Larry Ellison can be forgiven for sometimes making a mistake, particularly when it comes to marketing new, or not so new, concepts. His statement yesterday that Oracle was unveiling its "first-ever" hardware product is factually challenged by the 1996 launch of the Network Computer, Oracle's real "first-ever" hardware product. The NC, as it was called, was a diskless, Internet-only computer that looked pretty good on paper, but failed to take the world by storm for two reasons. One: The Internet was a pretty primitive beast in 1996, and running a business using only the Internet was like chiseling an epic poem in stone a la AD 25, as in slow, painful, and not particularly productive. Reason two: an NC cost $800, while a PC at the time could be had for $1200 or so, if my memory serves me. Not a big enough difference to justify a wholesale switch in the enterprise market targeted by the NC.
The resulting failure of the NC, and the subsidiary that Oracle set up to handle the flood of NC business that never arrived, is one of those chapters in Oracle's history that Ellison would want to forget anyway. Shortly after the NC was introduced, Microsoft, IBM, HP, and Gateway -- all scared diskless by the thought of the end to the PC franchise -- came up with an alternative standard that took some of the buzz effect of the NC away. But the final straw that broke the camel's back was the reality of the Internet's lack of productivity. Very few companies had email -- at the time companies like HP eschewed email, and made it hard for employees to have an email address -- and even fewer had high-speed connections. And then there was the software, or lack thereof, to actually do something in a completely diskless environment. Game over.
I wish Larry luck in his next endeavor, though I think he's playing a potentially losing game of catch up this time, as opposed to playing a losing game of being too far out in front of the market. Not only is the Oracle database appliance very me-too, considering all the competitors who are already in the market, but putting a relational database on a dedicated hardware appliance is never going to provide the kind of throughput that a column-based database can provide, and that's before you run a column-based DBMS in RAM, the way SAP is now doing with its BIA in-memory data warehouse. Seeing Larry on stage with a massive chunk of metal and calling it an innovation made me think of all the times Detroit responded to a sales crisis by unveiling yet another over-sized SUV. Game...over.
I have sinking feeling this isn't the last time Oracle will unveil its first-ever hardware product. First there was the NC, and now there's the Exadata. Third time's a charm, Larry. Good luck.