A method for shortening the length of time gaps between data units in a data switching network; the method comprising: arranging a group of K data units received consecutively at an input of a first device in K time units, where K is a positive integer; sending the arranged K data units to the inputs of M devices, where M is a positive integer; transferring a group of Hj data units of the arranged K data units by a device j of the M devices from its inputs to its outputs within a time unit Tx in the K unit time period, where Hj is a positive integer and the sum of Hj for j=1, . . . M is equal to K, where HjK; arranging the data units sent to the devices of the M devices transferring at time unit Tx such that no two inputs of the devices of the M devices transferring at time unit Tx receive the same data unit at any moment; and repeatedly combining and arranging the K data units from the outputs of the M devices by a second device such that the K data units appear at the output of the second device consecutively.