Please use this identifier to cite or link to this item: http://hdl.handle.net/10362/61678
Title: Replication and Caching Systems for the support of VMs stored in File Systems with Snapshots
Author: Silva, Luís Miguel Teixeira da
Advisor: Lopes, Paulo
Medeiros, Pedro
Keywords: VDI
Btrfs File System
Snapshots
Replication Middleware
Cache Servers
Defense Date: Dec-2018
Abstract: Recently, in a relatively short timeframe, there were fundamental changes in the way computing power is used. Virtualisation technology has changed both the model of a data centre’s infrastructure and the way physical computers are now managed. This shift is a consequence of today’s fast deployment rate of Virtual Machines (VM) in a high consolidation environment with minimal need for human management. New approaches to virtualisation techniques are being developed at a surprisingly fast rate, leading to a new exciting and vibrating ecosystem of platforms and services. We see the big industry players tackling problems such as Desktop Virtualisation with moderate success, but completely ignoring the computation power already present in their clients’ infrastructures and, instead, opting for a costly solution based on powerful new machines. There’s still room for improvement in Virtual Desktop Infrastructure (VDI) and development of new architectures that take advantage of the computation power available at the user’s desk, with a minimum effort on the management side; Infrastructure for Client-Based Desktops (iCBD) is one of these projects. This thesis focuses on the development of mechanisms for the replication and caching of VM images stored in a local filesystem, albeit one with the ability to perform snapshots. In this work, there are some challenges to address: the proposed architecture must be entirely distributed and completely integrated with the already existing client-based VDI platform; and it must be able to efficiently cope with very large, read-only files, (some of them snapshots) and handle their multiple versions. This work will also explore the challenges and advantages of deploying such a system in a high throughput network, with both high availability and scalability while efficiently supporting a large number of users (and their workstations).
URI: http://hdl.handle.net/10362/61678
Designation: Master of Science in Computer Science and Informatics Engineering
Appears in Collections:FCT: DI - Dissertações de Mestrado

Files in This Item:
File Description SizeFormat 
Silva_2018.pdf6,45 MBAdobe PDFView/Open


FacebookTwitterDeliciousLinkedInDiggGoogle BookmarksMySpace
Formato BibTex MendeleyEndnote 

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.