Like it or not, commodity servers will force big changes in integrator business models.
For years, mid-range Unix integrators and their suppliers spat the word commodity out with comfortable contempt when referring to PC products. Lately, the tone has changed to fear and indignation. Why?
Why, indeed. Frank Gens, senior vice president of research at IDC, has identified "commodity servers" as the third-largest IT opportunity through the year 2000. (The first two are Internet/ intranet services and network equipment.) Most servers sold today are under the $US100,000 price point, he reports. This is the largest segment of the market and the second fastest growing. In the next three years, servers running Windows NT are expected to roar ahead at a 44 per cent compound annual growth rate, while Intel-based servers will increase at a 23 per cent clip.
Poised to grow
As a result, server revenue driven through indirect distribution channels will leap from a mere 20 per cent in 1995 to 42 per cent by 2000 - and force many mid-range integrators with a product-centric business model to grow or die, eat or be eaten. Conventional wisdom explains the merger-and-acquisition frenzy currently gripping the integration community as an effect of the IT labour crunch. But that analysis ignores or underestimates the far-reaching efficiency demands of commodity computing.
A quick history lesson might help. Like NT-based commodity servers today, electric motors of the 1860s tended to be overlaid on to existing, centralised systems. It wasn't until about 1910 that motors were cheap enough and easy enough to use for individual machines. Over the next 40 years, that distributed power would drive labour productivity up 50 per cent.
But when electricity was first introduced and for many years after, labour productivity in manufacturing slowed because the technology was difficult to integrate and exploit - shades of IT's "productivity paradox".
We're barely 20 years into the Third Industrial Revolution, its beginning pegged at 1974, three years after the birth of microprocessors, when IT equipment prices started plummeting and product cycles shortened from years to months. Finally, server-based computing is poised to shoot up a steep "diffusion curve" - the time line between a new technology's introduction and its widespread use.
On average, new products take about 15 years to rise from a 10 per cent to 90 per cent diffusion level. (The ratio of desk workers to desktop PCs in the US business market is already 1:1 or 100 per cent diffusion.)It's the last five years or so in the curve that are a bitch - when product prices sink the fastest and lowest, when suppliers that fail to create critical efficiencies suffer most, when price can no longer be equated with scarcity or complexity.
This pricing dynamic is what many Unix-platform vendors and integrators have a tough time accepting. "Unix products shouldn't be priced to compete with Wintel products," they gripe.
"Unix is far more complex" (or "powerful" or "scaleable"). The premise here is Marx's theory of intrinsic value - that there's a "fair" or "just" price for everything, above the law of supply and demand.
Nonsense, of course. The value of anything is determined subjectively by what any given buyer is willing to pay - in this case, middle-market IT buyers with strong needs and relatively weak budgets.
Clearly some integrators know this and have embraced commodity computing with a ven-geance. Witness Ernst & Young Technologies' new "palletised solutions" initiative for NT-based mid-range applications, emphasising low cost and rapid deployment. (Java-based platforms are next.) Other large integrators, such as Andersen Consulting and MCI Systemhouse, are busy building similar business units.
And services, naturally, will not be exempt from the efficiency demands of the commodity model. SAP America is clearly on to something big with its "accelerators" concept for lower-cost implementation of R/3 applications.
Likewise, HP's simplified enterprise computing seeks to streamline NT deployment and integration, shrinking total cost of ownership.
Such productising - dare I say it, commoditising - of IT services is already the next key comparative advantage among integrators. Those who fail to wring the costs out of building intellectual capital and putting it to work probably won't be toasting the new millennium.