Back to Business: Wintel Problems

There is a huge potential market for OS X in business, where Unix is strong. This article looks at Wintel problems from the Unix standpoint. Next week's article will look at IT departments, the other Unix vendors, and the opportunities for Apple.

Microsoft's problems

Microsoft has real problems in penetrating the Unix stronghold of the large systems market. Windows doesn't scale well, currently lacks a 64-bit version, and has an appalling record for security. Until these problems are sorted out, major organisations will steer clear from deploying their critical systems on Windows.

Indeed, Microsoft's bad reputation for security is likely to derail .NET. Companies won't pay the Microsoft tax unless they feel transactions and systems are secure. Perhaps Microsoft has realised this, and this is behind the announcement that they will examine their millions of lines of code for security problems.

If, after they announce their review is complete, security bugs continue to appear regularly, Microsoft will lose much of what credibility it has in this market. In the meantime not many companies are likely to deploy new systems based on unproven and probably insecure technology when they know the systems will have to be changed or updated in a matter of months.

This lack of security shows up in the widespread viruses in the Windows world. For companies these are costly to eradicate whenever they get a foothold and, if they are malicious, expensive in terms of replacing lost work and lost files.

IT workers also have to install security patches on Windows servers. As these patches are issued frequently, most IT departments are trapped between the large expense of continually updating large numbers of servers or the potential cost of dealing with a major problem if the patches aren't installed.

Microsoft is also annoying companies with XP pricing. Most companies like to deploy a single version of an operating system or program. Then, a few years later, when there have been significant improvements (from the company's point of view), they will upgrade. Many companies currently skip versions and buy licenses when they want them. This reduces software costs and staff time spent upgrading systems.

With XP, Microsoft is charging a yearly levy on the number of licences. The charge includes the upgrades for Microsoft's PC software, but where companies don't continually upgrade software it will be -- surprise, surprise - more expensive.

This XP levy is making more companies look seriously at Linux. They already have the hardware, but few, outside technical areas, will accept Linux as a desktop replacement for Windows. Linux should, however, continue to make gains in the server space.

Remember that although Microsoft has problems, these are only problems limiting growth for the next year or so. Nothing is going to stop Microsoft growing its cash mountain and acquiring companies in strategic areas whenever it feels it needs them.

The Justice Department, under the current US administration, has already shown that it has no wish to hold Microsoft back.

Intel's problems

In the 32-bit market, Intel is successfully pushing the MHz envelope with their Pentium 4. Its chips have now reached 2.2 GHz and the roadmap shows speed increases for the foreseeable future. It owns the x86 desktop and the low end server markets. Its only successful competitor is AMD.

In the 32-bit market, margins are tight on chips for consumer PCs and standard desktops, which are fine for most commercial users, and the markets for the latest high performance chips are limited.

Intel wants to take over the 64-bit market. It has gathered an impressive bunch of OEMs: Dell, HP, Compaq, IBM, SGI, NEC, etc. The problem is that few organisations want to buy the Itanium.

According to Gartner, in the quarter ending September 30 only a few thousand Itaniums were sold through to customers. Most of these went into two 1,000 processor systems sold by IBM. Sales figures for the following quarter suggest that under 5,000 Itanium-based workstations and servers had been sold.

Intel has tried to gloss over this by suggesting Itanium is for evaluation (after an estimated expenditure of $1bn+) and that McKinley (scheduled 3Q02) is when the mass buying of systems will start.

Although the Itanium clock speed is fine, customers buy Unix systems on overall performance. In this area Itanium was generally rated below the OEMs proprietary chip ranges. As the Itanium was only marketed with various Unices and is, from the customer's viewpoint, version 1.0 of the chip, they clearly felt Itanium was not suitable to replace existing workstations and servers.

When the 64-bit version of Windows appears on McKinley, the Windows groups in IT departments will also be a market. But will they really want to go with v1.0 of a new Microsoft product?

According to a recent San Jose Mercury article there is a skunkworks project in place, just in case the stampede doesn't happen. Intel would then look at using a similar 64-bit approach to AMD, running the 32-bit Windows code native instead of emulation. These alternatives won't be in place before 2003. Then the OEMs will need to decide whether to go with AMD, with Intel, or if it makes more sense to continue with their own proprietary chips for the time being.

If McKinley fails to gain enough sales, Apple is likely to have little new competition from the Wintel world for the next 12-18 months in establishing Mac OS X machines as workstations in the Unix market.

Next week: IT departments, other Unix vendors, and Apple's opportunities.