Imagine the following scenario:

you have striped 10x 100 GB files sequentially over four harddisks and a stripe size of 1 MB - you created one file after the other
the client is not IO bound: each client can accept data faster than the aggregated maximum performance of all hard disks
each client does sequential file reading issuing 16 MB block read requests (the request are issued in parallel to all concerned disks)

Hint for the next two questions: figure out the IO pattern happening on a single disk. You can assume that concurrent block reads (of 4MB) on a single disk are executed sequentially and not overlapping. Do you see random or sequential IO for the two cases?