Tried-and-true platforms are safe and comfortable places for IT to build. But platforms aren't rocks; they're icebergs. The place you choose to stand might remain steady and anchored for all eternity. Or you might wake up one morning to find a crack under your feet and have to make a hurried choice between staying put and risking a long jump to another berg. Maybe the ice will split and crash into the sea, taking you with it.
If you believe what you read in the papers, the field of software and hardware platforms changed little, if at all, in 2003. Clients run Windows XP, little servers run Windows or Linux, and big servers run AIX. The prevailing software platforms are Windows and Java. That shallow view of the platform space frees IT to deal with larger issues, but IT will soon have to worry about platforms again.
Relationships between companies like Sun Microsystems Inc. and Oracle Corp., Microsoft Corp. and Intel Corp., and Hewlett-Packard Co. and Cisco Systems Inc. had allowed one vendor's sales team to combine partner products into solutions that seemed to come from one source. More than technology or customer demand, partnerships had determined which elements linked together to make a platform. But such tight partnerships fell out of style during the recession. Makers of servers, operating systems, frameworks, and enterprise applications have broken or loosened their bonds with each other. Vendors claimed the freedom to make money anywhere they could, forging ad hoc partnerships as required to land accounts.
Another popular tactic was to reprice and repackage enterprise solutions to reach smaller businesses. Microsoft split Windows Server 2003 into functionally limited bundles, such as the Web and Standard Editions. IBM created a limited version of its e-business platform called WebSphere Express. Sun promoted its rich Java server platform -- then called Orion and now sold as Java Enterprise Server -- as an affordable, all-inclusive alternative to the Java licensees' per-feature pricing.
In the short term, changes like these create new choices for customers. But in many cases, vendors are hedging their bets in ways that make platforms unpredictable. Sun's decisions to revive Solaris x86 and to build servers around AMD's Opteron are insurance against the potential decline of Sparc. But it's expensive to evolve and support multiple architectures. So if one doesn't sell, it won't live. You can dig in your heels and refuse to leave a retired platform like Windows NT 4.0 or Solaris 2.6. But when applications go stale, and then scarce, you'll wish you had spent your time preparing instead of protesting.
During 2003, conflicts and miscommunication among IT, vendors, and the media have created havoc in platforms. Vendors see the promise of fresh revenue, but it's not clear how to go after it. The tendency, most evident at Sun and increasingly at Microsoft, is to try everything. This makes it harder for IT shops to place safe bets.
As sleepy as 2003 was in other ways, it was a seminal year for platforms. Apple Computer Inc. blazed its own trail by sticking with a unified hardware and software platform while competitors were expanding customers' choices. The 64-bit Power Mac G5 and the Panther version of OS X did expand the Mac platform; however, the Mac platform remains consistently implemented across Apple's entire product line. That's a unique approach that harkens back to the Unix heyday, when IBM, Sun, and HP were platforms unto themselves.
By demoting .Net from a platform to a framework, Microsoft re-established Windows as a platform unto itself. The mystery of .Net gave way to the opaque muddle of the Windows Server System. The matrix of integration points and dependencies among the parts of the Windows Server System virtually forces customers to purchase enterprise software through Microsoft partners, including Itanium giant HP. That's a huge change in the way Microsoft does business, but it brings enterprise Windows in line with competing enterprise platforms from IBM and Sun.
IBM's profile in the chip market rose sharply with the shipment of the Power Mac G5, which is based on IBM's 64-bit PowerPC 970 processor. This foreshadows the delivery of the Power 5 CPU, which IBM expects to shake the enterprise system market to its foundations. Customers running IBM's minicomputers and mainframes are among the few who can be certain of platform stability. But even with fast-evolving systems based on the Power architecture, IBM is exceedingly careful to avoid marooning any customers. If you're looking for a rock to stand on, IBM is it.
New processors in the x86 space from Transmeta Corp., VIA Technologies Inc., and Advanced Micro Devices Inc. put fresh pressure on Intel as the CPU supplier for mainstream hardware platforms. AMD's AMD64 architecture, the first 64-bit x86-compatible CPU and garnering a lot of attention for that fact, is particularly troublesome for Intel. For all of the x86-32 players, and for the 64-bit architectures from AMD and Intel, Windows is where it's at. Linux and BSD continue to rise slowly as alternatives, boosted by events such as Novell's acquisition of SuSE and Apple's use of BSD as the foundation of Mac OS X. Even so, hardware platform makers spend much of their time firming up their Windows compatibility.
I won't say that's not the half of it, but there was more platform action in 2003 than I could describe here. The coming year is set to offer customers more platform choices than they've ever had. The ice under our feet is splitting and shifting, and it's anyone's guess what it will look like when it settles again. Whether that dramatic change engenders terror, confusion, or delight depends on your organization. IT should brace for change; no matter what, the landscape around you is going to heave and drift.