Interested to know how you handle accessing the data, after you've stored it. Fetching the entire XML = far more IO, memory and netowrk usage than might be required, if only a small section of data, like who completed the document, and when, needs to be extracted. If you are using the XML data, how are you seeing performance? It's been pretty poor in our examples of high-volumes, especially if you need to use nodes and elements as part of your critieria, so I'm interested to know who you've handled that - I'm sure I can learn from it...