Authors:
Armen Dzhagaryan
and
Aleksandar Milenković
Affiliation:
Electrical and Computer Engineering Department, The University of Alabama in Huntsville, Huntsville, AL and U.S.A.
Keyword(s):
Cloud Computing, Lossless Compression, Energy Efficiency, Performance Evaluation, Cost Efficiency.
Related
Ontology
Subjects/Areas/Topics:
Distributed Intelligent Agents
;
Embedded Communications Systems
;
Mobile and Pervasive Computing
;
Networking and Connectivity
;
Pervasive Embedded Networks
;
Telecommunications
;
Ubiquitous Computing Systems and Services
;
Ubiquitous Multimedia
Abstract:
An increasing reliance on cloud and distributed processing of scientific and big data in commercial, academic, and government institutions necessitate new approaches to optimize file transfers. Lossless data compression and decompression is essential in improving the overall effectiveness of file transfers between edge devices and the cloud by increasing communication throughput, reducing connection latency, making effective use of cloud storage, and reducing costs. This paper experimentally evaluates effectiveness of common and emerging general-purpose compression utilities for file transfers between edge devices and the cloud. The utilities are evaluated in terms of throughput and costs during representative file transfers between a workstation and the cloud, while varying LAN network conditions. The results show that the optimal compressed transfer modes improve both upload and download throughputs. For uploads, the peak improvements range from 5.16 to 25.6 times relative to uncom
pressed file uploads, and from 1.33 to 17.4 times relative to the default compressed uploads. For downloads, the peak improvements range from 3.82 to 19.57 times relative to uncompressed downloads, and from 1.8 to 13.8 times relative to the default compressed downloads. In addition, the best performing compressed transfer modes reduce the costs related to cloud computing.
(More)