Summary: Lame rant on the importance of buzzwords and fads in high tech.
John C. Dvorak, an award-winning technical columnist, is not known for avoiding (technical) controversies. Some of his opinions and predictions (such as suggestion that Apple would abandon MacOS and switch to Windows) were proved wrong (at least thus far); others came true (for example, he predicted that Apple would release iPod despite Steve Jobs' denial). In a recent column Dvorak responds to Sam Palmisano (head of IBM) who has just pronounced client-server computing dead. Says Palmisano:
"The PC client-server model has run its course."According to Palmisano, businesses must "escape the economic waste that has plagued traditional client-server architectures" and move to "a new architecture for data centers" coupled with "software as a service (SaaS) environments and service-oriented architectures (SOAs)."
Dvorak calls Palmisano's "new" approach nothing more than:
"what they do now, but with a fancy name"and concludes:
"In the next few years we will see more and more new monikers that rename what we already do but change absolutely nothing."You may discard Dvorak's opinion as irrelevant, but if you decide to analyze Palmisano's statement seriously, you will have to answer similar questions. Do client-server architectures really plaque businesses with economic waste? If they do, what (or who) can guarantee that the "new" architecture will do any better? By emphasizing the end of PC client-server model, does Palmisano mean to somehow distinguish it from non-PC client-server models? Does he suggest that non-PC client-server models (whatever they are) are alive and well? What is so particular about the PC (vs. non-PC) client-server model other than the fact that with the recent sale of PC unit to Lenovo IBM has exited the PC business? And which "traditional" client-server architectures is Palmisano talking about: 2-tier, 3-tier, n-tier, ...? Isn't Web-based architecture also client-server, as Dvorak rightly points out? Aren't all distributed application architectures based on the client-server architecture (even peer-to-peer can be viewed as client-server at certain angle, as well as SOA and others)?
I guess, Palmisano, who majored in history at school and specialized in sales at work, did not intend to be technically precise, or even correct. His likely goal -- and it's hard to argue with Dvorak about this -- must be promotion of the IBM's current lines of businesses (servers, consulting services), and buzzwords seem to be the best promotional drivers in the today's corporate world. In fact, if he substituted a new architecture for data centers with IBM-built hardware and software as a service (SaaS) environments and service-oriented architectures (SOAs) with IBM-built software, Palmisano's talking points would make more sense (at least, from the IBM's perspective), but they would not generate as much hype.