Oracle on Ulitzer

Mergers & Acquisitions on Ulitzer

Oracle finally closed on its delayed acquisition of Sun Tuesday, leaving local entities to shift for themselves according to local laws and sidestepping MySQL creator Monty Widenius' hopes of Russian and Chinese regulators stalling the merger.

Widenius will now presumably revert to his quixotic Plan B and appeal the European Commission's clearance last week, a green light that looked really iffy there for a while.

Oracle CEO Larry Ellison got out ahead of his own announcement Wednesday and started telling the press Tuesday evening that - contrary to Oracle's usual practice with an acquisition - he intends to hire more people at Sun than he fires.

The next day he tore into the "highly irresponsible" reports last week that claimed that Oracle would lay off half of Sun's 27,000 or 28,000 people, calling them "garbage" and scolding their Wall Street author and his minions saying they should be ashamed of themselves for making Sun suffer more angst.

It's still not exactly clear exactly how many more Sun people will get the ax. Heck, it's still unclear whether Sun cut the 3,000 it said it would in October or if the 27,596 people that worked there at the end of September are still there.

Depending on the moment - as is often the case with Larry - it appears Oracle will cut somewhere between 1,000 and 2,000 people, presumably folks in overlapping functions, and that it means to hire 2,000 more salesmen, engineers and support personnel.

Oracle executives, new and old, were sporting "We're Hiring" buttons on their lapels Wednesday and promising to pay new hires more than they're making now.

According to Ellison, "We're not cutting Sun to profitability. We think Sun's a growing business." He expects it to take back share in servers and storage.

Oracle's CFO Jeff Epstein mentioned something about paying on margins, not revenues. Still Oracle said it means to have the highest-paid sales reps in the business.

Ellison Tuesday struck a go-it-alone pose, seeming to burn his bridges with the IBMs and HPs of the world that sell Oracle's software on their systems. "It took us a while to decide that we would be better off with all the pieces, and not working with partners," he told the Wall Street Journal. He probably wishes he hadn't said that.

He's apparently intent on dumping Sun's resellers, though, at least those that don't add any value, "starting this week." He didn't define value and it's probably a pretty high threshold.

"Sun has wonderful engineering," he told the New York Times, "but they didn't seem to like selling very much. The partner model was disastrous, and we are immediately changing that."

Instead Sun will sell direct - at least to its 4,000 top customers which account for 70% of its revenues - using product specialists this time, not generalists and sales will include two new purpose-built pre-assembled systems designed for Oracle software that are supposed to come out this year. What exactly is unclear.

Oracle's Exadata data warehouse, the appliance that now runs on Sun hardware rather than HP's, Oracle's singular experience with hardware so far, reportedly has a $100 million pipeline - or maybe it's hundreds of millions. Larry was a little fast and loose with the number.

It may be delusional, but he claims all of Oracle's myriad database sites are Exadata candidates.

The Exadata box will be expanded into other uses and serve as the model for Oracle's Back-to-the Future vision of complete, purpose-built, task-specific, integrated systems whose components - from chips through applications - all come from Oracle like the mainframes of TJ Watson's IBM in those bygone days before PCs and industry-standard servers convinced people that they were Swiss Army knives.

Ellison is in the midst of a profound multibillion-dollar love affair with the IBM of the 1960s, which he calls "the most important company in the history of the earth." Users are supposed to take solace from the fact that Oracle's retread vision has been done before by Big Blue. It's supposed to result in all parts and systems being optimized for the purpose they are meant to serve and being cheaper than assembling best-of-breed components from multiple vendors. Oracle even means to replace users' IT administrators who screw things up when they change things. Not that Sun won't continue to sell general-purpose machines - it will apparently - but the point of the exercise is application performance.

Oracle claims it's the only company around that can deliver complete systems: IBM lacks the applications (not to mention that its database is a decade behind and uncompetitive except on mainframes). Microsoft doesn't have the hardware, management or vertical apps. HP doesn't have the apps, middleware or a database and its virtualization is thin. SAP just has its horizontal software, a dash of middleware and a thin database.

Frankly Sun under Oracle doesn't sound that much different than Sun pre-merger.

Like an American Indian making use of all the parts of the buffalo he just killed, Oracle doesn't seem to be discarding any of Sun's widgetry - and it took it five hours and scads of overheads just for it to hit the bullet points of its salvage job. At least for the customer-calming moment everything, it seems, even overlapping products, will be sucked up into Oracle's integration scheme (with Oracle's remaining ascendant) and in the process Oracle means to jack its R&D budget from $2.8 billion last year to $4.3 billion now that Sun's on board.

Before Oracle bought PeopleSoft in 2004 and started on its world conquest R&D cost it just $1.5 billion a year. Sun spent $1.6 billion in the year ended last June. Apparently Oracle figures to do more than Sun with a tad less.

The increase in R&D spending is supposed to start in Oracle's 2011 fiscal year, which begins in June.

The investments will be applied across-the-board to Sparc chips (there's more on the roadmap now sans Rock), Solaris and Linux, the Sun Ray thin client (ah, remember Larry and the network computer, he's finally got one), Java middleware, 7000 ZFS storage, Flash, archiving, virtualization and software.

Oracle vowed to improve the problematic open source database MySQL that was almost the merger's undoing and thrust OpenOffice onto the web with an online version dubbed Oracle Cloud Office aimed at the same enterprise crowd that Google, IBM and Microsoft are shooting for with similar widgetry.

MySQL, part of a global business unit dedicated to open source along with InnoDB, is supposed to be made part of the Oracle stack and integrated with Enterprise Manager, Secure Backup and Audit Vault.

As much as Ellison hates the name cloud computing, Oracle is now in the cloud computing business offering the building blocks for both public and private clouds. He also trashed VMware as "point solution," lacking Oracle's integration. "VMware's not integrated with anything," he said.

The Java programming model is supposed to be extended to emerging application development paradigms like RIA. Java projects like HotSpot, JRockit, NetBeans - even Glassfish, the Java application server, despite Oracle's acquisition of BEA - live on but they it won't be Oracle's enterprise cards. JavaFX is another matter. Java ME and Java SE APIs are supposed to be unified to recapture Java's old "write once, run anywhere" formula and ME optimized for new platforms like IP TV, Blu-ray and emerging embedded devices.

Ellison claimed that it doesn't matter if Sun doesn't monetize Java. BEA and Oracle make money on Java and now Oracle is bigger than IBM in middleware. "Where the money comes from is less important," he said.

Oracle continues to maintain that it can squeeze at least $1.5 billion in operating profits out of Sun year one despite the billion in losses the company has wracked over the last decade so it's been widely assumed that the only way it can do that is by slashing and burning its way across the Sun campus.

Ellison, however, claims making Sun profitable quickly is "very easy to fix." If so, former Sun CEOs Scott McNealy and Jonathan Schwartz are sure gonna look dumb. Schwartz, by the way, is out; Ellison's looking for a place to put Scott. Apparently they're still talking about what his job might be. Can you see Scott working for Larry? Hmmm.

Anyway, one of the ways to this profit nirvana - other than limiting the number of configurations sold - a move reminiscent of 20 years ago when Oracle railed against the cost of supporting so many Unix databases - is to change Sun's build-to-stock policy to a build-to-order one, a supply chain renaissance that could take until June.

The shift - and Oracle didn't detail what the savings would be - will involve shipping from a cutback number of plants that make the hardware out of components coming from half the number of suppliers Sun used - and doing away with Sun distribution centers. The plants will drop ship.

Oracle gave the impression that Sun pissed away a lot of money on excess parts inventories, obsolescence, inaccurate forecasts and the freight to return systems that needed to be retrofitted to meet what the customer ordered.

Oracle also figures that leveraging its infrastructure will lower the cost of finance, legal, marketing, HR, procurement, IT and other back-office activities.

Oracle is supposed to use Sun's line of x86 servers only for high-end clusters rather than compete with HP and Dell for low-margin commodity sales and focus on its high-end Intel-bucking Sparc/Solaris machines. Support is supposed to be automated, standardized and simplified by the fact that the whole package comes from a single supplier that knows all its secrets. Oracle figures Sun will do better if its support attach rates are improved. MyOracleSupport will be the access portal for both Sun and Oracle users.