June 25, 2004
HP announced this week that it has helped more than 100 companies worldwide move business-critical applications from mainframe systems to HP solutions. According to the company, newer mainframes have roles to play in some specialized niche environments, but HP’s solutions and Adaptive Enterprise strategy offer more cost-effective and flexible business and IT synchronization. To that end, HP is pitching solutions including Proliant, Integrity, and Non-Stop systems, with claims that for many customers, mainframe-housed data can be better utilized via Microsoft, BEA, or Oracle applications running on HP servers. Additionally, HP is also providing IT assessment and financing services tailored for mainframe client migrations. Among the companies HP said it has helped are CitiStreet, Atmos Energy, Aquila, Hyundai, GM Daewoo, KorAm Bank, KCM, and ABSA Bank. No information on the size, timing, or completion of these migrations was included in the announcement.
0Attacks on mainframe computing are regular occurrences in the IT world. A couple of times a year, a vendor such as HP or Sun, which lacks anything resembling a mainframe, noisily argues its theoretical superiority and trots out a handful of glassy-eyed converts. Overall, such antics tend to qualify as second-rate theater, with an ulterior motive standing in the wings; snatching a piece of the mainframe market which, depending on who is talking and how much they have had to drink, encompasses some 60-80% of all computerized data worldwide. That is a lot of motive any way you slice it. Curiously, announcements regarding the imminent death or overthrow of the mainframe tend to be a bit sketchy on details, though often with good reason. HP’s top two featured migration clients, CitiStreet and Atmos Energy, are good examples. HP originally announced CitiStreet’s shift to Itanium-based SuperDome servers in November 2003, but a coinciding news story explored the company’s original move to HP in 2001, and explained that the latest news concerned an upgrade from PA-RISC-based SuperDomes to newer models. According to Interex, an International HP Customer Community newsletter, Atmos began its move from mainframes to HP 9000 servers in 1996, long before Adaptive Enterprise was a twinkle in HP’s eye. The specifics of HP’s other migration deals are unknown, but we hope they are a bit less ossified than this pair.
PR blather aside, is there anything to HP’s claims that its offerings can adequately replace “aging” mainframes? It depends. IBM celebrated the fortieth birthday of the S/360 mainframe this year, and the fact is that older IBM and other mainframes continue to do their work worldwide. IBM encourages customers to upgrade aging mainframe systems to new eServer zSeries solutions and has developed offerings aimed at businesses with limited or legacy mainframe needs. At the same time, it is natural enough for those customers to consider (often with the assistance of IBM competitors) alternatives. Given the ongoing evolution of IT of every stripe, replacing mainframes with lower cost systems is an unoriginal strategy that can be espoused by virtually every enterprise IT vendor including IBM. In fact, we expect that in shops with limited mainframe requirements, such offerings might seem pretty decent on the surface, but they come at a price beyond the bottom line. In 2002, Carly Fiorina made her then-controversial assertion that many cost-conscious IT customers will always gravitate towards “good enough” solutions. While today’s Intel-based solutions are considerably more sophisticated than those available when older mainframe systems were deployed, they still do not offer mainframe-class availability, security, resilience, or reliability. However, so long as significant numbers of traditional mainframe customers dither about their upgrade paths, we expect some vendors to continue to trot out this melodramatic homily time and again. At the end of the day, “good enough” solutions may be good enough for HP, but it is unclear how many enterprises that have relied on mainframes for years will feel the same way.
Microsoft indicated recently that it is on target to deliver its own branded antivirus offering, but gave no specific dates for when the product would be available. Microsoft said that it would sell the product separately from its Windows offerings, and not include the product in its core desktop offering. Late in 2003, Microsoft acquired a Romanian antivirus company, GeCAD Software, with the intention that the company would develop a Microsoft-branded product. In related news, news reports have stated that Microsoft was on the verge of buying Network Associates, one of the largest of the antivirus and security companies. That news report apparently prompted an email from Network Associates’ management to employees denying that the company is on the block or that Microsoft is going to purchase it. Microsoft has made no comment on the possible acquisition; the management-to-employee email was the only means by which Network Associates has denied the reports.
Microsoft needs to make some sort of public relations noise about security since it has been battered over the past few years concerning security breaches and vulnerabilities. Microsoft has responded by announcing that security of its products is a top priority and has been very active in distributing patches and fixes for the Windows XP and Windows Server 2003 operating systems. These automatic updates – most in response to discovered vulnerabilities – have become a near-daily part of the Windows computing experience. While each new update is an ongoing attempt to deal with new vulnerabilities, each was also is a reminder that the Windows operating system has apparent weaknesses that can be exploited.
This makes us wonder. Will a Microsoft-branded antivirus be appealing to the market? Antivirus products are largely a commodity. They have detection engines and constantly updated virus definitions. Some new features or functionality have been added over the years – including automated updates as new threats become known – but for the most part all the vendors offer similar products and research behind the products to minimize the time between the discovery of a piece of malicious code and the distribution of the appropriate antidote to the user’s computer. In this way, the antivirus market looks quite a bit like the soft drink market, with brand loyalty the determining factor when choosing between similar products. If Microsoft did indeed buy Network Associates, they would be foolish to do anything that would diminish the McAfee AV brand. NA has had their own strugglers in this arena, seeking to expand the company beyond antivirus while still keeping the valuable McAfee brand alive. Regardless of what Microsoft does in the antivirus marketplace, given equilibrium in how antivirus software is marketed, we believe people will have little incentive to move away from the brands they now use, be it Trend Micro, Panda, F-Secure, Symantec, or McAfee.
Top500 announced the latest results of its semi-annual ranking of the world’s fastest supercomputers, the twenty-third edition of the list compiled by scholars at German and U.S. universities and research labs. The top ten systems include the stalwart NEC Earth Simulator, which has held the top spot since 2002, as well as prototypes for IBM’s BlueGene/L system in the fourth and eighth positions. An Itanium-based system at Lawrence Livermore national Laboratory holds the second position, while tenth place is occupied by the first Chinese entry to enter the top ten: an Opteron-based system built by Dawning. Trends in the list include notable increases in overall performance, with almost half of the systems listed exceeding the 1 Tflop/s mark. In addition, the system ranked 242nd on the list published six months ago now stands at 500th, marking a record turnover rate in the list’s eleven-year history. IBM led the list with 45% of the systems and 50% of installed performance, while HP was second with 28% of the systems and 18% of performance. The number of Intel-based systems jumped to 287, up from 119 in the list published a year ago. IBM’s POWER processor (#75) was the second most common platform, followed by HP’s PA-RISC (#57). AMD’s Opteron processor showed the greatest overall gain, moving from four systems in the previous list to thirty-four, four of which placed in the top 100 systems. In an unrelated event, news reports stated that Microsoft has confirmed plans to develop an OS for supercomputing environments. Windows Server 2003 HPC Edition will include features for supporting clusters of machines connected by high speed networks. Microsoft has enlisted the support of vendors including IBM, Intel, HP, Dell, and AMD, and expects to ship the new product in the second half of 2005.
Beyond providing benchmark testing geeks something to argue over, the Top500 list offers the IT industry the nearest thing to a concrete assessment of the evolutionary progress of computing. At the same time, the list provides various vendors regular opportunities for joy and/or despair. Along with overall trends towards more/faster supercomputing progress across the globe, the twenty-third edition of Top500 gives IBM and Intel a lot to smile about. IBM has become the obvious horse to beat in this space, profiting both from the performance of its own native POWER platform and Intel’s surging fortunes. Intel also has much to crow about, with HPC trends clearly playing to the company’s core strengths. However, we noted one dent in the company’s apparently unblemished façade; while the overall number of Itanium-based systems on the list more than tripled (from nineteen to sixty-one) over the past six months, only one was added to the top 100 (from #5 to #6). Given Intel’s increasing promotion of Itanium as an HPC powerhouse, we find this a bit curious, especially in light of the considerable progress AMD’s 64-bit extensions-based Opteron platform has made on the list.
Beyond slicing and dicing performance data, we find Microsoft’s interest in the HPC space notable. As the price/performance improvements continue to drive supercomputing technologies downstream, we expect them to garner the increasing attention of a variety of commercial ISVs, including Microsoft. However, there is one little word that likely lies at the heart of Redmond’s interest in what is an essentially nascent business opportunity: Linux. To date, cutting-edge supercomputing and HPC development is largely conducted at universities and research labs, which are also hotbeds of Open Source enthusiasm. Unless Microsoft wants to risk exclusion from upcoming supercomputing commercial markets, it needs to get off the dime and call in a few markers from partners and pals. That seems to be the path the company is pursuing, but it is far too early to tell whether supercomputing developers and potential customers will be interested in tagging along.
MediaLive International, the organizer of the COMDEX trade show, has announced that the North American event has been cancelled for 2004. The company indicated that falling attendance from both vendors and attendees led the company to decide to cancel this year’s Las Vegas event. The company stated it would try to put on a show in 2005, but only if a consortium of vendors would rally support for the show. MediaLive said that international versions of the show would go on as scheduled. COMDEX was first held as the COMputer Dealers EXposition in 1979. MediaLive is the latest incarnation of the Key3, which emerged from Chapter 11 bankruptcy last year.
COMDEX’s demise does not really come as any sort of shock to most people, as the trade show became less and less relevant to IT vendor marketing efforts. While many may point to the show’s demise beginning in 2000, it seems that the show was on increasingly shaky legs prior to that, only to be sustained by the mania surrounding the Internet boom. In other words, COMDEX was kept alive by means of artificial life support.
The seeds of COMDEX’s demise seem largely to reside it its most prominent claim to fame: its size. COMDEX at its peak was larger than any other North American trade show, in terms of vendor booth holders and attendees. It was annually held in Las Vegas because that town was the only one in North America that had convention and hotel facilities to handle the crowds of 200,000 plus that dutifully trudged across the seemingly endless air-conditioned landscape that was the show. At its peak, COMDEX simply became too large and without any real focus. Vendors of all types — many with only the slightest of connections to IT — plunked down good money for booths they manned with scantily clad bimbos or magic acts. Thus COMDEX became a big tech and high-priced version of a traveling flea market, one in which interested buyers had to wade through a lot of junk to find a real prize. Vendors trying to announce important new products found their messages buried in a blizzard of press releases; there too, quantity overtook quality. As for its real value to end users, COMDEX largely became a perk for good employees, an opportunity to play among the bright lights of Sin City, a value proposition that faded as time passed. More importantly, in an age where online product demonstrations are as easy as an Internet connection, the allure of watching a canned demo at a noisy vendor booth while in the clutches of an overly aggressive sales rep lost its allure as well. Goodbye, old friend, and good riddance.