Open source adoption within companies has occurred in a series of waves, each moving free software successively closer to the heart of the enterprise. First, there was Apache, whose rise is documented vividly in the Netcraft server survey. After the Web server, open source began to find favour for file serving, typically using GNU/Linux and Samba. GNU/Linux was also used to run proprietary databases like Oracle and DB2, but more recently open source databases like MySQL have proved increasingly popular with companies.

Although the open source desktop is clearly reaching a tipping point in terms of broader adoption - not least thanks to the maturity of offerings like OpenOffice and the Firefox browser - it is arguable that the next bastion of proprietary software to fall will be that of the application server.

The program most likely to achieve this is JBoss. The company behind it claims it is the leading product in the Java application server market, and the fastest-growing in terms of market share in the overall application server sector - it has already been downloaded over 5 million times. There is more on the JBoss "vision", a Wiki, and a detailed roadmap for the product.

Much of the success - and increasing importance - of JBoss can be attributed to the ancillary projects that have sprung up alongside the main application server, and the fact that JBoss the company employs many of the top open source hackers working in these areas.

For example, JBoss-IDE provides debugging and monitoring of JBoss servers using the Eclipse framework - another open source project that is just starting to break through into the mainstream and challenge proprietary offerings. JBoss has a content management system called Nukes, which JBoss's co-founder Marc Fleury believes has better scalability than the free PostNuke system, based around PHP (yet another open source success story). There is even work going on to turn JBoss into a fully-fledged email server.

An indication of JBoss's rising status are the alliances it has been forging. It is teaming up with MySQL, and has recently received an investment from Intel. The Intel deal is more than a matter of money: it brings considerable kudos, and some interesting historical parallels. For at the end of 1998 Intel made another investment, this time in a small open source company called Red Hat. A year later, Red Hat's IPO valued the company at $3 billion.

One of the reasons that first Apache, then GNU/Linux and now the open source desktop have all become serious options in the corporate sphere is because, one after another, IBM has backed them. It has been able to derive healthy revenues while adopting freely-available applications by selling products further up the software chain, as well as offering related services. The cornerstone of this approach is IBM's WebSphere software platform.

But as the history of open source shows, once commoditisation has begun in the lower layers, there is a constant pressure to move up the stack. It is rather ironic that having begun this process in June 1998 by replacing its proprietary Internet Connection Server with Apache to support its WebSphere offering and enhance its services business, IBM may find itself forced to replace the WebSphere application server too - with JBoss being the obvious candidate.

Glyn Moody welcomes your comments.