TY - GEN
T1 - Efficient cloud resource management using neuromorphic modeling and prediction for virtual machine resource utilization
AU - Li, Zhe
AU - Ma, Xiaolong
AU - Li, Ji
AU - Qiu, Qinru
AU - Wang, Yanzhi
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/6
Y1 - 2019/6
N2 - With the rapid development of Cloud Computing and Data Centers, Virtual Machine consolidation has become an important issue to achieve economic scale. In order to support such feature, a robust scheme for Virtual Machine resource demands prediction is critical. Previous prediction models such as Auto-Regressive Moving Average model lacks the ability to give an acceptable accuracy for a large prediction window and conventional machine learning based methods suffer from high complexity problem. In this work, a neuromorphic system based on cogent confabulation is built to predict the resource usages from statistics of historical records in comprehensive dimensions. The system exploits the correlations between observations in multiple dimensions and models a probability network to be finely tuned to support the prediction application. The experimental results show the cogent confabulation model based Virtual Machine resource utilization prediction gives a better accuracy than previous work and has an intrinsic advantage in dynamic prediction with a wider prediction window. Using the accurate confabulation-based VM resource prediction, the cloud resource management can improve the energy efficiency (in terms of electricity price) by up to 26.52%.
AB - With the rapid development of Cloud Computing and Data Centers, Virtual Machine consolidation has become an important issue to achieve economic scale. In order to support such feature, a robust scheme for Virtual Machine resource demands prediction is critical. Previous prediction models such as Auto-Regressive Moving Average model lacks the ability to give an acceptable accuracy for a large prediction window and conventional machine learning based methods suffer from high complexity problem. In this work, a neuromorphic system based on cogent confabulation is built to predict the resource usages from statistics of historical records in comprehensive dimensions. The system exploits the correlations between observations in multiple dimensions and models a probability network to be finely tuned to support the prediction application. The experimental results show the cogent confabulation model based Virtual Machine resource utilization prediction gives a better accuracy than previous work and has an intrinsic advantage in dynamic prediction with a wider prediction window. Using the accurate confabulation-based VM resource prediction, the cloud resource management can improve the energy efficiency (in terms of electricity price) by up to 26.52%.
UR - http://www.scopus.com/inward/record.url?scp=85070887614&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85070887614&partnerID=8YFLogxK
U2 - 10.1109/ICESS.2019.8782503
DO - 10.1109/ICESS.2019.8782503
M3 - Conference contribution
AN - SCOPUS:85070887614
T3 - 2019 IEEE International Conference on Embedded Software and Systems, ICESS 2019
BT - 2019 IEEE International Conference on Embedded Software and Systems, ICESS 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2019 IEEE International Conference on Embedded Software and Systems, ICESS 2019
Y2 - 2 June 2019 through 3 June 2019
ER -