TY - GEN
T1 - Online optimization in the Non-Stationary Cloud
T2 - 53rd Annual Conference on Information Sciences and Systems, CISS 2019
AU - Maghakian, Jessica
AU - Comden, Joshua
AU - Liu, Zhenhua
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/4/16
Y1 - 2019/4/16
N2 - The rapid mainstream adoption of cloud computing and the corresponding spike in the energy usage of big data systems make the efficient management of cloud computing resources a more pressing issue than ever before. To this end, numerous online algorithms such as Receding Horizon Control and Online Balanced Descent have been designed. However it is difficult for cloud service providers to select the best control algorithm dynamically for resource provisioning when confronted with consumer resource demands that are notoriously unpredictable and volatile. Furthermore, it highly possible that it might not be the case for any one algorithm to consistently perform well over the months-long contract period. In this paper, we first exemplify the need to address non-stationarity in cloud computing by showcasing traces from MS Azure. We then develop a novel meta-algorithm that combines change point detection and online optimization. The new algorithm is shown to outperform existing solutions in real-world trace-driven simulations.
AB - The rapid mainstream adoption of cloud computing and the corresponding spike in the energy usage of big data systems make the efficient management of cloud computing resources a more pressing issue than ever before. To this end, numerous online algorithms such as Receding Horizon Control and Online Balanced Descent have been designed. However it is difficult for cloud service providers to select the best control algorithm dynamically for resource provisioning when confronted with consumer resource demands that are notoriously unpredictable and volatile. Furthermore, it highly possible that it might not be the case for any one algorithm to consistently perform well over the months-long contract period. In this paper, we first exemplify the need to address non-stationarity in cloud computing by showcasing traces from MS Azure. We then develop a novel meta-algorithm that combines change point detection and online optimization. The new algorithm is shown to outperform existing solutions in real-world trace-driven simulations.
UR - https://www.scopus.com/pages/publications/85065157790
U2 - 10.1109/CISS.2019.8692890
DO - 10.1109/CISS.2019.8692890
M3 - Conference contribution
AN - SCOPUS:85065157790
T3 - 2019 53rd Annual Conference on Information Sciences and Systems, CISS 2019
BT - 2019 53rd Annual Conference on Information Sciences and Systems, CISS 2019
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 20 March 2019 through 22 March 2019
ER -