>My import was about 70,000 documents totaling about 50GB. The process worked with documents up to around 300MB, although I wasn't doing anything like encoding/decoding. ISTR one file that was about 325 - 350MB crapping out, so in the import I limited file size to 300MB and logged any files encountered that were larger than that (typically video recordings of meetings).

So you were using FILETOSTR() to push a document into SQL server (I take it). Not sure I would need to do the encoding/decoding if I was reproducing this document system and cutting out the web engine in between. How does the document get stored in SQL server i.e. what data type?

Albert