The Four Hundred

Midrange Madness is not just the name of the softball team on which the San Diego contingent of Guild Companies plays. It is also the latest hyperbolic mania to grip the IT community. You've been ignored for years, and now everybody wants to be your friend. The reasons are simple. Midsized companies, which generally buy midrange servers, are in the sector of the IT market that is still growing. The midrange is also the biggest identifiable target.

It's easy enough to identify the 20,000 or so large businesses in the world. Just look at the financial newspapers and see who is being written about. Look at the stock exchange listings. With the exception of a few thousand large private companies that have seen the wisdom of remaining private, these are the large enterprises (with more than 1,000 employees) that are said to rule the world. They don't rule the world so much as they dominate global politics and bend governments and peoples to their ways on a super-macro scale. They also don't spend as much on IT as they get credit for. All told, large enterprises spend about $250 billion to $350 billion on IT hardware, software, and services globally. That's $10 million to $15 million a piece, on average. These are the big, easy targets. Unfortunately, large enterprises have been cutting IT expenditures for several years, and it doesn't look like they are going to stop.

Small businesses have fewer than 100 employees, and identifying the 25 million to 90 million of them in the world is so hard that there is no agreement on how many there are or how many IT products and services they consume. How do you tell the difference between a small business with one employee and a consumer who does a little work on the side on a home PC at night? Small businesses flit in and out of existence like quantum particles in the vacuum of space, which also makes it hard to characterize what they are doing. Nonetheless, my best guess is that, in aggregate, small businesses probably account for $200 billion to $400 billion in IT spending, which is somewhere between a few grand and $15,000 a piece per company, depending on which statistics you want to use.

This is not a target-rich environment in terms of profits, especially among very small businesses with less than $1 million in sales and fewer than 10 employees. It is too hard to reach these companies, and they aren't going to spend a lot of money. Some small companies do spend what I still consider is a lot of money on a server, but they only do it every three, four, or five years, until they grow into larger companies. For instance, a lot of AS/400 Model 170, AS/400 Model 250, and iSeries Model 800 boxes go into companies at the upper end of this small-business category. Similarly, Dell and Hewlett-Packard rule in this space with their Windows-on-Intel servers. It's unclear whether the small businesses of the world are increasing IT spending. Some people think they are, but it might just be people starting small businesses after they lost their jobs. It's not an increase in spending as much as an influx of people who can't find work and have to make it for themselves. (I say good luck to you all. Been there. Done that. Got the golf shirt with the Guild Companies logo on it.)

If any sector rules the IT world, it is undoubtedly the midrange. There are probably 250,000 to 300,000 companies with between 100 and 999 employees worldwide. (That's a little larger than the OS/400 customer base, although there is not a 1 to 1 correlation. A very large number of companies in the OS/400 installed base are, technically, small businesses.) The midrange, as defined by number of employees, accounts for between $500 billion and $750 billion in IT spending. That's about two to three times that of large enterprises. The average midrange company spends between $1.5 million and $2.5 million a year on IT products and services. This is truly the belly of the market, and the funny thing is, IT spending is increasing in this part of the market, not decreasing.

The IT market is not, as people often incorrectly describe it, a pyramid. It has a much more odd shape than that. Here's an image for you that is funny, even if it isn't a perfect comparison. If the height of the following objects is a measure of the number of companies in a particular category, and volume is a measure of the aggregate money that these companies spend on IT, then the global IT market for large, midsized, and small companies is shaped like a plate balanced on a soda can, both of which are balanced on a pin that is 60 to 180 feet long. Definitely not a pyramid.

All the IT players are turning their attention to the midrange not just because there is growth there but because they have been neglecting it. The historical midrange, which came about with the advent of proprietary minicomputers like the DEC VAX and the IBM System/3 in the 1970s, was much neglected, too. And the more vibrant midrange market that was kicked into high gear by the advent of Unix-based midrange servers in the 1980s was also a market that most people didn't pay much attention to until Unix came to dominate the server market in the early 1990s. Unix, and particularly midrange Unix, still dominates the server market in terms of the aggregate amount of money companies spend on servers and related systems and applications software. But this is changing, and Carly Fiorina, chairman and CEO of HP, got caught off guard by a rapid swing in the midrange market in the latest quarter, which caused HP to miss its financial targets.

Fiorina said that one of the reasons why HP wanted to acquire Compaq was that over the long haul HP believed that small and midsized companies would migrate away from traditional midrange gear--like HP 3000 and HP 9000 minis, as well as IBM pSeries and iSeries Power boxes, Sun Microsystems Sparc boxes, and just about any other box running Unix or a proprietary platform--and toward Intel-based servers running either Windows or Linux. HP also expected there to be aggressive moves at the high-end of the Unix market, where companies on proprietary mainframe or midrange machines would move to big Unix or Windows iron and to more modern applications built in .NET, C, or Java. Still other companies with midrange boxes of all types would consolidate workloads on big iron, cutting down on server counts and system administration costs. The advent of partitioning and dynamic provisioning was part of this vision.

This is all logical. For 10 years, workloads have not been growing anywhere near as fast as the performance improvements in servers of all makes and models. There is tremendous pricing pressure in the entry server market, thanks to Intel, and the power of a so-called entry server rivals a big box of only a decade ago and a midrange box of only five years ago. The price/performance in the entry market--which is arguably the new midrange market--is not just compelling; it is magnetic and mesmerizing. Add to that the efficiencies that come from virtual and logical partitioning and dynamic workload management, and you can get an efficient two-way or four-way entry Intel server to do as much work as an inefficient four-way or eight-way Unix or proprietary midrange server. Small wonder people are thinking small when it comes to servers. And as 64-bit processing comes into vogue on entry machines and processor speeds go up and up on both 32-bit and 64-bit machines, the need to buy a so-called midrange server will further diminish.

The problem for HP, says Fiorina, is that this bifurcation of the server market is happening a lot faster than anyone at HP had expected. Consequently, HP's midrange Unix businesses (it has two of them: one on PA-RISC, the other on Alpha) are in the tank. Sales of big Superdome Unix servers were up by 64 percent in the latest quarter, and sales of Intel-based ProLiant servers were up by 4 percent, on shipment increases of 16 percent. What is happening to HP will undoubtedly happen to IBM, Sun, and any other established midrange player. Dell doesn't have a midrange or high-end server business, and is benefiting from the trend toward smaller machines. Actually, it has pushed that trend as much as it has followed it. The point is, this is not just something HP is grappling with. This is how it is for everyone.

In the past, midrange customers (in terms of their IT solutions, not their employee count) were defined by the machines they bought to do their work. But, as I have said in recent years, the word midrange is almost meaningless these days. In fact, it has some negative connotations among IT companies and customers.

In the coming years, saying "midrange" will mean more about the customer than about the customer's iron, and that is a good thing. So is the fact that many big IT players and their reseller and software partners are willing to bend a little (though not completely backward) to help you, the midrange customer, solve some of your persnickety IT problems. A few years ago, you couldn't get them to call you. Now you won't be able to make them stop. That is an improvement, if you don't mind lots of voicemail. It's either feast or famine, midrange shops, so you might as well feast. Figure out what your problems are and make these IT players help you solve them. They are ready, willing, and able.

And if you are a reseller or software house that has been fixated on the middle of any particular platform, including the iSeries, you had better brace yourself for a shift toward a customer base that is dominated by lots of small machines and a few very large machines. This will be something of a shock to the OS/400 software community, which has traditionally targeted the big 9404-class and small 9406-class machines. The middle of the IBM midrange was always the sweet spot. If the iSeries base follows what is happening out there in server land, this is going to change. No one is saying that the iSeries is going away, but the types of machines that people will need to buy are changing, and you have better shift your products, practices, and pricing to reflect this change.