Non-clairvoyant online scheduling of synchronized jobs on virtual clusters

Khorandi, Sina Mahmoodi; Sharifi, Mohsen

doi:10.1007/s11227-018-2262-4

Cited by 1 publication

(1 citation statement)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Finally, fl competitive algorithms for online nonclairvoyant job scheduling and resource allocation problems in virtualized clusters are studied in [35] and [36]. However, the problems considered, while having some features in common with the one addressed in this paper, are considerably simpler and thus more suitable to be analyzed with the tool of competitive analysis [37].…”

Section: Jobs Schedulingmentioning

confidence: 99%

A Path Relinking Method for the Joint Online Scheduling and Capacity Allocation of DL Training Workloads in GPU as a Service Systems

Filippini

Lattuada

Ciavotta

et al. 2023

IEEE Trans. Serv. Comput.

View full text Add to dashboard Cite

The Deep Learning (DL) paradigm gained remarkable popularity in recent years. DL models are used to tackle increasingly complex problems, making the training process require considerable computational power. The parallel computing capabilities o ered by modern GPUs partially fulfill this need, but the high costs related to GPU as a Service solutions in the cloud call for e cient capacity planning and job scheduling algorithms to reduce operational costs via resource sharing. In this work, we jointly address the online capacity planning and job scheduling problems from the perspective of cloud end-users. We present a Mixed Integer Linear Programming (MILP) formulation, and a path relinking-based method aiming at optimizing operational costs by (i) rightsizing Virtual Machine (VM) capacity at each node, (ii) partitioning the set of GPUs among multiple concurrent jobs on the same VM, and (iii) determining a due-date-aware job schedule. An extensive experimental campaign attests the e ectiveness of the proposed approach in practical scenarios: costs savings up to 97% are attained compared with first-principle methods based on, e.g., Earliest Deadline First, cost reductions up to 20% are obtained with respect to a previously proposed Hierarchical Method and up to 95% against a dynamic programming-based method from the literature. Scalability analyses show that systems with up to 100 nodes and 450 concurrent jobs can be managed in less than 7 seconds. The validation in a prototype cloud environment shows a deviation below 5% between real and predicted costs.

show abstract

Section: Jobs Schedulingmentioning

confidence: 99%