Pizza-box servers, what's really behind IBM's blades?

We're about to pitch off the Peak of Inflated Expectations and into the Trough of Disillusionment. Matt Loney knows the way out.

Over the past year, we've heard a lot about blade servers -- servers contained on a single card that plugs into a rack mount chassis so you can run dozens side by side in a very small space.

One prediction I'll make about blade servers is that we'll hear a lot more. Why? Because the economy is slow, server sales are low, and with the dot-com boom now bust, server manufacturers are desperate to find a new excuse to sell more servers. But do we really need them? Servers, like every other piece of tech equipment, go through phases. The best way to understand these phases is, I think, a little diagram drawn up by analyst firm Gartner Group, which they call the Hype Cycle of Emerging Technologies.

Gartner's diagram maps the visibility of an emerging technology against its maturity; it basically takes us on a little journey through the minds of marketeers. What the diagram tells us is that very early in the life of an emerging technology we reach a place called the Peak of Inflated Expectations. This is where visibility of the emerging technology is very high but it remains very immature, which basically means it is virtually unusable.

Following the life of any emerging technology we quickly fall of the edge of the Peak of Inflated Expectations and fall into the Trough of Disillusionment. This is where all the early adopters rush out to buy the hot stuff and find out it's incompatible with just about everything else ever created -- including system administrators.

From there it's a long slow haul up the Slope of Enlightenment. I like to think of this as the process of resigning ourselves to the fact that now we've bought the technology and have realised that it's pretty useless for the job we wanted it to do, we have to find a real use for it to justify the expense. That, and the realisation that we will now have to keep buying more, because whatever it is it won't work with anything else.

And finally, once we've been dragged by lock-in contracts up the Slope of Enlightenment (the sales department tick the 'Enlightened' box next to our name), we reach the Plateau of Productivity, which means that we eventually find something that the new technology will be useful for. Like warming up lunch. In the emerging field of blade servers, I suspect we are only just beginning our hike up to the Peak of Inflated Expectations.

There is lots of talk about blade servers; everybody is launching one, but nobody seems to have quite figured out why we need them. Every marketing department has its theory and some, you may be surprised to hear, are still developing them. When I told a marketing chap at one Big Server Maker recently how the marketing chaps at another Big Server Maker plan to differentiate their blade servers from their pizza box servers, he spoiled a very nice napkin writing it down.

So when the sales rep comes along saying how blade servers are great for server consolidation within a company, whereas pizza box servers are better for ISPs and other places where customers want their own physical server, you know who to blame.

The truth is that it's a fine distinction -- possibly too fine. Blade server makers, most of whom also make pizza box servers, tout the manageability of dozens of blades all in one box. But they also sell management products for their pizza box servers, and they often sell these on the fact that they make it seem as though all these servers are in one, giant, thick crust pizza box.

I'll accept that blade servers do make it very easy to fix faults and they do take up less space, and consume less power, than an equivalent number of pizza box servers, but they suffer from one major fault which does not look like being fixed for several years. They are proprietary. You can't take a blade from one maker and stick it in another chassis. HP, bless them, tried to introduce a standard earlier this year but got shot down by the other server makers because it wanted everyone to use a standard that was developed for the telco industry and which say competitors, is unsuitable for corporates.

I suspect that sometime next year you will all join me in the Trough of Blade Server Disillusionment. There is a way out but this time it's not us that need enlightening; it's the manufacturers, and they will only reach enlightenment by following the path that leads via open standards to a place called compatibility.