Despite industry doldrums and the glacial pace of IT spending, 2002 saw continued innovation and technology adoption at every level of the IT stack. From processors to worldwide networks, crucial transitions—such as those from 32- to 64-bit computing and from intermittent to always-on connections—gathered momentum. Acceptance of these and other new technologies may even have been aided by IT builders respite from the relentless expansion—with its lack of spare time to evaluate new options—of the previous boom.
After 17 years at the core of the PC platform, Intel Corp.s 32-bit iAPX386 instruction set may be ready to retire from the heavy lifting of new Web services and data mining applications—but not without a war of succession between the designated heir, Intels own 64-bit Itanium, and the challenger from Advanced Micro Devices Inc.
The release in 2002 of the Itanium 2 demonstrated Intels commitment to an ambitious platform with a completely new instruction set, demanding a major new software investment, but the Itanium has met with a lukewarm reception from system builders and software vendors. Although the Itanium has a promise of support from every major database provider, Itanium systems have yet to prove their cost-effectiveness in eWeek Labs tests of the Itaniums limited application portfolio as of years end.
Meanwhile, AMDs Athlon 64 and Opteron processors—introducing the 64-bit x86-64 superset of the 386 instruction set for desktops and servers, respectively—ended the year on track toward the companys promised combination of next-generation performance and previous-generation compatibility. Still open, though, is the question of x86-64 support from system software and enterprise middleware providers, a much-needed follow-on to the immediate, enthusiastic response of multimedia and gaming software developers.
Not waiting for either of these options to get dry behind the ears, a wide range of tasks formerly hosted on supercomputers moved in 2002 to Linux clusters. DaimlerChrysler AG, for example, estimated a 40 percent savings by shifting vehicle-collision simulations (representing 70 percent of its engineering groups computing workload) to a Linux cluster of 108 Intel dual-Xeon workstations. Meanwhile, Oracle Corp. and other major vendors rolled out key products targeting the cost-effective Linux cluster platform, and developers enjoyed low-cost, cluster-compatible tools from vendors such as Macromedia Inc. with its midyear launch of JRun 4.
At the opposite end of the hardware spectrum, efficient chips such as Intels 200MHz PXA250 extended the reach of the ARM instruction set beyond the Pocket PC platform into the competing realm of Palm OS 5—the long-overdue software response by Palm Inc. to the superior enterprise capabilities of Microsoft Corp.s Pocket PC 2002. The year ended with the release of Palm OS 5 devices, such as Sony Corp.s $600 PEG-NX70V, which set the pace for innovative handheld design and superior ergonomics with features such as an integral digital camera and a swivel-flip keyboard.
Keeping the pressure on Palm Inc. and its licensees, Hewlett-Packard Co. was quick to arrive at Comdex in November with a new iPaq—priced below $300—that successfully combined the best features of the former Compaq Computer Corp.s established PDA (personal digital assistant) line and HPs now-retiring Jornada into an attractive, affordable Pocket PC.
At this price point, HP competes head-to-head with a new PDA line from Dell Computer Corp. A bit of cost analysis, however, suggests that Dell is buying market share by taking a loss on every unit sold, while failing to understand that the enterprise and consumer PDA markets demand distinct solutions—a lesson finally learned by Palm (as shown by its launch of the Tungsten high-end brand) and already well-understood by HP (whose Comdex offerings also included new high-end units, with features such as integral fingerprint scanners to protect sensitive enterprise data).
Pocket PC Integration
We also saw the first integration of Pocket PC capability into communications devices from providers such as T-Mobile USA Inc., with positive response from early adopters and plummeting prices (quickly falling to as low as $299 after rebates).
All forms of wireless connection enjoyed growing enterprise mind share in 2002, with built-in 802.11b antennas becoming checklist items in new laptop machines and with new antenna technologies—from energy-focusing phased-array installations to Pringles cans—enhancing the range and improving the privacy of transmissions.
Wireless insecurity, more a function of sloppy administration than a fundamental flaw in the technology, hampered deployments in 2002 but will be addressed in 2003 by stronger encryption and more-streamlined management tools.
More wireless bandwidth is also on the way. While 802.11b clearly became a pervasive standard this year, as defined by such informal metrics as breaking the $100 price point at Frys, the higher-capacity 802.11a (54M bps in the 5GHz band) and 802.11g (with speed comparable to 802.11a but more compatible with 802.11b because both use the 2.4GHz band) gained traction with equipment providers and standards bodies. By moving to the higher radio frequencies, 802.11a escapes much of the growing congestion on the band also used by everything from Bluetooth to baby monitors; equipment builders have been quick to provide dual-band (802.11a and 802.11b) devices that may blunt the appeal of 802.11g before it enters the market.
Higher on the IT stack, the operating systems arena saw Windows XP selling 32 million retail and pre-load installations in its first six months, gaining quick acceptance in the home market but a cautious response from enterprise buyers. Intrusive license provisions in areas such as digital rights management, introduced by stealth along with the first Windows XP Service Pack, cemented the resolve of many enterprise buyers to retain Windows 2000 indefinitely—or consider non-Microsoft alternatives.
The subject of service packs remained a sore point with enterprise IT administrators, who told eWeek Labs that a growing fraction of their time was being consumed in identifying, evaluating, accommodating and deploying software updates.
Kevin Baradet, an eWeek Corporate Partner at the S.C. Johnson School of Management at Cornell University, in Ithaca, N.Y., used apocalyptic language to describe the problem. “The patches came up all over the land of computing,” Baradet told eWeek, “and settled in all the territory of applications and OS; they were very numerous. There had never been so many patches, nor would there be again any less.”
We noted during the year, however, that the September 2001 Nimda outbreak is still being cited by many vendors of patch management and other security tools, leading us to wonder if prevention and remediation—firewalls for the former, improved tools for the latter—are breaking the back of this problem. If so, an oft-cited argument for adopting an alternative platform will lose some of its vigor, especially given the increase this year in successful exploits aimed at non-Microsoft targets.
One such alternative, receiving surprisingly little attention from all but the Apple Computer Inc. faithful, was Apples Mac OS X. In 2002, the company did what the IT community has been demanding from other providers for years: It upgraded the Unix-based Mac OS X this fall to whats now clearly a ready-for-prime-time Version 10.2, which Apple will promote in 2003 to become the default Mac OS environment rather than an early-adopter experiment.
Apple is offering developers a reliable Unix-family platform with substantial open-source elements, fronted by a mainstream graphical interface and an end-user market complete with applications including a native version of Microsoft Office. Large-scale Mac OS adoptions, such as the state of Maines statewide seventh-grade iBook program (funded in part, ironically, by a Gates Foundation grant for associated teacher training), should be watched closely by enterprise buyers who want to know if they still have an operating system choice.
For those who want to exercise their choice and cut their desktop costs to the bone, Red Hat Inc. announced plans for an early-03 desktop distribution of Linux. With Mozilla browser code and OpenOffice productivity applications also emerging as ready to roll this year, and with the K Desktop Environment and GNU Network Object Model Environment GUIs giving those applications a comfortable home, 2002 was the year in which desktop Linux became a credible proposition.
Having done the unlikely by putting its Unix-based operating system on both desktop and portable personal systems, Apple did the obvious by entering the Unix server space with its sleekly designed Xserve: a rack-mountable 1U (1.75-inch) server that inspired the adjective “unparalleled” from eWeek Labs for its combination of functionality, usability and affordability.
This year, the workload borne by Xserve and its server-room brethren tilted toward the pumping and processing of XML-punctuated text. XML continued its relentless expansion as the foundation of exchanging data and loosely coupling distributed systems by means of Web services protocols. Microsoft announced that its Office 11 would make extensive use of XML to enable highly automated interactions—and also controversially announced that this Office upgrade would leave Windows 9x users behind.
For developers who preferred to build their applications on less rapidly shifting foundations, Java 2 Enterprise Edition tools such as Oracles JDeveloper (the first product of that name to offer Oracles own tool set, not a tweaked and relabeled version of Borland Software Corp.s excellent JBuilder) began the year by raising the bar for extensive use of XML throughout the tool environment as well as in applications.
We were less impressed by the XML visualization tools in Microsofts otherwise-outstanding Visual Studio .Net, but we got an early Christmas present with Version 5 of Altova Inc.s XML Spy editor and its graphical schema editing tools, grid-based XML data editor and graphical Web Services Description Language file editor that augments XML Spys useful Web services debugger.
Keeping all that XML where it can be efficiently used became the high-profile task of database servers this year: Oracle9i combined XML storage with strong relational query performance, although were still looking forward to 2003s expected emergence of an XQuery standard in preference to this years tentative implementations.
In the meantime, enterprise users are liking the new climate that XML is encouraging. “The good thing about XML is, its allowed us to use a standard engine, an XML parser, to look inside definitions and data stores—although we still have a lot of work to do to make sure that were interoperable between two people in the same industry,” said Gary Gunnerson, an eWeek Corporate Partner and IT architect at Gannett Co. Inc., in McLean, Va., in qualified praise for XMLs rapid progress.
Adoption of Web service application models proceeded at a brisk pace in 2002, although this success was partially obscured by overemphasis on user-oriented services—for example, Microsofts Passport and HailStorm (the latter universally acknowledged as perhaps the worst code name ever). Internal application integration proved a far more credible proving ground for these tools, as business models for their use in the open market continue to take their time emerging.
One Web service that received mainstream attention was elgooG, the search site that spelled Google backward and accepted reverse-spelled search terms. The service used new Web service APIs to return reverse-spelled results—a ploy to evade Chinas Internet censors and an elegant demonstration that the Internet refuses to be contained.
Lest the industry get too frisky in adopting platform-neutral standards and loosening the ties to its platforms, Microsoft—fresh from its antitrust victory in the same week—launched at the beginning of November its long-awaited Tablet PC platform, complete with a resource-intensive data type that only Tablet PCs can use.
What XML-based Office 11 promises to grant, the Tablet PCs Digital Ink could take away. It will make Microsofts OEMs happy, though, by consuming more processing power, memory, disk space and bandwidth to send the same notes that might otherwise have been typed, e-mailed, stored and indexed at a fraction of the cost.
In eWeek Labs November tests of six Tablet PC designs, about the nicest thing said of any of them was that the Tablet features didnt actually interfere with normal PC tasks. None of the six tested systems delivered what the Labs believes to be a necessary combination of accustomed laptop PC performance, go-anywhere notepad light weight, convenient switching from keyboard to stylus-only configuration and competitive price.
The computationally intensive Tablet platform wants to be a go- everywhere device, but that demands the kind of power parsimony previously promised only by purpose-built portable processors, such as the Pocket PCs ARM-family chips.
Transmeta Corp., in its Crusoe technology, sought to deliver comparable efficiency in hardware by handling x86 compatibility in software, with results in our first tests of HPs Compaq TC1000 Tablet PC that were literally an order of magnitude slower than those of Intel-based alternatives. We hope to find that the software shipped to customers in 2003 will improve on the pre-release version of Windows XP Tablet Edition that we tested on this unit, but it has a big gap to close.
Meanwhile, Intel announced its Banias initiative to design x86-compatible processors from the bottom up for low power consumption, rather than settling for trickle-down efficiency as a byproduct of shrinking feature sizes on successive generations of general-purpose chips.
Transmeta, therefore, goes into 2003 on the defensive, although its approach is paying surprising dividends in relatively predictable server workloads—and thereby making Transmeta an attractive option for high-density server blades, where power consumption translates to undesirable heat. Improved connectivity strategies and increased processing power will broaden blade applications in 2003.
More widely dispersed than server blades, but still representing a huge number of computational nodes, are the always-on but lightly used PCs in retail store displays. At years end, Gateway Inc. announced a plan to commercialize that power as a grid computing utility charging by processor minutes, a literal IT utility after years of hearing that phrase used as a figure of speech. IBM and Sun Microsystems Inc. also made substantial commitments to this concept during 2002.
IP networks are the framework that makes such things possible, and enterprise demands for voice communication and shared data storage are turning toward the IP network as the place where bandwidth can be most cheaply had and most readily managed. Throughout 2002, eWeek Labs has seen network equipment makers enhancing wiring-closet equipment with the IP quality-of-service and management capabilities formerly reserved for high-end switches.
During 2002, that IP plumbing carried a growing load of storage and voice traffic. In eWeek Labs tests, Cisco Systems Inc.s SN5428 router demonstrated the low cost but still relatively poor performance that still hampers IP storage acceptance. IP offload adapters boost performance but at a cost. IP storage price and performance gaps should be narrowed in 2003.
Instant messaging also contributed to growing Internet traffic in 2002, with enterprise IT adopters cautiously accommodating its popularity with the users they support. “There arent the things in IM that weve built into corporate enterprise e-mail,” said eWeek Corporate Partner Frank Calabrese, manager of PC strategy and services at Bose Corp., in Framingham, Mass. With e-mail, Calabrese said, “you get a message, you know its me; you send a message, and you know you sent it to me. IM has no audit logs, no message logs, no encryption. There are pieces missing that make it unacceptable as an enterprise product.”
We expect to see IM solutions in 2003, from providers such as Microsoft and America Online Inc., that directly address the concerns of Calabrese and other corporate users. In general, however, enterprise ITs demand for security and ease of management—rather than bells and whistles—is the post-Sept. 11 legacy that significantly shaped IT adoption during 2002 and will continue to influence IT spending even when budgets regain their vigor.
Technology Editor Peter Coffee can be reached at [email protected]