file size appears to me as a completely random metric to measure your code against. With the overhead associated with establishing a http connection, the actual

file size appears to me as a completely random metric to measure your code against. With the overhead associated with establishing a http connection, the actual byte-size is most likely of little importance. In fact, if this goes over TCP/IP (Which HTTP usually does), isn't there a frame size of some 1500 bytes?