Judy Guglielmin added a comment - 16/Aug/17 8:31 AM The Document object is the root object of a massive object tree.  It would be difficult to make the object serializable especially given how closely bound the object is to a large byte stream representing a PDF document.   The serialized object sized would likely be extremely large and wouldn’t be a good candidate for session replication. From a viewer point of view there is a small amount of data that needs to be capture to show a page in a new session: file or byte stream location, page number and any view specific info like zoom or rotation.  It is recommended that the user build a very small state object that would be serializable.  The document class is very cheap to recreate and the random access nature of the byte stream can make grabbing page data very efficient even if done from scratch on  different server node.