History: Why did the Eclipse Foundation choose OSGi?

Perhaps the main advantage of OSGi is the ability to have separate, separate packages. This allows a single node to hide its dependencies from the global classloader.

Many times, when building a large system, one component needs a particular library and another uses a different but incompatible version of the same library. In this case, choosing one version of the library over the other interrupts the other.

Separate classloaders for JARs and WARs were an early attempt and allowed some modulation and dependency management for separate and isolated components. If you look at older versions of WebLogic, etc., you will notice that they repackage commonly used libraries (like Xerces) in the weblog. * To avoid clashing with Xerces that your application might want to include.

With OSGI, each bundle can only expose one packet containing an interface to the kernel. Each package can then use its own libraries without worrying about the classpath problem described above.