Devil's Advocate: Revenge of the mainframe

From mainframes, to client-server PC revolution, back to mainframes again? It's one of the big ideas plenty of people are talking about. This week Martin Brampton casts a critical eye over the claims...

How dare they? We spent the last 20 years looking down our noses at 'legacy' systems as we all rushed into the PC revolution. And now those people at IBM are claiming we can do it all better by scrapping our dinky new boxes and returning to that old dinosaur, the mainframe.

They say the constant application of new techniques to the old fashioned mainframe has made it more efficient than ever. Price gains mean IBM is claiming a 25 per cent cost reduction for customers who get rid of those pesky computers from HP and Sun.

Did you think Linux was a grass roots revolution belonging to the PC age? Well, those fuddy duddy mainframe chaps at IBM have taken it up very seriously. They like Linux and have adapted it to run many times on one big mainframe. Bringing order to all those little computers has driven the hosting centre paradigm - squash them as much as possible and rack them up to make them tidy.

Tidiness is hard to achieve, though, with lots of separate units. The front panels look appealing, all neatly rack mounted, but there is usually a terrible mess of wires behind the back panel. Power distribution is a nightmare and each unit needs at least one network connection. Heat dissipation is a major problem. Blade servers improve matters to some extent but still involve a good deal of complexity.

Why implement things in hardware that are simply not required? The virtual machine was invented at the dawn of the computer age and first implemented on those old big iron systems. It makes sense to have lots of virtual machines all running inside a single large computer. Lights out operation? The mainframe people invented it.

So why ever did we start spreading lots of little computers around our organisations? Sadly, the virtues of the mainframe had become hidden by the sclerotic attitudes of all too many IT departments, backed up by a stodgy IBM managed from an ivory tower in Armonk. There was a lengthening queue for new applications and nobody seemed to have any idea how to make computing power available to the masses.

Along came the PC revolutionaries with the thought that the whole IT business could be turned upside down. And it has been. There has been unprecedented innovation in the user interface. Inventions such as the spreadsheet and the personal database have given computer power to the individual. IT departments have changed dramatically or been ousted in favour of third party providers.

Yet there has always been a sneaking suspicion we would pay a price for all this granularity. Most people do not want to become their own IT experts. Or, even if they do, the organisation would rather they got on with their own jobs. Management is painfully aware that server-based computing is far easier to handle than distributed systems.

It looks as though we could be at a turning point. We will not see it while it is happening. Turning points are only ever visible in retrospect. The PC sector is in danger of stultification, with a few - too few - powerful vendors who have a vested interest in the status quo. Much of the real innovation has stopped, to be replaced by mere feature creep.

Many people say the PC is still the only way to deliver the e-world to which we are all supposedly aspiring. But it could be that we will wake up one day soon to find that the computing in our lives is based around ubiquitous access to powerful servers that keep our valuable information safe. They are also far more attuned to the collaboration that is so vital to organisations. And how will we gain access? Through a variety of devices in public places or through pocket gadgets that really can go anywhere. Who needs a PC then?

What do you think? Is the mainframe about to mount a comeback?

** Martin Brampton is a director and founder of Black Sheep Research (www.black-sheep-research.co.uk ), an independent consultancy providing research, writing and speaking services on a wide range of business and technology subjects. Martin was previously a director at Bloor Research, and has worked with IT as a user and analyst for over 20 years. He is a frequent contributor to silicon.com's Behind the Headlines TV programme and can be contacted at silicon@black-sheep-research.co.uk.