It must seem like heaven to a CIO: A software product that only costs you something when you use it. A product that eliminates software upgrade angst and support requests. The neon sign over the pearly gates reads "Software as a service", better known as SaaS.

A recent McKinsey report quoted research estimating that 10 percent of the market for enterprise software will migrate to a pure SaaS model by 2009. McKinsey believes this delivery method is attractive to CIOs because it offers them flexibility (to switch vendors) and an easier upgrade path.

I'm not so sure.

I know SaaS is something that many local CIOs are interested in and I frequently get asked how I see it progressing. I'm conscious that salesforce.com is one of the great success stories of the IT industry in recent years. Yet I still wrestle with how practical the concept will be in reality. Will it really be possible for suppliers to price, support and deliver everyday software via this methodology?

Let's start with the basics. What about Australia's notoriously backward Internet facilities. Without a fast, reliable Internet connection everything else in SaaS is academic. Whoever wins the coming election I have doubts that a swift Internet service will be generally available across even the capital cities in Australia by the end of 2009.

Then there's how you use it. How elegant will it really be for spreadsheets, word processing and presentation software? Often you find you want to create these document types on occasions when you have time to kill. This might include waiting for a plane or in a coffee shop between meetings when you are travelling. Typically in such circumstances you might not be able to get a simple, cost-effective or reliable Internet service. Yet without it you will have no applications, so your laptop will be useless.

Next we need to look at the charging mechanism. SaaS is a bit like a toll road. However, with a toll road the approach is relatively simple. You can pay before you use it or use a tag to debit your credit card. It is reasonably straightforward for an individual to keep track of their transactions. How well, though, would that translate in a corporate world with multiple individuals? I've got a feeling that it could offer a CIO the ultimate payback opportunity for any CFO who has been making their lives a misery. Imagine the challenge some poor soul in accounts would have trying to reconcile all those transactions in a SaaS model for a desktop application in a large multinational enterprise. Many CIOs have difficulty keeping track of the equipment covered by a lease. I suspect the volume of such transactions in a SaaS model would be beyond the capabilities of even the most sophisticated corporate accounting system.

That said, I'm not against SaaS. The model of paying for something through use rather than a licence has appeal. I am conscious that many CIOs are extremely frustrated by the cost and challenges of the never-ending software upgrade cycle. SaaS should put an end to these hassles.

However, the challenge will be to make the concept work in reality. How can the charges be equitable and transparent? What are the infrastructure requirements to support it? Which applications are the best candidates for it? The early iterations of SaaS will shed light on these questions. Over time these experiences will enable the SaaS model to be refined in to an effective working arrangement. When that happens I believe that quite a few CIOs will think they are in heaven.

Peter Hind is a freelance consultant and commentator with nearly 25 years experience in the IT industry. He is co-author of The IT Manager's Survival Guide and ran the InTEP IS executive gatherings in Australia for over 10 years