Abstract: Size-based scheduling policies, such as shortest-remaining-processing-time (SRPT) and shortest-job-first (SJF), have been extensively studied for decades. Yet, almost all analytical results in the literature remain exclusive to single-server queues with infinitely patient jobs and under exact job-size information, which are assumptions that clearly do not hold in service systems such as call centers or healthcare facilities. In this work, we provide the first analytical results on the performance of size-based scheduling in multiserver queues with abandonment and inexact job-size information.

Based on joint work with Jing Dong from Columbia Business School