Hardware

OPINION

Linux, Microsoft and the 64-Bit Decade

Linux remains overmatched in the 32-bit world on the desktop. There is just too large a base of Windows users and related applications for Linux to dominate the market in a reasonable period of time. However, transitions breed change, and the industry is about to take a big step into the 64-bit world, where the playing field is either level or actually might favor Linux.

Unix still owns the 64-bit server space. Microsoft, as always, has moved into this space with an impressive bundle of tools, but 64-bit servers power the most conservative and mission-critical areas of computing — areas that are much more concerned with security and reliability than with the benefits associated with using Windows. In fact, 64-bit server environments tend to be highly customized, which nearly eliminates the core benefits — including ease of use and compatibility — normally associated with Microsoft’s platform.

This is not to say that Windows is not competitive. However, the dynamic that exists in the 32-bit world, which favors Windows, is upside down in the existing 64-bit world that favors Unix. Linux, meanwhile, is currently the most popular Unix derivative on low-cost hardware.

The Big Linux Mistake

Recently, Germany made a big move to Linux on the desktop. This shouldn’t surprise anyone because the country also made a big move, long ago, to OS/2. Autocratic by design, IT shops in Germany have a level of control and authority that should be envied by the rest of the world. As a result, when they make mistakes, the mistakes tend to be big ones.

I’m not saying the move to Linux is the mistake; the mistake is the move to 32-bit Linux or to Linux on a 32-bit-only platform. Both might have an inordinately short life.

Governments in general — and in Germany in particular — like to keep desktop products stable over long periods of time. We are now entering the transition period to 64-bit computing; Linux will move more quickly than Windows into this environment. This means that, once again, Germany is jumping too early and, as a result, might regret this move, much as it learned to regret the early move to OS/2.

The perception that these departments keep making the same foolish mistake likely will not sit well with the folks to whom these IT decision-makers report.

On the Desktop

So, why do we need to move to 64-bit computing on the desktop? We could have asked the same question of 32-bit computing, and we likely will ask the same question of 128-bit computing a decade from now. The right answer probably has more to do with the need to step away from the mistakes of the past than anything else.

This move to 64-bit computing will let Microsoft create a product that steps away from the unreliable and unsecured parts of its current code base without incurring the wrath that typically would be associated with such a move. In addition to the opportunities presented to the company, there is a compelling argument for Microsoft to change directions and make better decisions.

Microsoft executive management undoubtedly would like to be able to look back at the dawn of the 64-bit era with no regrets. The company also would like to live in a world in which its products were once again revered rather than reviled. To do this, it must step sharply away from the code base that has spawned its most pressing problems. Clearly, the company appears to be moving in the right direction.

However, Linux, which doesn’t have as large an installed base of hardware or as massive a number of packaged applications on the desktop, can more quickly move to the 64-bit platform, establish a strong beachhead and potentially place Microsoft in a subservient position.

This idea has to be compelling for those who are driving the Linux agenda. If they can strengthen the GPL and overcome the distraction that SCO represents — and SCO really is just a distraction — they have a shot at making this a really interesting decade.

Intel vs. AMD

Just as Linux is currently the favored challenger to Microsoft, AMD is the favored challenger to Intel. When we did an unscientific review of large-scale Linux deployments a few weeks ago, we found that an inordinate number were using AMD rather than Intel-based hardware.

This was generally because these companies, once they decided to move from one of the branded Unix variants — virtually all were migrations from Unix, and most were from Solaris — also decided that they wanted to save as much money as possible, and the result was AMD-based hardware.

On the desktop, the argument could be even more compelling because, unlike Intel’s Itanium, which is optimized for 64-bit performance and is clearly not a desktop component, the AMD Athlon 64 blends high 32-bit performance with 64-bit capability, making it a better transition component, and one that is ideal for a market in transition. In addition, the AMD Athlon 64 is priced well for the desktop.

For those who are living in a 32-bit world but are developing or expanding into a 64-bit world, the Athlon 64 is a better choice. Once the market has moved fully, probably around 2007 or 2008, it is likely that a more dedicated technology will be needed, but the Athlon 64 gives the AMD processor a solid five or six years of life, while the Intel Pentium will start to look out of date in the next 24 months.

The 64-Bit Decade

Clearly, the Athlon 64 won’t be as mobile as Intel’s Centrino offering and will fall more into the high-performance desktop replacement market, where most early adopters can be found. However, what is making the AMD platform even more compelling is that it promises a single image for both mobile and desktop implementations — whether we are talking Windows or Linux.

By now, you are getting a sense of why I’m calling 2000 to 2010 the 64-bit decade, which might also be the decade for Linux. This is the decade in which Linux will either make it or — like OS/2, the Mac OS, NetWare and branded Unix — fade into the world of might-have-beens.

Regardless of whether or not Linux steps up, Microsoft clearly is doing so. If nothing else, we are undoubtedly looking forward to a much healthier hardware market, not to mention more competition, than we have enjoyed for some time.


Rob Enderle, a TechNewsWorld columnist, is the Principal Analyst for the Enderle Group, a company founded on the concept of providing a unique perspective on personal technology products and trends.


4 Comments

  • The other argument favoring Linux not only over Windows but all other proprietary OS’s in the 64-bit space is its multi-architecture support. Linux (like NetBSD) supports a whole host of 64-bit processors including Itanium, Opteron, UltraSPARC, Power, Alpha and others. You really can consildate your entire enterprise on a single OS across all hardware platforms.

  • Referring to Linux as a "derivative of Unix" is likely an error that is bothersome to many, including myself. It’s a workalike–not a derivative. What’s particularly irritating about that mistake is that it’s what SCO has been saying, because they believe some of their code was illegally copied into it, after the 2.2.x kernal series. Hence, it sounds like the author of this article is legitimizing their claim.
    .
    Those German cities may not regret their moves–it depends on how they deployed. SuSE is very well refined desktop system and works great. What ever Windows applications they need, that are not available for Linux can be provided in a variety of ways–most notably via terminal services. That’s a far virus safer and more manageable way of providing applications, anyway.
    .
    In the article, the author also seemed to imply that Windows was working toward 64-bit and that Linux could catch up, easily. Linux was the first OS to compile on Itanium and for 64-bit on the Opteron. Linux had long offered 64-bit support on other platforms, such as the Alpha (very prominent) but also all other major processor types…and most minor ones.
    .
    Microsoft had support the Alpha processor, but mostly in 32-bit mode. They spend many years attempting to get Windows NT (the predecessor to Windows 2000, XP, etc.) to work more and more in 64-bit mode but made very little progress. It was a story of high hopes and mostly failures.
    .
    Matthew

  • The 64-bit transition actually argues in *favor* of a move to Linux now — since Linux and Unix have been established on 64-bit systems for years. Linux is actually a more stable long-term platform than Microsoft, given that Linux applications are a mere recompile away from working on just about any architecture, and application porting could and did start years ago.


    Later in the article, Enderle says,
    "This move to 64-bit computing will let Microsoft create a product that steps away from the unreliable and unsecured parts of its current code base without incurring the wrath that typically would be associated with such a move. In addition to the opportunities presented to the company, there is a compelling argument for Microsoft to change directions and make better decisions."


    So, a New Windows. That’s not a bad thing, considering all of the bad design decisions, emulation and legacy cruft in everything up to and including XP. But, if Windows is changing its APIs and Linux isn’t, isn’t it better to move to Linux now to help ensure forward application compatibility?

    http://www.flyingbuttmonkeys.com/n/index.php?itemid=22

  • To QUOTE:
    So, why do we need to move to 64-bit computing on the desktop? We could have asked the same question of 32-bit computing, and we likely will ask the same question of 128-bit computing a decade from now. The right answer probably has more to do with the need to step away from the mistakes of the past than anything else.
    UNQUOTE:
    No, Rob, the reason we moved from 16 to 32 bit computing is that 16 bit computing limited the CPU to a maximum linear address segment of 64k, while 32 bit computing expanded that to 4 gigabytes. Quite a change. While 64 bit computing lets us break the 4 gig barrier with ease, the fact that current 32 bit machines can already address more than 4 gigs (albeit non-linearly) with much more ease than their 16 bit ancestors could access >64k means that 64 bit computing is only really needed right now for things with a large memory model.

Leave a Comment

Please sign in to post or reply to a comment. New users create a free account.

More by Rob Enderle
More in Hardware

Technewsworld Channels