I am not sure how a file this big is being generated, but I can certainly tell you that 100 MB is way too much for a typical XML document.
The bigger the file is, the parsing process will become more resource intensive. You could optimize the process by using small foot print SAX parsers, but the long term solution should be to generate and consume smaller files. If the (large) size of the file is driven by the requirements, then
you should consider some kind of custom parsing or using plain text-comparison tools to deal with such documents.
Hope that helps,