By default, when Trading Networks receives a document, it keeps the document content in memory during processing. If a document is large, Trading Networks might encounter memory constraint problems. These problems can occur when Trading Networks tries to execute an XQL query against an XML document during document recognition or tries to perform pre-processing or processing actions.

You can set a property to specify when a document should be considered large. When Trading Networks receives a document that fits your specification, it processes the document as a large document. Trading Networks does not try to read content for large documents into memory; rather, it writes the document content to hard disk drive space (called tspace) and stores only a reference to the document content in memory. When Trading Networks needs to access document content during processing, if the content is in tspace, Trading Networks either retrieves a certain number of bytes of the document content or uses a Java InputStream to read the document content. The document content remains on disk until the service that processes the document, and all services invoked from that service, complete, and the time to live period you specify expires.

The Trading Networks features listed in the table are not available for large documents.