A point largely overlooked by other reports on the recent pairing of IBM and Gluecode is the fact that Gluecode's product is more than just another J2EE application server. Gluecode provides an open-source entry point for the entire IBM Websphere product suite. The Apache-based Gluecode application server also includes Apache Derby database (formerly IBM's Java-based Cloudscape, which was recently donated to Apache), Apache Pluto portal framework (see: WebSphere Portal) as well as a message queuing protocol similar to IBM’s Websphere MQ. In my opinion (which counts for little, save in this blog), the fact that Gluecode’s offering has direct parallel’s with IBM’s WebSphere platform provides an insight into Big Blue’s decision to forego purchasing the better-known JBoss group in favor of the lesser-known Gluecode.
The Gluecode-specific portions of the app server are not open source per se as much as they are shared source; if you choose to subscribe to Gluecode's service package, the source for the JDBC drivers et al is included. That Gluecode already has a subscription-based service model dovetails nicely with IBM's current revenue strategy which is based more upon services than deliverable products.
Offering an open source application server also serves to complete the open/closed source product tiering already in place with Eclipse/WSAD and Derby/DB2. Get developers/IT shops hooked on the freebie, then - when their needs scale to the enterprise level - migrate them to the pay-for products and services.
On the surface, the Gluecode purchase appears to be just another salvo in the M&A volley currently underway in the tech sector. However, paired with other recent moves over at Big Blue, this tidbit brings with it possibilities which portend a shift in the products that IBM currently offers.
With the Gluecode buyout, IBM is also acquiring the development staff - many of which are top contributors to Gluecode's underlying Apache Geronimo app server. This link gives IBM leverage on defining Geronimo's functionality, meaning subsequent releases of Geronimo will have more in common with WebSphere, making a client migration from Geronimo or Gluecode to WebSphere less painful than, say, a migration from Geronimo to BEA's WebLogic.
Being open source also gives IBM a safe haven to test extensions to their WebSphere app server without polluting that product line. For instance, IBM could push the envelope by extending support within the Gluecode product to applications built on PHP or Python (via Jython). Should this strategy gain traction in the marketplace, it would be academic to replicate this functionality in WebSphere.
Two Birds, One Open Source Project
One final point on the subject of Java, open source and IBM, regarding the press release by the Apache group about Project Harmony, whose goal it is to create a completely open-source JVM.
While I agree completely with fellow ADT blogger Matt Stephens that open source Java for the mere sake of being open source is a waste of time, from IBM’s perspective an open source JVM resolves two issues.
First, IBM would no longer have to pay their competitor Sun royalties when using the JVM as the underpinnings for their enterprise product line. Second, certain pundits in the open source community refuse to endorse or include a JVM as a component in Linux distributions; if there were an open source implementation of the Java runtime environment, IBM would have the ideal open source technology stack competitor to Microsoft’s .NET-on-Windows: OS-Java-on-Linux. Imagine every Linux OS from RedHet, Novell, and Mandriva including IBM’s Gluecode running atop Harmony. Java is IBM’s preferred delivery vehicle for their enterprise software offerings. If the complete stack were open source, the natural migration path for enterprise-scale apps would be to run atop IBM’s WebSphere platform.
Regardless of the logic behind abandoning a project to create an open source JVM, IBM will ensure that the project succeeds if only to divorce itself completely from Sun, and to create a completely open source stack competitor to Microsoft.