Server Scheduling Benchmark Instances

Published: 08-03-2018| Version 1 | DOI: 10.17632/ph95d337dj.1
Rodrigo A. Carrasco


Data for job scheduling in a server. The data is divided into 5 ZIP files. Each zip file contains a collection of text files, where each file contains the information of all jobs arriving on a day to the server. The text file structure is as follows: <instance ID> p <list of CPU cycles required> w <list of priority weights> r <list of release dates> pr <list of precedence constraints in pars of the form [parent, child]> The information in p, w, and r, follow the format of dictionaries in Python (job ID: information), whereas pr has the format of a Python list. The "results.xls" file, has 10 Excel sheets, with the best lower bound, and upper bound (i.e. schedule value) known for that instance. There are two sheets for each set of instances, one with the results considering release dates (ends in "wr") and one with the results without considering release dates (ends in "nr"). In all cases, the schedule was evaluated as total completion time, plus total energy consumption as described in "Resource Cost Aware Scheduling" by Carrasco, Iyengar, and Stein (