That depends on where the document volume comes from.
If the document is bulky because of bulky images – it ok. PD4ML can temporarily unload static resources to cache and read them again at PDF writing phase.
If it is bulky because of a lot of HTML content: it can be problematic. HTML rendering assumes a lot of overhead – even a single standalone whitespace is represented in RAM as a number of Java objects (content itself, HTML attributes, CSS properties, layout info etc). In the case RAM requirements can be too high.
An experimenting with eval version can help a lot to estimate the max doc volume for your particular case.
Also do not forget a possibility to split the the doc to multiple parts (it it is possible for your doc structure), to convert them separately and to finally merge to a single PDF doc from just generated pieces.
