

Today, there are clearly two types of consumers. One group chooses X86 technology to gain an open product that they, the customer can control. The other group choses the most reliable technology. This group will accept higher prices and proprietary, vendor controlled systems to achieve expected performance and reliability.

It's important to note at this juncture that the lack of capability in the X86 server market is not because X86 technology is not capable. Consider a Nortel or Avaya PBX. Although they are based upon standards based servers and Linux technology, both have multiple processor boards that will enable a system to stay functional during a component failure or upgrade.

The PBX example is a rare case where the two markets have combined. Telecommunications customers expect phone systems to be highly available and they expect affordable hardware options. Nortel and Avaya have thus used X86 technology to deliver both the hardware and performance that the market demands. Unfortunately, the Nortel and Avaya efforts don't fully constitute market Glasnost. Both companies still sell a closed box, one that is not likely lo run the new Nortel compatible communications software soon to be released by Microsoft.

When market Glasnost is finally achieved, customers will be able to purchase an off the shelf mainframe class X86 system and then choose their applications software.

The key take away from this article is this:

There are two perceived choices in the technology industry:

At one end of the spectrum there is large, heavily integrated and tested systems that are highly available. These include products such as HP's NonStop and IBM's Z-Series systems. Such systems provide the best technology available including redundant processing, integrated virtualization and system management facilities needed to maintain performance and availability. But they are also constrained by utility based pricing, proprietary maintenance, and closed architectures that limit customer flexibility. Even when manufacturers appear to have "opened" their systems, there are usage constraints applied, such as the $100K minimum price to install Linux OS Specialty Engine on an IBM mainframe.

At the other end of the spectrum there are smaller and midsize X86 servers available from nearly all computer system manufacturers. These products require little initial investment, are open to development and run tens of thousands of commercial applications. But they are constrained by single point of failure (SPoF) processing and small data buses.

At this point, the reader may be considering an obvious question, "Does there need to be a choice?"

Is it possible to build systems with the latest high availability and virtualization technology using X86 processors? The answer is of course, "Yes."

So why hasn't the market demanded this? Why is it that CTO's everywhere don't grab their HP reps, point to their Nortel or Avaya PBX, pull off the front cover and say,
"Look, inexpensive X86 hardware, free Linux OS ... the production half dies, the other half takes over without a dropped call. It just never stops. Give me a Proliant that does that!"

Consider the possibility that the entire market has been conditioned to expect proprietary controls and high prices for certain types of technology.

Consider that if HP spent a few thousand per server to make a $10K Proliant server work more like an Integrity NonStop system, HP might not sell many $170K Integrity systems.

Consider that Dell Computer, known for the lowest R&D of the major X86 server manufacturers spends as much on R&D (approx. $500M) as the $3B HP/Tandem/Integrity unit that invented the "NonStop" computer.

Finally, consider the possibility of a new choice ... one that combines the best technology with open, affordable systems.

Talk to your manufacturers, tell them you would prefer not to be a Luddite. Tell them that you want to see more powerful, and more reliable systems. Tell them you want versions of these systems to be based upon X86 technology.

And what if they say no?
Then I suggest you build your own.

RELATED CONTENT
Part 1: Revisiting the Mainframe Utility Computing Model
Part 3: A New Model - the X86 Mainframe
X86 Mainframe Home Page

Tags: X86 Mainframe, Efficiency