SUNORACLE: RETURN OF THE KING by Charlie Martin
There once was a company that would sell you — or lease you — a complete computer system, front to back, soup to nuts, hardware, software, training, even operate it for you if you wanted. It was called IBM, and back in the ’60’s it was the best, most profitable computer company in the world.
Today, Oracle announced their plans for the long-delayed acquisition of Sun Microsystems, and with it their vision for the future of the combined company. Oracle and Sun are going back to the 1960’s, Tom Watson and Fred Brooks IBM.
Today, in a five hour (yes, five full hours, plus a break for lunch) conference call, Oracle laid out its plans for Sun after the acquisition that finally closed yesterday. For an old Sun guy like me, it started with a lot of fears — starting with the layoffs. Doesn’t matter much to me directly, I’ve been laid off from Sun twice so far and I’m not currently working there so they can’t lay me off again, but I do have a number of friends there, and pundits like the Fake Steve Jobs were saying things like this:
Larry is (a) too smart to keep you Sun folks around; and (b) too craven to be honest with you about that. So tomorrow will be the night of the long knives for Sun employees. My Little Pony [Jonathan Schwartz, the last CEO at Sun] is totally in on it and, brave soul that he is, agreed with Larry that they could use the Apple event as a way to distract attention.
Amusing as FSJ is, he’s not necessarily a great source: every one of the Oracle/Sun executives who spoke was wearing a “We’re Hiring” button. Oracle announced plans to add 2000 people in sales alone, and make them the best paid salespeople in the business; remaining layoffs will be limited to about a thousand total.
Oracles’ plan now is to move back to a model where they sell hardware and software in integrated packages, all certified to work together, all with built in management and monitoring software that works across the whole platform, and all built on top of open systems, including Solaris and Linux, and using Java.
Somewhat astoundingly, this is actually a very good plan. Sun’s technology is first-rate; they just never could figure out how to sell it. Oracle sells really well, but they have to deal with many different platforms, many different operating systems. By buying Sun, in Oracle’s words, they will be able to sell their software on any platform — but they can also show you an “integrated stack” that is a complete system, built to work together.
Sun had already been pushing toward that model — the Fishworks/Amber Roads appliances that became the Sun Storage 7000 series Unified Storage Systems are built to make new technology like ZFS easily available in a single package.
Why is this good? Simply, because the economics of data centers have becomes wildly unbalanced. When IBM started selling computers, they were expensive and the people who could run them, configure them, maintain them, were very rare. Someone buying a computer for business needed help just to bring it in. IBM solved that by selling or leasing packages, along with IBM’s then-famous service. (Our IBM salesman and field engineer became friends of the family — the salesman attended by high school graduation, and I bought my fist car from the field engineer.) The difficulty of finding skilled labor was a major issue for data processing in the 80’s, and IBM helped solve that by providing whole systems and the operators and administrators and field engineers to run them.
When Sun started, a single significant computer cost $50,000 or more; filling a data center, with the good old aquarium computer room, raised floors and all, could run into the millions. Systems administrators were more available, and there were many vendors, lots of different hardware. Hiring administrators to run a data center, doing scripting and backups and network configurations, made some sense, even if it cost another $200,000 a year to pay them all. A lot of those administrators jobs came down to putting random pieces of software and hardware together, and making them work. Labor was a major cost, but the hardware and software to run your data center had become the dominant cost.
Now, however, a server more powerful than anything Sun built in 1985 costs $500, and companies won’t pay $100,000 a year for administrators to operate a $500 computer. Labor has become the dominant cost, and like in the 60’s it became the limiting factor again.
In the computer business, what goes around, comes back. The market needs computers that provide all the functions, but that are easy to configure and easy to manage. An integrated, vendor-built, certified, tuned complete package is the answer once again.
Everything else follows more or less directly from that. Oracle will keep MySQL and improve it. They will continue development on the OpenOffice tools, and extend them for “in the ‘cloud'” enterprise use. (They’ll call it something else, though — Larry Ellison hates the “cloud” terminology.) Oracles total R&D budget will nearly double, and the SPARC processor family will be developed and extended. (John Fowler was wearing a “We’re Hiring!” button as well, and made a particular pitch for chip designers.)
Why does this follow? Because, it turns out, most Oracle installations have MySQL as well, and because Oracle can be tuned to run very fast on the SPARC architecture.
The biggest news, though, was at the last when Larry Ellison took the stage. Beyond the news on his plans for the World Cup and his hopes to buy a sports team, he talked about the finaincial position of Oracle when he was asked when Oracle would begin to be profitable with this new acquisition.
Ellisons’ answer was “February.” Not next year, next month.
That, for a Sun Guy, was the clincher. Everyone who has worked for Sun has known in their heart that Sun had plenty of good products and technology, but somehow we could never figure out how to really make money on it. Ellison’s plan sounds good, and if he makes good on his promise to start being profitable immediately, then, well, the Sun could rise again, by imitating the IBM of the 60’s.
Tom Watson was a very smart guy. It could be that his business model will once again dominate the computer business like it did in 1965.