All Posts in Organizational Effectiveness

|In: Industry Commentary|Last Updated:

Two recent Microsoft communiques have sent my Absurd-O-Meter off the scale.

First there’s the Microsoft ad imploring you to install Windows NT Workstation because it’s faster, more robust, and yada yada yada. What will you be upgrading from? Windows 95, of course, which I guess must be slower and more fragile.

Who writes this copy? News flash to Microsoft: You’re supposed to badmouth your competitors’ products, not your own.

Then there’s the new Palm PC, for which the company that once claimed ownership of the word “windows” borrowed “Palm” from a popular personal digital assistant. It’s going to run — I’m not making this up — a “stripped down version of Windows CE.”

Shows what I know. I thought Windows CE was the stripped down version. WinCE indeed.

To punish the perpetrators of this nonsense, this week we’ll discuss defenestration (OK, I’m reaching) strategies.

We can’t throw out Windows on the desktop. That really is too risky. Instead …

Microsoft has built its Windows NT Server strategy on Moore’s Law (which predicts that the bang-per-buck ratio will double every 18 months) and on the average CIO’s overall nervousness regarding Unix.

In the minds of most CIOs, NT is safer and easier to learn than Unix’s notorious grab-bag of in-joke commands and semi-compatible versions. It’s more versatile than NetWare, since it can act as both file and application server.

And, it doesn’t threaten MVS (now OS/390) because comparatively, it’s still a toy.

That’s fine with Bill Gates. With every iteration of Moore’s Law, NT can take over more of OS/390’s turf, even without Microsoft investing in product improvement.

But NT is, of course, far inferior to many of its head-to-head competitors, at least from the perspectives of performance and stability. Be honest. Isn’t there a part of you that wishes you could use Linux instead? Too bad you can’t take the risk.

Well, you can run part of your business on Linux with no risk at all. All you have to do is break free of the we-gotta-have-a-standard mentality and replace it with a what-standards-do-we-need mentality.

When it comes to Web servers you don’t need a standard operating system, because Web server software shields everyone from the OS. You can run your whole corporate Intranet on Linux (or Netware, or Solaris, or a different OS on every server) with no compatibility or integration worries. And since most of your NT alternatives can handle at least twice the processing load as NT on a given piece of hardware and are more stable besides, there’s a direct business benefit. (Truth in packaging department: In saying this I’m relying on reviews and the opinions of knowledgeable friends, not on direct experience.)

How about the dreaded cost of training? I’ll bet you have a few adventurous employees who’d be absolutely delighted to invest personal time learning Linux on their own, and some Netware hold-outs who’d be thrilled to extend that environment rather than have you phase it out despite its technical superiority.

Cost of administration? C’mon, these are Web servers. You don’t have thousands of logins to administer … and besides, Unix and Netware both have very strong tools for administration.

Okay, I hear you say, but what if Linux (or Netware, or whatever) vanishes from the landscape?

No problem. Simply install Cairo (it should ship by then) and copy the files. Since you’re dealing with true cross-platform standards, you’re safe.

Each platform decision you make has its own risk/reward dynamics. When you enforce a one-size-fits-all strategy you lose your ability to optimize.

The beauty of this defenestration strategy is that while your company benefits, you help preserve a diverse operating system marketplace.

Comments

|In: Industry Commentary|Last Updated:

Some economist analyzed all of the leveraged buyouts of the 1980s and made a startling discovery: In about 70 percent of the cases the companies — generally profitable when acquired — did not generate enough cash to pay off the debt taken on in financing the LBO.

Everyone involved in these transactions knew they would fail, but it didn’t matter, since they all ended up wealthier than when they started.

Why do groups of people who are individually smart seem to act stupidly so often? A group is smart when its members’ personal short-term interests coincide with the group’s long-term interests. Otherwise the group will do stupid things.

That, far more than a bunch of dumb CIOs caught off-guard when a century sneaked up on them from behind, explains why we have a huge year-2000 problem.

The year 2000 is a crisis because until it reached crisis proportions, making this year’s numbers was more important than investing resources in surviving past 1999. Understanding these dynamics should change how you set IS priorities.

I remember raising the year 2000 question back in 1994. We spent exactly one sentence discussing it. Since we lacked the resources to satisfy immediate demand, the year 2000 could wait until 1995, when with luck we would all have different jobs and it would be someone else’s problem. (No, none of us actually admitted to that thought process … maybe I was the only one that shallow.)

What problems do you have with your systems architecture or applications portfolio that you could fix with little stress if you started right now? I’ll bet you have at least one.

Maybe your particular problem is an ancient system with old, fragile code nobody really knows anymore. You can fix it by rewriting modules that require maintenance instead of simply tweaking them, even though the rewrites will take twice as long.

Or maybe you operate a batch legacy environment, and you know in your bones that the company needs to operate in real time. Figure out how to switch one class of updates at a time from the batch queue to online transaction processing and incorporate this into your overall IS strategy.

Use your year-2000 crisis as a lever to sell a long-term program of technology grooming, so you never again experience this kind of difficulty. I know that right now, if you’re a typical CIO or IT director, the year-2000 problem seems like your greatest challenge and these other projects seem distant at best. And I have almost no useful advice to give you on solving your immediate problem except for these very obvious points:

  • Make sure you have enough programmers assigned to the project. Then add a few more for good luck.
  • Assign or hire the absolute best project manager you can possibly find and afford to run the project.
  • Pay more attention to testing than to any other part of the project.

The immense cost estimates for the year 2000 fix are misleading. Yes, organizations will spend huge sums. These sums will have little adverse impact on the economy, though, because for every company that spends there are others, or individual employees, who receive.

The real impact of the year-2000 problem will be to reward far-sightedness: Competitors that have already fixed the problem can now afford to invest in projects that yield a competitive advantage.

Comments