OpenMike: A Case for Commodity Computing

A Case for Commodity Computing

OpenMike is a guest editorial space and talking point for the 3000 community to express its views. In an era where emotion and analysis run abreast for the community’s attention, we want OpenMike to be a forum for the way you feel about the future and what you believe is important to the 3000 community. Send your contributions of less than 1,500 words for consideration to editor@3000newswire.com.

By Duane Percox

You have to know what you need and want. Don’t assume you need what you have now.

I suggest that HP is viewing the computing industry’s “parts” as a commodity that can be put together in creative and inventive ways to create value, while still achieving lower cost and improved reliability.

Check out the IA-32 server space with regard to what the various manufacturers are doing. A few are just putting together basic systems. But a number of them are building very nice server systems that are priced like commodity items (relative to proprietary server costs), at the same time being differentiated by how the “commodity parts” are packaged.

I believe the marketplace wants flexibility and cost containment at the same time. This is driving us toward the enterprise model we are seeing more and more — clusters of dedicated servers running standardized hardware and software so they are reasonably interchangeable.

This is also driving the architecture of applications. Deploy your application to a TCP-connected RDBMS and you get yourself an entire set of deployment choices without having to re-architect or re-write your application. For example, you could use Linux on IA-32 with PostgreSQL as your database server — and without any coding change you could move one or more databases to another Linux IA-64 box.

My study of the marketplace and industry indicate that the HP e3000 isn’t delivering the value it once did. And it’s only going to get worse.

There are many reasons for this. Some include the industry catching up, some include the reduction of sustained innovation on the platform, some include HP’s reluctance to organize their business in a way that supports mature products. Keep in mind this last point is historical and not something new at HP.

I’m no economics wizard, but it seems not getting support revenue into CSY sure guarantees that as the product matures you have less and less revenue for sustaining the product — which guarantees only products with a growth path will survive. Installed-base products, therefore, wither and die.

HP isn’t organized to allow the stream of support revenue to get funneled to CSY for platform improvements. The Microsoft model, which isn’t support revenue-based, requires continued purchases of upgrades and new products to sustain growth.

By growth I mean increased demand, which has been an issue for CSY-supplied products (HP e3000). The division’s business has been mostly installed base for quite some time. This revenue stream isn’t enough to sustain a platform in an ever-changing technology world. You can complain all you want about the change, but it’s happening and even driven by companies to force their competition into a competitively disadvantaged position.

Also, over time hardware costs must decline to stay competitive. If the HP e3000 is not experiencing growth in unit sales to offset the required decline in cost, then the revenue stream will decline. As that revenue stream declines the investment in the platform reduces, and the potential for increased sales diminishes. You can play out this scenario and you end up with what HP has determined: by 12/31/2006 you better be somewhere else.

HP is trying, with IA-64, to turn the tables on its competitor Sun by leveraging what they hope will be a universally-adopted server chip. That chip will be mass produced by the king in chip production (Intel) for significantly less than Sun can make its SPARC chips. Sun was an original player in the IA-64 space, but later pulled out for reasons not fully, truthfully disclosed.

Duane Percox is a founding partner of Quintessential School Systems, an HP 3000 application supplier to K-12 educational institutions in the US.