Information Age: News, analysis & insight for IT & business leaders

After a seemingly never-ending round of negotiations with prospective suppliers, the technology decision-maker decides to test a mobile application. The trial goes well, but persuading the board to roll it out takes an age. Frustration takes hold. Eventually, the board gives the scheme the go-ahead, which heralds the moment to integrate the application into existing systems. And then the real problems start.

Arguably the biggest challenge for any leader of a mobile working project is finding the best way to blend new mobile applications with older systems without breaking the bank.

Optimising the efficiency of workers who spend much of their time out on the road means tying them much more closely into business information systems designed mostly for use in the back office. "Often, businesses have two distinct environments," says Jaye Isherwood, product manager for mobile technologies at Cap Gemini Ernst &Young. "They have people out in the field dealing with clients on a one-to-one relationship, which is paper-driven and hugely inefficient. In the back office, you have all the computer processes that drive efficiency, but it is highly impersonal. We now have the technology to bring this together."

She cautions, however, that businesses have to choose between a plethora of phone operating systems, middleware environments and integration models if they are to exploit mobile access to their back-office systems to the full.

Perhaps surprisingly, given the fierce commercial rivalry between smartphone and PDA platform vendors Microsoft, PalmSource and Symbian, the choice of the platform itself is not the most critical decision for enterprises. Most applications either support multiple platforms directly, or they can provide data in an open format that can be read using a web or even a WAP browser.

The way the device integrates with the business application itself is the more strategically important choice, mobility experts argue. Here, companies have four main options: use mobile connectors provided by application vendors; build some form of mobile middleware in-house; opt for a commercial mobile middleware technology; or provide access through existing infrastructure, such as remote access from a web browser interface or a corporate portal.

Web-based front ends are perhaps the easiest to deploy, especially for so-called 'out of the box' access to software tools such as email, contacts databases and calendars. Microsoft's Exchange Server 2003, in particular, has advanced support for remote clients. But web-based access is bandwidth-intensive, offers no offline access - for when users on cellular connections cannot find a signal - and works best with large screen clients such as laptops, rather than PDAs or smartphones.

Enterprises creating their own mobile middleware can build a highly tailored environment, but they have to choose between developing a bespoke middleware layer, typically using Java 2 platform, Mobile Edition (J2ME) or Microsoft's .Net Compact Framework. This offers the most customisation and, depending on the expertise of a company's in-house development team, it can be a cost-effective and quick way to 'mobilise' a handful of key business applications.

An alternative that is gaining ground among enterprises is to buy from one of the growing number of mobile middleware vendors. Oracle, Sybase-subsidiary iAnywhere and Aether Systems are regarded as 'market leaders' by analysts at Forrester Research, although the sector has more than a dozen key suppliers.

Top-end middleware companies provide a range of important features including managed synchronisation between a mobile device and the host application, as well as support for offline working.

The key advantage, though, is that these products remove the need to write or implement either device or application-specific connections.

If companies change their mobile devices, or want to add mobile access to new back-end applications, they should just need to add the connectors from the middleware to the device or the program. For large businesses with multiple applications, this offers the prospect of significantly reduced development and testing times.

But that has to be set against the upfront cost of acquiring and deploying middleware, and the fact that the direct support for mobility in key enterprise applications has improved markedly. "One or two years ago you would have had to put in third-party middleware for an application to work with a laptop or PDA, especially if you didn't have Ether-net bandwidth," says Peter Andersson, European mobility business development manager at Hewlett-Packard.

He accepts that competitors including SAP, Siebel and Microsoft have made progress in improving the way their applications work with mobile clients. "If you only have one or two applications, building a specific connector is the way forward. But if you have the flexibility to put in an architectured layer using J2ME or .Net, then that is the way to go."

As yet, though, relatively few businesses appear prepared to make the necessary investment.

Analysts at Ovum recently surveyed enterprises' work in the mobile arena, and found that companies are favouring either a piecemeal approach or the web browser route over third-party or bespoke middleware.

"You would expect enterprises to take the integrated approach and create a uniform mobile integration layer," says Ovum analyst Jessica Figueras. "But if you can only make a business case for a very specific application, you are not going to spend money on an architecturally beautiful solution, because you won't get the best out of it." For a large number of companies, mobile deployments so far consist mostly of giving staff mobile data cards. "They are using it for exactly what they already use their laptops for," she says. The difference is that they are no longer constrained by the need to find a phone or Ethernet socket.

Software suppliers' big mistake has been to label mobility as one clearly defined sector, she says. "In reality the picture is fragmented. Companies will use mobility for specific things in specific cases, such as a pharmaceutical company using mobile devices to gather drugs test results - something that is expensive to do on paper. But that is a completely different business case to putting WiFi in the office."

Both are valid projects but both require IT managers, or line-of-business heads, to make a totally different investment case. For enterprises, the way forward seems to be to test the concept of mobile access to business applications by deploying 'point' solutions, where the business case is clear and where the technology process is not so involved or complicated that it is hard to measure the results. But if a company plans a longer-term commitment to mobile working, the CIO will be faced with an even bigger integration challenge. Ultimately, that may even lead to a complete overhaul of the way IT systems and business processes are managed.