AF: EAGER: Scheduling with Resource Contraints

Columbia University, New York NY

Investigators

Abstract

Even as the availability of computing power, network bandwidth and other resources increases at a rapid rate, users and applications increase their demand for computation and their use of resources at roughly the same pace.  Thus, no matter how much progress is made  on hardware efficiency, we will always need efficient algorithms to manage these resources. The philosophy of managing additional resources in an intelligent manner goes beyond computer systems, and is relevant to many other  scientific and industrial areas. In various real-life systems, processing times may be controllable by allocating resources, such as additional money, overtime, energy, fuel, catalysts, subcontracting, or additional manpower, to the job operations. In such systems, job scheduling and resource allocation decisions should be coordinated carefully to achieve the most efficient system performance. Applications arise in many industrial areas. The PI plans to study several algorithmic problems that arise in scheduling with additional resource constraints.   While this field has received much attention over the past ten years and beyond, the PI will focus on two areas that have received very little attention in the computer science community -- scheduling when the benefit of the schedule and the cost of energy or other resources are monetized, and scheduling in models that go beyond the typically studied computer systems setting.  Because these have received very little attention, this work is somewhat speculative.  If successful, this work could have a high impact, with applications into many new areas. The PI will design simple, low overhead algorithms.