Thanks to the people who responded to my original post. 

I have to say I was a little disappointed/puzzled by the lack of responses. I 
am not sure why this was but I can only assume there is either little 
interest in transforms over the few 100MB size or little interest in 
discussing the problems this presents. 

It is not often I get the chance to discuss stuff we are doing on the 
processor for obvious commercial reasons. I thought this subject might be 
suitable for a wider audience as anecdotal evidence suggests that large XML 
file usage is growing and and along with that comes some expectation that 
tools can cope. 

Regards,
Kev