Arbortext Publishing Engine consists of two major components, the Arbortext PE Request Manager and a collection of Arbortext PE 하위 프로세스. As described earlier, the Arbortext PE Request Manager is a Java servlet; it receives HTTP requests from the servlet container in which it runs and returns a response for each request. Internally, it consists of a number of components that cooperate in processing each request. The Arbortext PE Request Manager is designed to be extensible; users can write custom components that take part in the request-handling process. Such components are called Dynamic Components.

The Arbortext PE Request Manager and its Dynamic Components have little or no knowledge of XML technologies. To import, publish, export, and otherwise manipulate XML documents, the Arbortext PE Request Manager uses separate programs called Arbortext PE 하위 프로세스. Each Arbortext PE 하위 프로세스 runs in a separate process. The Arbortext PE Request Manager maintains a pool of Arbortext PE 하위 프로세스 that are idle, ready to work on its behalf. When, in the course of processing a request, the Arbortext PE Request Manager determines that it needs some document-manipulation function performed, it allocates an Arbortext PE 하위 프로세스 to do so and waits for the Arbortext PE 하위 프로세스 to complete its work.

Like the Arbortext PE Request Manager, an Arbortext PE 하위 프로세스 can be customized. Users can write Arbortext Publishing Engine applications in several languages. These applications run in Arbortext PE 하위 프로세스 just as Dynamic Components run in the Arbortext PE Request Manager.