Now that chief executive officer Leo Apotheker has made his worldwide tour of Hewlett-Packard's sites and big customers – and dodged the ongoing lawsuit between rival Oracle and former employer SAP – he is settling down to remaking HP into something more closely approximating his own image of the IT racket.
According to a report in the Wall Street Journal, Apotheker is going to rejigger HP to chase areas of the business with higher profits – and trim back on chasing unprofitable business just for the sake of padding its revenue each quarter. This is a tactic used by IBM – which was the reigning champ of the IT space for decades before HP took the crown two years ago – in the wake of nearly going bust in the early 1990s under outside chairman Louis Gerstner. Oracle has done the same kind of streamlining in the wake of last year's acquisition of Sun Microsystems, making cuts that Sun's founders and top brass were unwilling to make.
The WSJ report cites sources familiar with HP's plans, who have been briefed on Apotheker's strategic plan. The gist of the rumors is that HP will focus more on software, storage, and networking – and less on PCs and commodity servers.
It seems far more likely that HP would simplify its PC lineup than cut back on its ProLiant and BladeSystem line of x64-based servers; it has already radically cut back on its Itanium-based Integrity server line and, quite frankly, there is not much more cutting to do. But there is one possible exception.
If HP had plans for very high-end Superdome machines, which would go far beyond the capacity of the Superdome 2 machines announced in April 2010 and shipping in the fall of last year, these machines might be thrown into the volcano if there is not enough customer demand for big Itanium iron. HP has never confirmed that such machines are in the works, but there is an expectation that the company will deliver something to compete against IBM's Power 795 machine, which packs 256 cores in a single system image, and Oracle/Fujitsu's M9000, which also have 256 cores.
HP's current Superdome 2 machines are based on a modified 18U c7000 BladeSystem chassis that puts eight blades, each with a two quad-core Itanium 9300 processor, into a single system image with the sx3000 chipset created by HP. That's only 64 cores – half as many cores as could be brought into a single system image with the dual-core Itanium 9100 processors using the sx2000 chipset in the current Integrity Superdome machines. A lot of high-end HP-UX customers were probably looking for a 256-core system, which is four times as much oomph as the Superdome 2 blades offer. It all depends on how many HP-UX shops are pushing up against the performance window and how well or poorly HP can ramp up its clustering software for HP-UX to compensate if it decides to back off on big SMP iron.
Apotheker could take a more radical move and actually sign off on a port of the company's HP-UX operating system to the Xeon 7500 chips or their future "Sandy Bridge" equivalents from Intel. It would take a really good emulation environment of some sort to make this happen, of course. But Apotheker is a software guy and would no doubt throw software at a hardware problem. The one thing that HP cannot afford to do is alienate its thousands of HP-UX, OpenVMS and NonStop customers – or the suppliers of third-party applications for these platforms. If Apple can make a transition from PowerPC to x64 chips using emulation – as it did several years ago for Mac OS X applications – then HP can do it, too. (It's a pity that IBM owns Transitive, which supplied Apple with key emulation software that eased that transition painlessly and invisibly.)
The Journal report suggests that the top people at HP will be shifting jobs and responsibilities, with some executives being shown the door. Randy Mott, chief information officer at HP, was named as an executive who would be departing the company once Apotheker's strategy is set in motion. This may have something to do with a broader push by HP into cloud computing, and it is likely that someone from within its EDS services acquisition – who actually knows how to run data centers – would be tapped for such a position. HP would probably stop short of starting up its own public compute and storage cloud, since that would compete against its customers, but by the same token, EDS offers hosting and outsource services that once moved to virtualized server, storage, and networking infrastructure will compete against those cloud-builders.
If cloud computing takes off, HP is damned if it does, and damned more if it doesn't. The margins will be in the services and in the operating of the clouds, not in the peddling of the iron supporting them. If anyone knows that, it is HP. And if HP is building its own clouds – and has pretty much ceded the custom server space to Dell's Data Center Solutions unit – then it might as well go become an infrastructure cloud provider and make the money.
In fact, what HP should do if it is serious about cloud computing is acquire Rackspace Hosting. Rackspace has teamed with NASA to create the OpenStack cloud management fabric, an open-source alternative to Amazon Web Services' EC2 compute cloud and related storage clouds. Rackspace has a market capitalization of $4.4bn and would be a pricey acquisition.
The rumors reported in the Journal suggest that Ann Livermore, who runs HP's Enterprise Business, will be tapped as vice-chairman, and the services and technology businesses will be broken apart, with Dave Donatelli taking control of Enterprise Servers, Storage and Networking and Tom Hogan taking over HP Services. Both are expected to report directly to Apotheker once this change is completed. ®