Fundamental trade‐offs between resource separation and resource share for quality of service guarantees. Issue 1 (1st March 2014)

• Record Type: Journal Article
• Authors: Shioda, Shigeo
• Is Part Of: IET networks. Volume 3:Issue 1(2014)
• Page Start: 4
• Page End: 15
• Publication Date: 2014-03-01
• Subjects: resource allocation -- Internet -- quality of service -- telecommunication network topology -- telecommunication traffic; resource separation -- quality of service guarantees -- resource‐allocation strategies -- QoS -- Internet -- network resource -- resource‐sharing strategy -- prioritised traffic -- statistical multiplexing gain -- IntServ -- network topology; Computer network architectures -- Periodicals; Computer network protocols -- Periodicals; Information networks -- Periodicals; Telecommunication systems -- Periodicals; 004.605
• DOI: 10.1049/iet-net.2013.0120
• Languages: English
• ISSNs: 2047-4954

Abstract

Abstract : There are two possible resource‐allocation strategies for the quality of service (QoS) guarantee in the Internet: the resource separation and the resource share. The resource‐separation strategy allows a prioritised flow to occupy its own network resource, which is physically or logically separated from resources used by other traffic. The resource‐sharing strategy secures some portion of network resources for prioritised traffic, but secured resource is shared by all of prioritised flows. This study compares these two resource‐allocation strategies in terms of the provided QoS level. It is found that the two strategies form a striking contrast to each other in terms of the resource usage. The resource share can benefit from the statistical multiplexing gain, but it increases the envelope of each individual multiplexed flow. In contrast to this, the resource separation does not increase the envelope of flows so much, while it does not benefit from the statistical multiplexing gain. That is, the general belief that the resource separation like the IntServ provides better QoS is not always true, and various conditions including network topology or QoS target would determine which strategy is preferable.