Gartner reports scary Q3 sales news for the server business Oracle took control of when it acquired Sun Microsystems, but Larry Ellison's focus on optimized systems such as Exadata is the reason Oracle needn't worry.
By Alexander Wolfe, InformationWeek
Dec. 2, 2010
URL: http://www.informationweek.com/story/showArticle.jhtml?articleID=228500162
Oracle CEO Larry Ellison is betting all his hardware chips on the emerging category of optimized systems, at the expense of the commoditized server business Oracle took control of when it acquired Sun Microsystems. That's the clear message I get by juxtaposing the shocking slide of Oracle's Sun servers against Ellison's intense emphasis on optimized hardware-software combos, notably Oracle's Exadata Database Machine.
Yet I believe this is all good for Oracle.
First, let's unravel some data points. Gartner recently released its worldwide server shipments numbers for the third quarter of 2010. Those figures are good news for the economy as a whole--worldwide Q3 shipments grew 14.2% year-over-year, and revenue was up 15.3%.
The totals also show that HP's executive-suite distractions have not impacted its server business. Hewlett-Packard remained the market leader, topping IBM with a 32.1% market share, as compared to 30.2% for Big Blue.
Note that HP's dominance was stoked by a year-over-year shipment bump of 16.3%, driven largely by HP's ProLiant family. ProLiant is an industry standard commodity line. That's not to denigrate it, just to note that it plays in tiers with tight margins, where a company can never rest on its laurels. (Indeed, in commodity markets, laurels don't matter. Low price does.)
Which brings us back 'round to Larry Ellison. Here's the money quote from Gartner's Q3 server press release:
"Oracle was the only vendor of the top five to show a decline year over year. 'Following the acquisition of Sun’s hardware business, the company now faces the challenge of preventing further declines in the hardware segment,' Gartner research director Adrian O’Connell said."
At first glance, this reads as bad news. In Gartner's Q2 figures, released in August, Oracle was number four server by revenue--after HP, IBM, and Dell--with an 8.1% market share. Going by shipment volume, Oracle was number five in Q2--after HP, IBM, Dell, and Fujitsu--with 48,000 units shipped for a 2.2% market share.
As O'Connell's comment telegraphs, things were much worse this cycle. In Gartner's Q3 tally, Oracle's "by revenue" market share fell from 8.1% to 6.2%. More glaringly, Oracle dropped entirely off of the top 5 volume list. (They've been displaced by NEC.)
Ace In The Hole
So why I am so blithely sloughing over the sagging Sun server numbers? It's because building mostly commodity servers doesn't provide a long-term strategic advantage. Integrated offerings like Exadata do. (I use the phrase "mostly commodity," because I don't want to lump UltraSparc systems in the low-end bucket.)
Integrated offerings are the future, particularly when you're talking about a vendor with an enterprise software portfolio, which is trying to make a case for the high value of its product. That's even more important in a world where a second wave of commodity pressure has emerged on the apps front, via cloud and SaaS.
So far, Exadata is looking like a smart play, the pressure from IBM notwithstanding. (Actually, what IBM is doing further validates this area.) Earlier this year Ellison said the value of Exadata deals in the pipeline approached $1 billion. That was in June, so if things remain on track, Oracle's optimized systems sales will presumably make up for much of the revenue lost by the drop in commodity servers.
There's also a technical advantage attendant to Exadata. Integrated and optimized systems have a spillover effect in that they support the more rapid advance of compute platforms than would otherwise occur in a grind-it-out commodity market.
I got some perspective on this earlier this year when I talked with John Fowler, Oracle's executive vice president of systems. He talked about some of the internal optimizations being designed into Exadata, aimed at reducing latency (the amount of time needed to access data): "What we did with Exadata is design a set of flash components, which became part of the storage server, that get used by the partitioned database query engine. . . With Exadata, we've a constructed a completely integrated appliance where we put together storage, Infiniband fabric, and servers together with the software. We've designed the storage of Exadata so a portion of the database queries are actually executed on the storage side. In fact, [we've] partitioned some of the database and put it on the storage side. So this is a complete holistic design."
Now, it's true that focusing on decreased storage latency isn't unique to optimized systems; regular servers are taking this approach, too. Still, I do believe that the high-performance patina surrounding integrated platforms drives this stuff forward very rapidly.
I also don't want to pretend that Exadata makes everything smooth sailing for Oracle on the business front. As Global CIO guru Bob Evans relates in his latest column, IBM is going directly after Oracle customers in the market for integrated and optimized systems.
But this is a good battle, at least from the perspective of people who are interested in a scenario where advancing technology lifts all boats.
Of course, the danger in a shift towards integrated hardware-software platforms is that the server pendulum swings back from the commodity towards a market filled with what are effectively proprietary offerings. Good for vendors profits, not so good for customers' pocketbooks.
Heard From Hurd?
Finally, one prospective wild card in the equation is Mark Hurd, the new Oracle co-president who formerly ran HP. Hurd knows a lot about selling server hardware, having come over to Oracle from the server sales leader.
My thinking here is that Hurd, being an astute business operations manager, might resuscitate some of Oracle's lost commodity server volume. One can argue that Oracle is leaving money on the table simply because it's been neglecting a business which wasn't managed all that well to begin with. This is another way of saying that focusing on Exadata doesn't mean you have to let your lower end lapse.
However, Hurd undoubtedly won't change Ellison's maniacal focus on optimized systems, and that's a good thing, both for systems technology and, strategically, for Oracle.
