Virtualized dynamic resource allocation algorithm for the internet Diffserv domains

Al-Wasity, AJL 2017, Virtualized dynamic resource allocation algorithm for the internet Diffserv domains , PhD thesis, University of Salford.

[img]
Preview
PDF - Accepted Version
Download (86MB) | Preview

Abstract

The Differentiated Services (DiffServ) architecture has been proposed for providing different levels of service to the Internet Protocol (IP) traffic. Current discussions in the DiffServ networks are focused on managing resources dynamically according to the traffic conditions of the DiffServ router (Per Hop Behaviour). Software Defined Networks (SDN) and Network Function Virtualisation (NFV) technologies have recently emerged in the research agenda to support researchers in managing network domains and to achieve better use of domain resources. This thesis introduces a new scheduling algorithm called “Dynamic Resource Allocation Management - Network Function Virtualization (DRAM-NFV)” to allocate the service classes resources in the proportional delay DiffServ domains. DRAM-NFV algorithm manages the resources among service classes within the edge routers of the DiffServ domains dynamically according to their traffic conditions and manages these resources between the DiffServ domains in the event of congestion based on their traffic conditions at the egress routers of the upstream domain and ingress routers of the downstream domain. The NFV executes the DRAM-NFV algorithm on a virtualized - Network as a Service (NaaS) - cloud infrastructure to manage the SDN controllers for the edge routers of the DiffServ domains through monitoring the traffic conditions in the service classes at the edge routers and reallocating the out-link resources of the edge routers among service classes.

A number of test scenarios were conducted in this research in order to test the performance of the DRAM-NFV algorithm. The performance of DRAM-NFV algorithm is compared with the performance of the DWFQ algorithm by comparing the average End to End Delay for service classes traffic and links utilization. The DWFQ algorithm cannot manage resources between DiffServ domains but can manage the resources locally and dynamically for each DiffServ domain separately.

The network simulator NS3 has been used to implement these test scenarios and to test the performance of the DRAM-NFV algorithm. The results show that with the DRAM-NFV algorithm, better balance for DiffServ domains resources can be achieved through monitoring the bandwidth hungry service class at the downstream domain and managing its resources at the upstream domains. As a consequence of this, the utilizations of some service classes traffic are improved and the average End to End Delay for overall traffic are also reduced. An example of the improvement that was achieved by managing resources between (upstream and downstream) DiffServ domains dynamically, in test scenario 3- Case Study 2, the average utilization for the highest priority class (SC1) for whole period of simulation at the destination end is increased by 0.175% and the average End to End Delay for overall traffic is also reduced by 800 msec. As a result of reducing the average End to End Delay for overall traffic and improving the utilizations of service classes traffic, the QoS of applications traffic can be improved during the congestion periods in DiffServ domains.

Item Type: Thesis (PhD)
Contributors: Al-Yasiri, Adil (Supervisor)
Schools: Schools > School of Computing, Science and Engineering
Depositing User: Ahmed Jumaa Lafta
Date Deposited: 19 Feb 2018 14:30
Last Modified: 19 Mar 2018 01:38
URI: http://usir.salford.ac.uk/id/eprint/43695

Actions (login required)

Edit record (repository staff only) Edit record (repository staff only)

Downloads

Downloads per month over past year