Learning to dynamically select cost optimal schedulers in cloud computing environments
File(s)
Accepted version
Author(s)
Tuli, Shreshth
Casale, Giuliano
Jennings, Nick
Type
Conference Paper
Abstract
The operational cost of a cloud computing platform is one
of the most significant Quality of Service (QoS) criteria for
schedulers, crucial to keep up with the growing computa-
tional demands. Several data-driven deep neural network
(DNN)-based schedulers have been proposed in recent years
that outperform alternative approaches by providing scal-
able and effective resource management for dynamic work-
loads. However, state-of-the-art schedulers rely on advanced
DNNs with high computational requirements, implying high
scheduling costs. In non-stationary contexts, the most so-
phisticated schedulers may not always be required, and it
may be sufficient to rely on low-cost schedulers to tem-
porarily save operational costs. In this work, we propose
MetaNet, a surrogate model that predicts the operational
costs and scheduling overheads of a large number of DNN-
based schedulers and chooses one on-the-fly to jointly op-
timize job scheduling and execution costs. This facilitates
improvements in execution costs, energy usage and service
level agreement violations of up to 11%, 43% and 13% com-
pared to the state-of-the-art methods.
of the most significant Quality of Service (QoS) criteria for
schedulers, crucial to keep up with the growing computa-
tional demands. Several data-driven deep neural network
(DNN)-based schedulers have been proposed in recent years
that outperform alternative approaches by providing scal-
able and effective resource management for dynamic work-
loads. However, state-of-the-art schedulers rely on advanced
DNNs with high computational requirements, implying high
scheduling costs. In non-stationary contexts, the most so-
phisticated schedulers may not always be required, and it
may be sufficient to rely on low-cost schedulers to tem-
porarily save operational costs. In this work, we propose
MetaNet, a surrogate model that predicts the operational
costs and scheduling overheads of a large number of DNN-
based schedulers and chooses one on-the-fly to jointly op-
timize job scheduling and execution costs. This facilitates
improvements in execution costs, energy usage and service
level agreement violations of up to 11%, 43% and 13% com-
pared to the state-of-the-art methods.
Date Issued
2023-04-27
Date Acceptance
2022-04-01
Citation
2023, pp.29-31
Publisher
ACM
Start Page
29
End Page
31
Copyright Statement
Copyright is held by author/owner(s). This is the author's version of the work. It is posted here for your personal use. Not for redistribution. The definitive Version of Record was published in https://doi.org/10.1145/3595244.3595255
Identifier
https://dl.acm.org/doi/abs/10.1145/3595244.3595255
Source
ACM SIGMETRICS /IFIP Performance 2022
Publication Status
Published
Start Date
2022-06-06
Finish Date
2022-06-10
Coverage Spatial
Mumbai, India
Date Publish Online
2023-04-27