Treffer: Distributed resource provisioning for containers using machine learning and live migration

Title:
Distributed resource provisioning for containers using machine learning and live migration
Authors:
Gonzalez-Araya, Alejandro
Contributors:
Arzuaga, Emmanuel, College of Engineering, Rodríguez-Martínez, Manuel, Rivera-Gallego, Wilson, Department of Electrical and Computer Engineering, Baigés-Valentín, Iván J.
Publication Year:
2020
Collection:
Digital Institutional Repository @UPR (University of Puerto Rico - DiRe.UPR)
Subject Terms:
Cloud computing, Containers, Live Migration, Performance Monitoring, Data Streaming, Electronic data processing--Distributed processing, Application software, Computer communication systems, Algorithms
Document Type:
Dissertation thesis
File Description:
application/pdf
Language:
English
Relation:
https://hdl.handle.net/20.500.11801/2825
Availability:
https://hdl.handle.net/20.500.11801/2825
Rights:
Attribution-NonCommercial-NoDerivs 3.0 United States ; (c) 2020 Alejandro Efraín González Araya ; http://creativecommons.org/licenses/by-nc-nd/3.0/us/
Accession Number:
edsbas.BC6A9429
Database:
BASE

Weitere Informationen

Cloud computing uses pools of virtual machines to provide shared computing resources. Provisioning and management of these resources are usually done using statistical algorithms to help decide how to better utilize available compute power. Recently, this has been performed mostly by using live migration of virtual machines. Nowadays containers provide the flexibility to handle many software environments and tasks in a lighter-weight virtualization scheme, providing a more agile alternative to virtual machines for certain applications. Application checkpointing coupled with a container manager allows the live migration of a container. In this thesis, we use container live-migration and real-time monitoring to develop a cloud resource provisioning platform that enables an improvement in usage and execution of containers. As a result, the available resources can be tuned and distributed in a more efficient manner, providing a better use of cloud. This research explores the use of container live migration techniques to improve cloud resource provisioning. We present the design and development of our container live-migration, monitoring and provisioning toolset as well as a performance evaluation and characterization. ; La computación en la nube utiliza grupos de máquinas virtuales para proporcionar recursos informáticos compartidos. El aprovisionamiento y la gestión de estos recursos se realizan generalmente mediante algoritmos estadísticos para ayudar a decidir cómo utilizar los recursos de cómputo disponibles de una mejor manera. Recientemente, esto se ha realizado principalmente mediante la migración en vivo de máquinas virtuales. Hoy en día, los contenedores brindan la flexibilidad para manejar muchos entornos de software y tareas en un esquema de virtualización más liviano, proporcionando una alternativa más ágil a las máquinas virtuales para ciertas aplicaciones. El guardar el estado de una aplicación junto con un administrador de contenedores permite la migración en vivo de un contenedor. En este trabajo, ...