Ever notice how rarely we in the IT user community flex our collective software purchasing muscles? It must make the car manufacturers jealous – they live in constant fear of negative consumer reaction. Airlines are also busy scrambling to meet the demands of disillusioned passengers (albeit in their own ham-fisted way). They know their customers will bolt without a second thought.
By contrast, few IT companies I can think of, with the exception of Microsoft have felt the wrath of angry customers (and really, isn’t that all bark, no bite?).
Given this, it’s likely Scott McNealy, chairman and CEO of Sun Microsystems, gave little thought to his company’s decision last January to cease work on a x86 (Intel’s instruction set) version of its flagship Solaris 9 Unix platform.
The reason for axing the project? Simple economics. “We are focusing more on our bottom line,” said Graham Lovell, director of product marketing for Solaris at Sun, at the time. “We need to focus on immediate revenue where possible.”
It appeared to be an open and shut case for Sun: most enterprise Solaris users run the operating system on 64-bit UltraSparc processors. And the company promised the small but enthusiastic Solaris x86 community, mostly running Version 8, that they would continue to receive support for a better part of a decade.
Analysts, meanwhile, grimly offered their support. After all, cutting costs only made sense in such troubled times. Sure, a few customers might get peeved, but not enough to make the move risky, they reasoned.
Exactly when Sun officials realized that they were very, very wrong is unclear. Maybe it was when the Solaris x86 fans quickly informed Sun that they would be happy to pay for the product if they would only resume development. Maybe it was when they built a Web site to support their cause (www.save-solaris-x86.org). Regardless, by summer events started to take on an X-Files-like life of their own – a select group of Solaris-on-Intel users, known as “The Secret Six,” were flown to Sun’s headquarters to personally appeal to McNealy to rethink the decision. Customers were becoming surly – they began hinting that Windows or Linux could easily stand-in for Solaris.
By September, the heat was really on. John Groenveld, associate research engineer at Penn State University, shelled out US$9,292 to place an open letter in the San Jose Mercury News demanding a public debate on the matter. Sun tried to placate them with Solaris Express, a beta version of Solaris 10 that can run on any Sun Sparc or Intel server platform. But users weren’t impressed. “We are moving toward Linux on Intel in some areas because a Sun Solaris offering isn’t there,” said one frustrated customer.
Sun continued to waffle for a few more weeks, but it was clear the company didn’t have many options. In October, Sun finally relented and announced its intention to sell an unbundled version of Solaris 9 for x86 systems by the end of this year. The long, strange saga concluded in late November when that product was made available on Sun’s site.
Sun is now even talking about building versions of Sun ONE and its blade server platforms on x86.
The lessons in this story are self-evident. If a few loud voices and a newspaper ad can get a small group of Sun’s customers (a group Sun could have afforded to ignore, financially speaking) what they wanted, isn’t it high time you had a frank discussion with your vendors? You might be surprised at the results.