Please use this identifier to cite or link to this item:
Title: Hera Object Storage : a seamless, automated multi-tiering solution on top of OpenStack Swift
Authors : Höppli, Remo
Bohnert, Thomas Michael
Militano, Leonardo
Conference details: 2018 IEEE 8th International Symposium on Cloud and Service Computing (SC2), Paris, France, 18-21 November 2018
Publisher / Ed. Institution : IEEE
Issue Date: Nov-2018
License (according to publishing contract) : Licence according to publishing contract
Type of review: Not specified
Language : English
Subjects : Cloud Computing; Object Storage; OpenStack; Multi-tiering
Subject (DDC) : 004: Computer science
Abstract: Over the last couple of decades, the demand for storage in the Cloud has grown exponentially. Distributed Cloud storage and object storage for the increasing share of unstructured data, are in high focus in both academic and industrial research activities. At the same time, efficient storage and the corresponding costs are often contrasting parameters raising a trade-off problem for any proposed solution. To this aim, classifying the data in terms of access probability became a hot topic. This paper introduces Hera Object Storage, a storage system built on top of OpenStack Swift that aims at selecting the most appropriate storage tier for any object to be stored. The goal of the multi-tiering storage we propose is to be automated and seamless, guaranteeing the required storage performance at the lowest possible cost. The paper discusses the design challenges, the proposed algorithmic solutions to the scope and, based on a prototype implementation it presents a basic proof-of-concept validation.
Departement: School of Engineering
Organisational Unit: Institute of Applied Information Technology (InIT)
Publication type: Conference Paper
DOI : 10.1109/SC2.2018.00011
ISBN: 978-1-7281-0236-8
Appears in Collections:Publikationen School of Engineering

Files in This Item:
File Description SizeFormat 
2018_Höppli_Hera_object_storage.pdf866.78 kBAdobe PDFThumbnail

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