One of QlikView's strengths is working with memory. That make it faster. So basically the user loads all data (all document meaning all graphs and all data loaded from source) in his/her computer. If the server reloads, as far as the user is concerned, nothing happens. He/she keeps working with the data they had when they opened the document.
If you are afraid of locks, don't be, they don't happen.
Hope this helps!