Open AccessAn Attention-based Recurrent Neural Network for Resource Usage Prediction in Cloud Data Center
Author(s) -
Huidan Xi,
Yan Cheng,
Huixi Li,
Yinhao Xiao
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2006/1/012007
Subject(s) - computer science , cloud computing , data center , resource (disambiguation) , workload , trace (psycholinguistics) , encoder , artificial neural network , mechanism (biology) , resource allocation , data mining , real time computing , distributed computing , artificial intelligence , computer network , operating system , philosophy , linguistics , epistemology
To manage the computing resources efficiency is crucial for the Cloud data centers (CDCs). By predicting the resource usages of certain virtual machine (VM), the workload could be balanced before the resource overusing occurs. In this paper, we propse an attention-based time series prediction model, which contains an encoder of input attention mechanism and a decoder of a temporal attention mechanism, to optimize the efficiency of could data center. Experiments on Alibaba CDC VM trace dataset demonstrates that our proposed methods can outperforms the classic LSTM method, especially when the resource usage data is lack of relationship.