Computerworld - This version of this Q&A appeared in Computerworld's print edition.
Peter Sandborn is an expert on parts obsolescence planning. He has created tools to help sustain the electronics embedded in safety-critical systems, such as aircraft avionics, to help ensure that they can operate for decades.
When it comes to obsolescence, what's worse, hardware or software? Software is worse. It's potentially a lot more expensive and a lot scarier.
How fast do products become obsolete today? Things like memory can become obsolete within nine or 10 months. People who depend on putting Dell laptops into larger critical systems are stuck with the same problem. The laptops they've bought are not supported for more than three years, and they may be putting them into a system that [must be] maintained for 25 years.
What industries are most affected? The root of the problem is in avionics. Medical diagnostic equipment is another good example. They have to be sustained for 30 years, maybe more. You can't just change things in those systems without putting a lot of thought and work into it, because those are highly qualified and certified systems. These systems can have qualification processes that run tens of millions of dollars. It's very expensive to change anything.
What are the consequences of failing to plan for obsolescence? If you have to refresh the design of a system, you have to go back through some portion of the qualification. It can get quite expensive. I worked on a GPS radio for an Army helicopter, and if you changed the hardware such that it changed a single line of the software, it was an automatic requalification of the helicopter.
There also can be a cost in availability of the system. If you're flying airplanes in Iraq, you may have to retire a perfectly good aircraft so you can rob it for parts to keep the other ones going.
What's the alternative? If you forecast the lifetime of the parts early, you can strategically plan refreshes to deal with the problems and figure out what the optimum refresh frequency is.
Given all of the uncertainties, can you really predict that accurately? This is a decision-making-under-uncertainty problem. [The tool we developed] does simulations to handle all of the uncertainties: in the costs of resolving things, in the dates when something is expected to go obsolete, in how many spares you're going to need. It looks for a solution that is good in the context of all of the uncertainties.
How can IT organizations preserve their technology investments? Planning is king. You can certainly piggyback on the sort of mitigation approaches that people use for avionics, which work reasonably well in hardware situations. Either you're going to have to find an aftermarket supply chain or make lifetime buys and keep the replacements in inventory. For a small volume of things, you can make final orders and store parts.
The other thing people can do is try to consolidate demand and inventory. When a part goes obsolete, I'm not the only person who needs it. What people find is that I need a couple of them, the guy in the next building needs a couple of them, and there's a guy in a branch in Germany that has 10 that I never knew existed. There are parts out there, if you can just link up the people.
How have regulations, such as the EU ban on the use of lead solder, exacerbated the obsolescence problem? The EU ban is called RoHS, Restrictions on Hazardous Substances. You can't have lead in solder anymore. It has made the entire supply chain move to lead-free parts. In one fell swoop, you made obsolete all tin lead solder parts, and now you're stuck.
A lot of systems that IT folks have use tin solder parts. If you need to fix one of those boards, you may have to use a lead-free part. Now you're assembling a lead-free solder part onto a tin lead soldered board, and people question the reliability of that.
Then you have the tin whiskers that can grow and wipe systems out.
Tin whiskers? Yes. Traditional solder is lead and tin. When you add lead to tin, it [grows] these single crystal whiskers over time. They can be millimeters long, and they'll short things out. A couple of satellites have been lost due to tin whiskers. It's hard to understand how to stop it. It's just an example of the kind of problems you get into.
Is some of this obsolescence in the IT market planned? We're all trapped in this problem of "I have Office 2005, and I'm fine with Office 2005. It does more stuff than I'm ever going to use." But darned if documents don't start showing up that I can't open in Office 2005 because someone made them in Office 2007. Pretty soon, I get fed up with this and I'm forced to upgrade. I'm stuck in this cycle of needing to upgrade because the world pushes on and pretty soon I can't function, even though I don't need the new stuff. We've called this the Microsoft business plan.
If this wasn't happening, how would Microsoft stay in business? Their whole stock value is predicated on the idea that everybody needs a machine upgrade and a software upgrade at some average frequency. If they didn't, Microsoft couldn't grow, let alone stay the same size.
A lot of planned obsolescence is really a business plan to keep the level of need high enough that companies grow. They're strategizing on "How do we force people to continue to upgrade their version?" And they're good at it.
Read more about Hardware in Computerworld's Hardware Topic Center.
Free Insider GuideCopyright © 1994 - 2012 Computerworld Inc. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of Computerworld Inc. is prohibited. Computerworld and Computerworld.com and the respective logos are trademarks of International Data Group Inc.
