Method for optimizing joint operation of task scheduling and cooling control of data center

The invention discloses a data center task scheduling and cooling control combined operation optimization method, and the method specifically comprises the steps: carrying out the real-time monitoring and feedback of the operation state, including task information and server load information, of an...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: YANG CENYUAN, YAN HUAJIANG, SHI KUN, XIE DANYU, QIU JUN, DING ZHAOHAO, WANG JIAYING, LU CHUNGUANG, LI XINRAN, CHEN SONGSONG, ZHOU BIN, LI LEI, XIAO TAO
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention discloses a data center task scheduling and cooling control combined operation optimization method, and the method specifically comprises the steps: carrying out the real-time monitoring and feedback of the operation state, including task information and server load information, of an IT system of a data center; a temperature sensor regularly samples three temperature data of each server rack; a Markov decision process is adopted to model a joint operation process of task scheduling and cooling control of an IT system, and a task scheduler is trained based on a deep Q network algorithm with the purpose of minimizing the use efficiency and energy consumption cost of a data center power supply. According to the method, joint optimization of task scheduling and cooling control of the data center in a spot market environment is realized, the energy efficiency of the data center is improved without depending on an accurate and mathematically solvable system model, a depth Q network algorithm based on