In our modeling, we decompose a repair operation into threesteps carried out in a sequential manner: (i) read (i.e., readingchunks from the underlying local storage), (ii) transmission(i.e., transmitting the chunks over the network), and (iii) write(i.e., writing the repaired chunks into new existing nodes(for scattered repair) or hot-standby nodes (for hot-standbyrepair)). Let bd and bn be the disk and network bandwidths,respectively, and c be the chunk size. We calculate the readtime, transmission time, and write time for each repaired chunk.