Living a Long Afterlife
November 16, 2009
Eight years ago this week, your community lay in a state of shock. Some vendors were not surprised that HP announced the end of its HP 3000 business, but an overwhelming majority of customers and suppliers found themselves caught off guard. The approach of the 3000's afterlife began on Nov. 14, 2001. Like the horizon, of course, the complete exit of 3000 customers has remained out in the distance.
HP continues to find itself surprised at the pace of migration. Alvina Nishimoto, one of the few HP employees left who can help out with 3000-specific issues of moving to HP's alternatives, said as much during the roundtable discussion of this fall's e3000 Community Meet.
It's very quiet on the 3000 front at HP, she explained. But when asked what the surprises have been during the Year No. 8 of the 3000 Transition, Nishimoto said the unexpected continues to surface.
"They're migrating late, which is kind of surprising,” she said. “We have 9x7 customers coming out of the woodwork,” a data point that would seem to suggest more than 1,000 customers continue to use a 3000, because the 9x7s were first shipped 15 years ago. That's been a busy 15 years, since more than half of it has comprised The Afterlife.
We don't let this anniversary pass without reminding our community that HP predicted its demise with astounding inaccuracy. At first 80 percent of you would be migrated off 3000s by 2004. Then came revisions that put 25 percent of the community on the server at 2006, two years later with a larger group. "Reports of my demise are greatly exaggerated," homesteaders quipped. HP didn't even have a term for the customers who would stay put, long-term or intermediate. We provided that homesteader title, which HP eventually started to use. It had to; the majority of its customers were not migrated three years out from the end of 2001.
But the reality is that a very large portion of the customer base of 2001 is now using other platforms.
There was always migration underway before The Afterlife began. HP 3000 customer attrition started in the early 1990s, and some in the community peg the highwater mark for customer base in the late 1980s. HP tried to grow back customers in the late '90s with long-overdue enhancements to networking and Internet features. But the very event which postponed migrations, Y2K, also worked to stall HP's drive toward faster and better-connected systems. What HP called the N-Class and A-Class servers only made their debut six months before the vendor pulled its 3000 plug into the future. HP promised these servers for the year 2000.
That lab-based delay came out of unit managed by Winston Prather while the work started in the late 1990s. The same HP employee moved into the general manager post for the 3000 division, shaping the headcount, as the labs were forced to extend the deadline to a year beyond division estimates. It's little wonder why HP 3000 sales came to a standstill after Y2K. Customers were waiting on a promised product better than those 9x7s. It was time to upgrade, but the new generation was overdue.
Once HP announced it would exit your community, those 9x7 owners couldn't justify buying N-Class and A-Class servers. So that glorious day in the spring of 2001 when Platform Product Manager Dave Snow marched down the aisle at the Solutions Symposium with the first A-Class server -- a marvel of reduced size with increased power and efficiency -- didn't arrive soon enough.
HP was doing its own migration to deliver the final generation of 3000s. The PCI peripherals bus, already running and selling HP 9000s for more than a year, proved to be a complex transfer to the 3000. Some have pointed at the differences between IO handing in Unix and MPE/iX to explain the delays. More likely culprits were two elements that were too numerous and too few. HP 3000s supported a wide array of peripherals, since the HP 3000 credo was "leave no customer behind."
At the same time the HP 3000 lab headcount was being squeezed too small to manage both Y2K repairs and tests to MPE/iX, as well as hardware development projects for the PCI servers. Add those elements during an era when HP's CEO was mandating revenue growth as a way to stick to the HP product line, and you get a formula which delivered a late upgrade, which stalled sales and kept the 3000 from growing. The same manager whose lab direction had to juggle two major projects got to pull the plug on the 3000. Winston Prather has always said he made the call to cull out the 3000. It might be one of the few times when a GM at HP erased his own division.
Prather engineered a safe landing for himself and some of the engineers and managers of the group. As for his customers, many were not so lucky. Having spent their careers polishing their HP 3000 expertise, system managers and programmers suddenly got motivated to learn technology on other platforms. They would compete for these jobs against younger and less-costly technologists. The lucky ones retooled themselves. Few community members can point at a career that didn't take a hit in November of 2001.
Companies like HP don't step away from 28-year-old businesses very often. Your community's contribution to HP's knowledge about ending business relationship is worthwhile for a vendor who will nurture in-house technologies. Except that HP doesn't appear to be in that kind of business anymore for computing, given developments like buying its competition in networking with an acquisition of 3com. One day the HP-UX customers will suffer a day like Nov. 14, and HP will be more prepared than it was eight years ago. The community of 3000 customers was always teaching HP something until the day the vendor pulled its plug. Learning how to estimate the pace and impact of churn and change -- those are HP's lessons that entitle you to help and accommodation from the creator of the 3000.