arrow
Return

HOSS: Hybrid Object Storage System for Performance Acceleration

delete2022-03-01
delete3
PRE
AI
Y
Yeongjin Kim
T
Taehoon Kim *
DOI:10.1109/JSYST.2021.3097383delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
In this article, we propose a novel architecture for the object storage system in order to improve PUT object throughput without degrading availability and durability performances. Our proposed architecture resolves networking and/or processing resource bottlenecks of an arbitrary cloud infrastructure by flexibly performing erasure encoding of the object data in the following two tiers: first, application programming interface (API) gateway and second, storage servers. Simulation results demonstrate that our hybrid object storage system increases the PUT object throughput up to 2x compared to the state-of-the-art object storage systems where data encoding is performed in static, i.e., only a single tier among API gateway and storage servers can encode object data.
Keywords:
Servers
Logic gates
Encoding
Throughput
Downlink
Cloud computing
Uplink
Erasure coding
object storage system

Journal

I
IEEE Open Journal of Circuits and Systems
IF:
2.4
Papers:
4.5K
Citations:
387

Organization

Hanbat National University cover
Hanbat National University
Scholars:
2.1K
Papers: 2.2K
Citations: 1.9K
I
Inha University
Scholars:
1.1W
Papers: 1.1W
Citations: 1.1W