Technology News Technology News

Another Perspective


On March 21, after nine and a half days on the trail, musher Mitch Seavey won the Iditarod dogsled race, beating 86 other entrants and reminding modern mushers and their dogged fans of the sleds that brought diphtheria serum to Nome in 1925.  IBM, too, has been big on commemorative events of late.  It announced a new group of mainframes on April 7, the fortieth anniversary of the debut of the System/360.  IBM probably spends a bit more managing its empire than dogsled racers spend grooming their dogs, but they basically think the same way.

They like to win.  They understand how to manage teams.  And they know nothing gets the attention of the public as much as a big gamble.

When IBM's chairman Thomas J.  Watson, Jr., bet his company's future on the 360 forty years ago, livelihoods were at stake and, perhaps, the whole IBM Company.  A big bet — the System/360 cost approximately $5 billion to develop at a time when IBM only generated $3 billion a year in sales — and Watson called it right.  This year's mainframe announcement is pretty much a sure thing.  At issue is not whether it will be a success, only how much of a success.

Mushers in the Iditarod race aren't taking much of a chance, either.  It's hard and dangerous, but the races are closely watched and there are rescue teams on call.  That wasn't the case 79 years ago, when Scott C Bone bet on mushers.

Hundreds of lives were on the line, only a few of them the lives of mushers.  Diphtheria was spreading in Nome, and there was no serum in the city to keep it from becoming an epidemic that would infect its population, particularly the children.  There was plenty of serum in Anchorage, hundreds of miles distant and more than a thousand trail miles away.  Bone, governor of the Alaska Territory, reckoned that only a dogsled relay could get the diphtheria serum to Nome in time to stop the incipient epidemic.  The Alaskans put the fate of Nome in the hands of the mushers, and it worked out just fine.  Like Watson, Bone had called it right.

Mitch Seavey and his route
Mitch Seavey and his route
The Iditarod run may be the toughest race in the world

Commemorations of historic events can be more than reminders of the past.  They can also signal the beginning of a new era.  In the case of the Iditarod race, its relatively recent inception came as a reaction to triumph of the snowmobile, which ended the role dogsleds held in the lives of Alaskans since the days when all the Alaskans (and neighboring Canadians) were Inuit.  The mainframe celebration and coincident announcement similarly may mark a turning point in IBM's long history of computer systems manufacturing.

IBM's new zSeries 890 line consolidates the circuitry base of its flagship systems and positions IBM to move its mainframe processor technology in a new direction.  One possibility will be an eventual shift to CPU chips based on IBM's Power architecture, like the circuits used in iSeries and pSeries processors.  A move to Power technology would in effect virtualized the mainframe, much the way the transformation of the AS/400 CISC systems to iSeries Power RISC technology abstracted the salient aspects of the processor design that evolved from the old System/38.

The Power line is the only family of CPU chips IBM has successfully sold and licensed to outsiders, with one interesting exception.  IBM's last generation of mainframe engines, the ones in the z800s now eclipsed by the z890 line, were sold to Hitachi as part of a complicated arrangement that had the Japanese company making IBM mainframes as well as machines sold under the Hitachi brand from the same technology.  Power chips are not only incorporated in two of IBM's processor families, they are also used in Apple computers, bringing IBM about $500 million a year in additional revenue.  Intel chips in the xSeries may have the lead in unit volume among IBM's products, but Power chips bring IBM money, prominence, and some big performance benchmark wins.

Still, IBM doesn't sell computers with heavy emphasis on its chip technology.  On the contrary, it builds its marketing strategy around branded end products, and if tomorrow IBM announced it was going to make all its computers using different CPU chips from Intel or AMD or Taiwan Semiconductor, not one systems customer would think their chosen platform was going to be discontinued.

This emphasis on branding was the basis of the System/360, which was a family of computers based on a number of hardware designs that were made to look the same to software by layers of microcode.  These days, all computer makers think in terms of product families, so it's easy to forget that this is a concept put into commercial practice only 40 years ago.

IBM's most significant deviation from its greatest big idea came with the introduction of the IBM PC, which the company treated like some kind of parasite that would live off central systems or maybe get some use in small offices where mainframes were never going to find a home.  While it's hard to say what would have happened if IBM had based its PC line on a scaled-down mainframe (or System/36, System/38 or another proprietary architecture), it is entirely possible that Big Blue might not be dragging around a ball-and-chain of a PC business today.  If IBM had a What If Department, we'd love to ask it about this matter.  So, too, would be the handful of Windows users who from time to time may wish they, along with Bill Gates, would wake up tomorrow in such a parallel universe.

The IBM System/360
It was the dawn of the big iron age

Meanwhile, IBM has to constantly reposition its various product lines in order to maintain the degree of customer interest it takes to keep a $90 billion corporation in motion.  This month, mainframes lead the pack.  Maybe next month it could be some new iSeries machines or X86 Linux servers.  It all ads up to increased vitality for the company as a whole.

Similarly flexibility is what mushers bring to their dog team strategies.  In case you haven't mushed since your last trip in search of a clean stream loaded with arctic char, we'd like to remind you that the team of animals that pulls a dogsled is nothing like a team of horses that pulls a wagon, a team of oxen that pulls a plow, or a team of mules that pulls in buyers of Borax.

Horses, oxen, and mules are hitched together with relatively rigid harnesses and their motion is controlled with reins.  You can talk to the animals, but except in cowboy movies, when you want action, the reins reign supreme.

Dogsled teams consist of up to 16 dogs (and, rarely, more) loosely hooked by their tug lines to a towline.  The dogs are organized in pairs, one on each side of the towline, except that at the front of the pack there may be either one dog or two, depending on the musher's preference and the condition of the dogs.  So the engine count on a dogsled is similar to that in a mainframe G8 or iSeries Power processor group, a structure IBM calls a book.

The dog or dogs in front are called lead dogs.  The next two dogs, running in parallel, are called swing dogs, and they mastermind turns.  All the rest of the dogs are called pack dogs or just plain dogs except for the two dogs closest to the dogsled.  These two dogs are special and they are called wheel dogs.  They are almost always larger than any other dogs in the team and they have the extra hard job of keeping the sled from hitting obstacles as the team pulls it down what might not be much of a trail.

All the dogs can run into trouble in the form of sharp ice, rocks, and sharp frozen twigs.  Using a line system that allows them to dodge debris helps.  So does the universal practice of giving the dogs shoes, or booties as the mushers call their dogs' protective footwear.

So much for configuration: The key aspect of mushing that makes it so different from any other kind of animal team hauling is that there are no reins.  The musher talks to the lead dog(s) and the other dogs follow the lead(s), more or less.

We say more or less because it's not only the wheel dogs that do a little independent thinking.  All the dogs in a team exercise a degree of spontaneity, jumping back and forth over the towline as they see fit in order to avoid hazards in the trail or to get better purchase on the slippery trail.

The basic mushing vocabulary has a few words for go (mush! hike! all right! let's go!), one for right (gee!), one for left (haw!) and one for stop (whoa!).  Mushers that really get along with their teams can also call for u-turns to the right (come gee!) or left (come haw!).

The mush words all work pretty well except maybe whoa!  This is why a dogsled has brakes.  Without brakes and a musher who knows how to stop a sliding sled, the wheel dogs would be flattened the first time they halted in front of a heavily laden, fast-moving sled.

One more aspect of mushing that's very important: Lead dog is not a permanent job.  Mushers reconfigure their teams all the time, so today's lead dog might be tomorrow's ordinary pack dog.  Not all dogs can lead, but enough dogs on every team have the knack so a musher can keep moving even if a lead dog poops out.  A dog that gets tired or hurt is taken off the team and put in the basket, the part of a sled that carries the payload.  The musher does not ride on the sled but instead just keeps his feet on the back ends of the runners, holds tight to the handles and talks (or screams) his team down the trail.  Mush!

This basic concept of management is how tightly coupled conglomerates such as IBM organize their divisions.  At various times one or two divisions may be singled out as lead dogs for any of several reasons.  Trade shows provide good examples of this, such as Linux gatherings where IBM switches to a two-platform lead team, Intel and Power, to demonstrate its commitment to open software and its market agility.

Still, as is the case with the dogs in mushing teams, it is an error to equate prominence with importance.  Some of the heavier divisions may not get the most fanfare but, like wheel dogs, play a vital role.  IBM's credit operations and software division provide the annuity revenue to help the team get around obstacles on the fiscal trail, but are often kept out of the limelight.

Actually, software is one of the trickiest IBM dogs to watch.  From time to time, Software Group will get a short stint as a lead dog, such as when IBM feeds it a big acquisition, the main method Big Blue uses to boost the otherwise mundane growth in that segment.  IBM no longer has the patience to build new software products and raise them to a big enough size to pull its huge sled, so it has to buy its way into the big money software races.  We can't remember when IBM created a new commercial software product, even though it continues to make valuable contributions to software technology with small products or with contributions to others' efforts, such as its gifts to the Linux development community.

IBM's Global Services group, shown off from time to time at stock analyst gatherings, is another part of Big Blue that is often a wheel dog.  Sure, IBM puts out dramatic press releases when it signs big services contracts, but most of the time the company prefers public attention to be focused elsewhere, in part because the optimistic promises in many services deals exceed the eventual facts.  Services contracts have a way of fading out or ending up in acrimony.  They involve expectations on the parts of both buyer and seller that are subject to differing interpretations.  These deals can end up on the rocks like star-crossed marriages.  Every year, IBM reports a lot of new contract signings, a lot of revenue and, if you go over IBM's financial reports with great care, attrition large enough to power whole smaller services companies.  In fact, IBM's unsuccessful services deals probably do just that.

Attrition is a problem for mushers, too, and these days, with animal care a prominent issue in all walks of life on two or four feet, it is a factor in races such as Iditarod.  To run the 1,150 miles of the race, mushers use teams that vary in size from 8 to 16 dogs.  They must have 8 dogs running to win.  Along the way, dogs may tire or suffer injuries, and the mushers then have to leave the dogs in the care of people who will get the animals back to the starting point where vets provide the quality of care that makes for favorable press coverage.

IBM similarly adds, removes, combines, or splits divisions to make the most of its resources based on its assessments of external conditions, competitors and, of course, its own strengths and weaknesses.  These days, IBM does not seem to take many big risks.

The most dramatically risky moment in IBM's recent history came in 1993 when the company chose Lou Gerstner to be its boss.  And the most important decision made by Gerstner was his rejection of calls from some board members and influential outsiders to break the company up.  Whether that was the best possible decision is academic.  Gerstner wanted to run a gigantic IBM and he did what he could to make the not only possible but effective.  It was probably the best choice by a long shot.  If IBM couldn't find a single employee to run the business when it was whole, how could it find people to run each of several businesses that would emerge from a breakup?

Snowmobiles can be driven by an idiot, and often are,
which is why dogsleds are no longer basic transportation in the arctic

In January 1925 Governor Bone faced a more portentous choice as diphtheria began to spread in Nome.  There was plenty of serum on hand, but it was in Anchorage.  There were one or two bush pilots who might have been able to make the trip in the summer, but in the dead of an Alaskan winter, that wasn't possible.  The available airplanes, which had the open cockpits of the day, had been taken apart and stored for the season.  There was a rail line to carry the serum part of the way, north to Nenana, but after that, nothing but open trails.  With the help of Northern Commercial Company and support from the Army Signal Corps, a group of Alaskans including doctors, politicians, and 20 mushers organized a dogsled relay to get the serum from Nenana to Nome.  It arrived in time and only a few people were lost to the disease.

The current Iditarod race has its roots in a 1967 run much shorter than the current one; long races like this year's began in 1973.  While dogsleds have given way to snowmobiles for most Alaskans, the big race brings out dozens of mushers, mainly men and women from Alaska, but also entrants from as far away as Norway.  Like the mainframe and to a certain extent the AS/400, the dogsled is a legacy technology that isn't what it used to be but is still impressive and astonishing.

Mushers and their dogs, like mainframers and midrangers, are among the most tenacious folk on earth.  They're the kind of creatures who will face down 1,150 miles of subzero weather to get to a bar and a fireplug, or, in the case of mainframes, exhaust a working lifetime in the hope they can get to a locked-in pension before somebody sends them to the outsourcing glue factory that's the scary half of IBM you don't see featured in the press releases.

— Hesh Wiener April 2004

Copyright © 1990-2019 Technology News of America Co Inc.  All rights reserved.