It seems an important new buzzphrase has been christened by analyst Michael Coté of RedMonk, the open-source research firm. The Stackless Stack (a term first expressed in blog form by Coté's colleague, James Governor) is shorthand for what those of us who've been watching "big IT" fade have been wanting to find a term for, to no avail. It refers to an emerging trend toward enterprise infrastructure (and web-app infrastructure, particularly) that's lighter, more modular, and more services-oriented.

In some ways, "stackless stack" is a bit misleading, because no one would seriously suggest that people are going to move off of (for example) the LAMP stack any time soon. But plenty of people are moving off of the heavyweight WebSphere + DB2 + AIX type of stack (apologies to IBM) in favor of Tomcat + Derby, say, or a Spring-based approach. The Spring Framework, in fact, gets straight to the heart of the matter, because "stackless stack" is really about running what you need when you need it, in the lightest possible runtime environment (preferably service-oriented), with the fewest number of moving parts. Spring exemplifies that.

If you're still having trouble grasping stacklessness, consider the trend (over the past year or two) by Web CMS vendors toward embedding, bundling, or otherwise targeting Tomcat as a runtime framework instead of, say, JBoss. If all you need is a servlet engine and web server, why bring along EJB runtimes, a JMX framework, JAAS/JACC, and all the other scaffolding that comes with a full-blown J2EE appserver?

"Stackless stack" encapsulates many notions, but if you're an IT expert, you can think of it in terms of a shift of focus from component-based development to (as James Governor says) component-based production. If you're a developer, you shouldn't need to load 500 Java classes just to print "Hello World" to the console. Likewise, if you're running a Web CMS, you shouldn't need to wire together millions of dollars' worth of systems to light up a web browser. Even in "Web 2.0" world where you might want to push more interactive services to the edge of your network, you still don't want more services than necessary. This is, after all, 2008, not 1975.