In the context of team Cornet’s seminars, Francesco De Pellegrini (LIA) will present his research work on Coflow Fair Scheduling via Dynamic Progress, on April 1, 2022, at 11:35 in the meeting room.
Abstract : The average coflow completion time (CCT) is the standard performance metric in coflow scheduling. However, standard CCT minimization may introduce unfairness between the data transfer phase of different computing jobs. Static progress guarantees have been introduced in the literature to mitigate this fairness issue, but the trade-off between fairness and efficiency of data transfer seems hard to control. In this paper we introduce a new fairness framework for coflow scheduling based on the concept of slowdown to measure the performance degradation experienced by a coflow compared to isolation. This framework provides more flexible means to control the progress of coflows while minimizing the average CCT. We design an algorithmic solution chosen in the class of the sigma-order schedulers to solve the fair coflow scheduling problem in polynomial time. The algorithm is proved to be a 4-approximation w.r.t. an optimal scheduler. Our numerical results validate the proposed scheme and demonstrate that this approach can trade off average CCT for per-coflow slowdown.