• United States

IT’s big flops

Mar 27, 20068 mins
Data CenterIBMSecurity

Our editors and columnists nominate their favorite failures chronicled – OK, hyped – in Network World over the past 20 years.  ATM and OSI are two unsuccessful paths IT followed in the last two decades. For others, read on.

Next: Scott Bradner on OSI Scott BradnerColumnist, “Net Insider”

OSI: Count the wasted billions

It’s hard to top ATM for being a wrong path followed with vigor, but there is another data-network blunder of about the same scale: the Open Systems Interconnection protocol suite.

The OSI protocol suite was developed by the International Organization for Standardization (ISO) after it turned down an offer from the United States in the early 1980s to let ISO use as the basis of the data network standard for the world. More than 10 years and tens of billions of dollars of software development later (IBM and Digital both spent multiple billions on their own), the U.S. government stepped back to take a look at the state of data networking. Like most governments, it had been mandating support for OSI for quite a while, and it was common knowledge, including in this magazine and the IETF, that OSI was the way of the future. Mostly unnoticed by the governments, however, the Internet and TCP/IP kept chugging along.

In early 1994 the U.S. government said formally that it was OK to use TCP/IP when it could do the job. It took a while for other governments and industry to abandon the OSI trail and find their way back to TCP/IP, but eventually they did. OSI is now known mostly for the seven-layer model rather than an expensive and complex data communications protocol.

Next: Johna Till Johnson on ISDN > Johna Till JohnsonColumnist, “Eye on the Carriers”

ISDN: I Still Don’t Need

Remember ISDN, the original technology sans business driver? As every sentient telecom manager recalls, ISDN – also known as “I Still Don’t Need” – was the telcos’ original lame answer to voice and data convergence. With ultralow speeds and twice as much voice as data capacity, it demonstrated an astounding failure to envision the upcoming explosion in data traffic.

ISDN went on to become a back-up alternative for packet services (chiefly, ) and the infrastructure of choice for some videoconferencing deployments, but its predicted massive deployment was short-circuited by the rise of the Internet.

Full disclosure (at the risk of irreparably sullying my reputation as a pundit): One of my earliest publications was a technical paper in an IEEE conference on ISDN chip design. D’oh!

Then there was switched multimegabit data services: Last year I ran into an enterprise that had finally retired its SMDS network.

I think they may have been the only paying SMDS customer on the planet. On paper, SMDS was an engineer’s dream: Telecom designers had thought of every possible contingency and architected a data service that offered rock-solid delivery guarantees and scaled smoothly into multiple megabits per second. Providers including MCI and several of the invested millions of dollars to roll it out. There were only two problems: In the mid-1990s, when SMDS was introduced, most enterprises didn’t need that much bandwidth, and by the time they did, IP and Ethernet services had taken over. The upshot? No uptake. Oops.

Next: Kevin Tolly on IBM > Kevin TollyColumnist, “Tolly on Technology”

IBM: A history of flopettes

As I’ve mulled over the list of technologies that would fit under the headline of flop, it occurred to me that the biggest flop wasn’t a technology but a company: IBM.

As successful as it is today, it should have been – should have remained – the dominant computing company in the world. With all due respect to the many who labored to develop its various technologies, as a company IBM was the biggest flop of the last two decades when one considers what it could have accomplished. Look at the list of technologies that went nowhere:

  • Microchannel Architecture – the first smart PC bus software that could be configured without flipping hardware switches. It fought against EISA, and ultimately neither survived.
  • OS/2 – Announced in 1987 and originally a joint effort with Microsoft, this became a very solid operating system. It was technically far superior to Microsoft’s competing products but was outmarketed by Microsoft.
  • DisplayWrite – Before there were PCs, there were IBM DisplayWriters – purpose-built word-processing computers. IBM ported this software to the PC, to its midrange S/36 and even to the S/370. Yet this word-processing program lost out to WordPerfect and then Microsoft Word.
  • Distributed Office Support System – Long before there was Microsoft Exchange or even cc:mail, IBM had an architecture and products that allowed peer-oriented information transfer between office systems. User elements were implemented across multiple platforms and conformed to IBM’s Document Content Architecture for content and SNA Distributed Services for the store-and-forward delivery mechanism. It, too, lost its battle.
  • High-speed (100Mbps) Token Ring came along so long after Fast Ethernet had taken critical mindshare and work on Gigabit Ethernet was well underway that it became a classic example of too little, too late.

Next: Mark Gibbs on Newton, Bob and ME > Mark GibbsColumnist, “BackSpin” and “Gearhead”

A threefer: Newton, Bob and ME

My nominations in the category of spectacular product flops are the Apple Newton, Microsoft Bob and Microsoft Windows ME.

First, the Newton: Way ahead of its time, but compared with Apple’s current hardware designs, it must – like the camel – have been designed by a committee.

Next, Microsoft Bob: This friendly, brain-damaged user interface confused usability with cute and deserved to be terminated with extreme prejudice. Even Microsoft’s Clipit isn’t as objectionable, which is saying something.

Finally, Windows ME: Yuck. Given Network World‘s audience, I don’t think I need to explain that last one any further.

Now, you might argue that there are other flops I could have offered, but these three exemplify just how wrongheaded the computer industry can be. Let us hope these products were the last of their lines.

Next: Frank Dzubek on security > Frank DzubekColumnist, “Industry Commentary”

Security: A flop for the ages

For the last 20 years, more has been written about security in Network World than about any other topic, and a never-ending stream of security products continues to flow – yet almost all surveys still indicate that the No. 1 corporate and consumer concern remains security.

No matter what you call it – a fundamental industry flaw or using technology Band-Aids instead of solutions – the situation has not changed in two decades.

When private networks dominated the industry, security was an implicit design issue. In today’s public network world, security problems are endemic, and new problems appear almost every week.

No matter what vendors promise, no silver bullet may ever emerge.

Next: Joel Snyder on Token Ring > Joel SnyderColumnist, “Bottom Line”

Token Ring: Faster? No, flop

When is four bigger than 10? When IBM tells you it is. Armed with a pile of dubious math and the might of the most important computer company in the world, Token Ring’s 4Mbps was pushed as “better, stronger, faster” than Ethernet’s 10Mbps for years – and an astonishing number of companies bought into it.

Even when the market moved on, Token Ring’s religious zealots kept on buying, shelling out $500 or more for new and improved 16Mbps network interface cards (NIC) and thousands of dollars for MSAUs (the Token Ring equivalent of a hub), while everyone else was paying $50 for 100Mbps Ethernet NICs and $250 for managed switches. Token Ring was never a good idea, but it was proof that if you keep repeating the same thing often enough, loudly enough and with authority, someone will believe you.

Token Ring’s “four is more than 10” PR people moved on and almost sold us 25Mbps ATM-to-the-desktop on the “25 is more than 100” theory, a tremendous bargain requiring some of the most expensive equipment and interface cards ever sold – until everyone realized that Ethernet did all that and more for a tiny fraction of the price.

ISDN-to-the-home got the hype and was great technology, except that phone company economists and their 30-year-depreciation models rolled it out five years too late and five times too expensive.

By the time all the ISDN gear was in place, DSL and cable modems were starting to give people a taste of what real broadband was all about, and it was all over but the depreciation.

Previous story: Are you better off now than in ’86? | Next story: Our pundits peer into the future >