Using a typical enterprise architecture such as the one shown in the previous diagram, we can consider 10.000 document retrievals/hour in average. The downloading times will depend on the size of the documents and the customer's network capabilities. If needed, the retrieving capabilities can be escalated.
In the following image you can see number of request per hour in a typical enterprise production architecture (see first diagram).
You can see that from 12 o’clock (to 1 o’clock) the system received more than 26.000 requests. The downloaded data is 1,4GB/hour in this example, what is equivalent to downloading, approximately, more than 25.000 documents of 50Kb / hour.
As another example, you can considered:
In a production environment in which the system hosts 4TB of information, making the call from Spain, we can retrieve a document metadata in 71 ms (in this case we do not consider content downloading time, only metadata).