Information Technology employs various techniques that are dealing with the problem of Energy Efficiency in data centers. Among them are various implementations of artificial intelligence and machine learning techniques, as they can provide important contribution in energy and performance management of a data center, because they can deal better with the challenge of adapting to the complex and dynamic heterogeneous environment that modern data centers represent. In this thesis I have modified and expanded an existing goal-oriented model for energy efficient adaptive applications in data centers. Improvement of Energy Efficiency, and satisfaction of Quality of Service metrics are considered in the goal-oriented approach. The main benefit of the proposed modification, explained in this thesis, is general applicability of the goal-oriented approach for any set of metrics, in any data center. The relations among the metrics of interest, and their dependencies, are explored in more depths, taking in consideration the deployment and migration of virtual machines from one server to another.
La tecnologia informatica impiega varie tecniche che si occupano del problema dell'efficienza energetica nei data center. Tra questi ci sono varie implementazioni di intelligenza artificiale e tecniche di apprendimento automatico, in quanto possono fornire un contributo importante nella gestione energetica e delle prestazioni di un data center, perché possono affrontare meglio la sfida di adattarsi all'ambiente eterogeneo complesso e dinamico che i moderni data center rappresentare. In questa tesi ho modificato e ampliato un modello orientato agli obiettivi esistente per applicazioni adattive efficienti dal punto di vista energetico nei data center. Il miglioramento dell'efficienza energetica e la soddisfazione delle metriche sulla qualità del servizio sono considerate nell'approccio orientato all'obiettivo. Il principale vantaggio della modifica proposta, spiegato in questa tesi, è l'applicabilità generale dell'approccio orientato all'obiettivo per qualsiasi insieme di metriche, in qualsiasi data center. Le relazioni tra le metriche di interesse e le loro dipendenze vengono esplorate in modo più approfondito, prendendo in considerazione l'implementazione e la migrazione di VM da un server a un altro.
Discovering relations between metrics in a real data center using Bayesian networks
STOJKOSKA, DUSHICA
2016/2017
Abstract
Information Technology employs various techniques that are dealing with the problem of Energy Efficiency in data centers. Among them are various implementations of artificial intelligence and machine learning techniques, as they can provide important contribution in energy and performance management of a data center, because they can deal better with the challenge of adapting to the complex and dynamic heterogeneous environment that modern data centers represent. In this thesis I have modified and expanded an existing goal-oriented model for energy efficient adaptive applications in data centers. Improvement of Energy Efficiency, and satisfaction of Quality of Service metrics are considered in the goal-oriented approach. The main benefit of the proposed modification, explained in this thesis, is general applicability of the goal-oriented approach for any set of metrics, in any data center. The relations among the metrics of interest, and their dependencies, are explored in more depths, taking in consideration the deployment and migration of virtual machines from one server to another.File | Dimensione | Formato | |
---|---|---|---|
2018_04_Stojkoska.pdf
non accessibile
Descrizione: Thesis text
Dimensione
1.67 MB
Formato
Adobe PDF
|
1.67 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/140126