Mid-range machines: The Wizards of Mid

As the lines between different kinds of machine become increasingly blurred, mid-range suppliers need to pull more than a rabbit out of the hat to stay afloat, says Sean Hallahan

The term mid-range systems is something of a misnomer these days. It?s a useful catch-all description, but is no real way of judging a machine.

Ten years ago machines could be easily categorised as mainframes, minicomputers or PCs. Typical minicomputers were the Digital Vax, the IBM AS/400, the Hewlett Packard 9000 and the Data General Nova. These systems can be shoehorned into the mid-range category, but with less ease.

As the power of the PC increased it began to impinge on the space occupied by lower-end mid-range systems. That the PC companies now consider themselves suppliers of mid-range systems is not a new phenomenon. As far back as 1992, Compaq was ready to present itself as an alternative to the traditional suppliers.

In that year, management consultancy Touche Ross released The Mid-Range Systems Report. The report detailed the response of 11 companies to a hypothetical invitation to tender (ITT) document for seven different categories of mid-range system, rising in technical complexity and number of users from one to seven. Compaq was confident enough to submit itself as a candidate for the first four categories.

Compaq was the only vendor included in the report that was exclusively a supplier of PCs and ranged itself against the HP 9000, Vax, AS/400, ICL DRS3000 and 6000, NCR 3000, Sun Sparcserver, Altos System 3000, Bull/ Zenith Zserver, Olivetti LSX500 and Unisys U6000. With the exception of the Zenith-based system from Bull, all the offerings were in the traditional mid-range camp.

Not everyone is convinced that at that time Compaq was really a provider of mid-range systems. ?I think that Compaq was rather flattered to be included in that report,? says John Saw, commercial marketing manager at HP. But he believes that Compaq will be a major force in the future, even rivalling traditional mid-range players such as Digital. ?I don?t think Digital is a major force any more. The major force will be Compaq,? he says.

Mark Lillycrop, a consultant with IBM watcher Xephon, believes that the lines between mid-range, mainframe and high-end PCs have become so blurred as to be virtually meaningless. ?What really changed the shape of things was the move to client/server computing and losing the proprietary OSs,? he says. ?Unix is being squeezed by NT in the server market and there are some pretty ferocious PC servers in that sector. At the same time, IBM is pushing its mainframe sales downward to attack what was the mid-range market.?

One of the ways of defining a system as a mainframe or mid-range and, to a lesser extent a PC or workstation, was by measuring its performance in millions of instructions per second (Mips). But Lillycrop believes that Mips no longer offers a form of accurate measurement because of the blurring of the lines between the systems.

?The days when you could say a machine running at between so many Mips and so many Mips is a mid-range system is a thing of the past,? he says.

One of the factors that can distinguish a mid-range system from a PC-based system is the degree of support staff and support costs required. As IBM and the other suppliers push more of their products through the channel, the burden of providing the support has been shifted more on to the shoulders of the dealers, agents and Vars. According to Lillycrop, one of the key factors in choosing a system is how much support the customer needs in terms of manpower. Systems like the AS/ 400, which has an integrated database and a user friendly OS, require much less support than a Unix-based system, which he views as a more complex operating environment.

This analysis is shared by Paul Fryer, a senior consultant at IBM?s AS/400 software division. According to Fryer, IBM is planning even greater levels of integration with future releases of the AS/400 including Java and internet capabilities which will reduce the need for both internal and external support staff. Staff at IBM?s AS/400 design and manufacturing plant in Rochester in the US are working on more powerful versions of the AS/400 capable of supporting thousands of users.

?Instead of having two PC gurus on site, the users will have the benefit of 12,000 engineers at Rochester for support,? says Fryer.

He believes the hardware platform matters less these days. Up until 10 years ago, the choice of hardware ? be it mid-range, mainframe or PC Lan ? was the deciding factor in selecting a system. Today, with applications ported across all three platforms, it is the software that is important.

The cross-fertilisation of applications between mainframe, PC and mid-range systems means users can select the applications they want and then decide on the hardware system. The AS/400, for instance, supports 28,000 application packages, many of which also run on PC and mainframe platforms.

According to Fryer, many companies that are basing their business on the PC are living on a wing and a prayer. One of the distinguishing features of mid-range systems is that they contain levels of data security and data integrity not available on PC based systems, he argues.

This view is endorsed by Phil Payne, a consultant at Sievers Consulting. He also questions whether the term mid-range system means anything in hardware terms any more. ?Originally you had the minicomputer which you programmed yourself in machine code. I think of a mid-range system as something like an AS/400 which you plug in, write four lines of RPG and forget it for 10 years,? Payne says.

One of the distinguishing features of today?s mid-range systems which sets them apart from PCs and mainframes is that they come with packaged utilities such as back-up and security features. Payne also believes that the market is now application-driven rather than hardware-driven.

Both Fryer and Payne agree that one of the problems facing the mid-range software vendors and resellers is getting their pricing right. With hardware prices and operating systems licences continually falling, the application providers must eventually fall into line.

?They have to get their software pricing right. There is no way a company is going to pay #20,000 for a machine and three times that amount for an application,? Fryer says. According to Fryer, the advent of the network computer (NC) could provide a boost to the mid-range system providing users with an internet terminal.

Steve Larwood, business development manager at software house JBA, which specialises in applications for mid-range systems, is not as confident as Fryer that the NC will have such an impact.?The jury is still out, not least because we have not seen an NC yet. The real battle will be fought between the NT and the Java platforms,? he says. Nor is he convinced that Java will be as huge a success among JBA?s customers as its protagonists suggest.

?It is a risk to take them down the Java road. Our customers do not want to be at the bleeding edge of technology. They may want to be at the leading edge but not at the bleeding edge,? Larwood says.

JBA?s manufacturing and logistical software runs across a range of mid-range systems, the AS/400, HP 9000, Digital Alpha, RS/6000 and NT platforms. Saw shares some of Larwood?s scepticism about Java. ?It all depends on what you mean by Java. If you mean Java as a development tool kit then yes, there is a huge amount of interest. If you mean Java as an operating system it is a long way from showing that it is going to be an industry standard,? he says.

Analysts have expressed doubts about the future of the AS/400, with some predicting that it will merge with the RS/6000 to produce a single hybrid mid-range IBM system.

Although JBA applications run on other platforms, it is best known as an AS/400 house. But Larwood is not convinced. He is confident that the AS/400 will still be around in five years time, but is less sure of the future of the HP 9000. Now that HP has released its PA Risc 8000 chipset, the traditional heartland of the HP systems, it effectively joins Digital in the 64bit arena. But HP has based its future systems on the Intel Merced chip, which may alter its perception in the eyes of its customers.

Larwood also identifies the fusing of the mainframe and mid-range worlds. ?My view is that there is not a mainframe as such and there is not a mid-range as such,? he says.

Consultancy Bloor Research estimates that there are at least a million mid-range systems installed worldwide compared with only 100,000 mainframes, 10 million PC Lan servers and a hundred million PCs. The mid-range market has always been more competitive than the mainframe market, where IBM was, and is, the dominant force. Indeed, IBM almost missed out on the mid-range market altogether when the likes of Digital, HP and Data General came on the scene.

The advantage these early pioneers of mid-range systems had over the mainframe vendors was twofold. First, they were able to offer a lower cost per user than mainframe suppliers. Second, they attract ed medium-sized firms that could not afford or did not require the power of the mainframe.

The pattern was repeated when the PC was launched and brought computing power to small firms, departments and individuals. But increasingly all three platforms have ended up competing for the same space. Very few companies now manage to exist without some form of computing. This leaves all three camps targeting the same audience, the replacement or upgrade market or finding new applications, such as the internet, which will induce customers to buy new hardware, software and services.

Predicting which companies and architectures will make up the bulk of the mid-range systems in the next few years is a bit like the party game where a child is blindfolded and has to pin a tail on a donkey.

Moore?s Law, as propounded by Intel founder Gordon Moore, states that computing power will double and prices halve every 18 months. While this may be true in technological terms, it offers no real guide to the future of computing. Customer purchasing cycles do not mirror the technological cycles, they are much slower.

Although there have been several waves of change in the industry, few of them have been of tidal proportions. The first was the birth of the mainframe. The second was when Digital was founded in 1957 and established the minicomputer business. By 1996 Digital sales were approaching $8 billion. The third tidal wave was the arrival of the PC in 1981. Dismissed by many data centre managers and mid-range vendors as little more than a toy, the PC established a foothold in the market and began to eat into the mid-range vendors? traditional market. All three waves radically changed the way in which business and commerce functioned.

The industry could now be on the edge of the fourth wave with the internet, Java and the NC. This time round nobody is making the mistake of ignoring technology that is a potential rival to the established norm.

The distinction between mainframes, mid-range systems and high-end PC servers may be becoming blurred at the edges, but that does not mean mid-range suppliers are prepared to throw in the towel. Being squeezed by low-end mainframes and high-end PC servers is an uncomfortable place to be, but in some ways it strengthens the resolve of mid-range suppliers to press their case.

They may not be called mid-range systems in the future, but machines like the AS/400 and HP 9000 still have a future.