Research On Resource Preference-aware Routing And Content Caching Mechanism In SDN Networks

Software Defined Networking (SDN) and Information-Centric Networks(ICN) are the two general directions of future network development.SDN decouples the control plane from the forwarding plane, and enables the transition from traditional distributed network to the centralized network architecture.Inspired by the fact that the Internet is increasingly used for information dissemination. ICN is proposed to enable the tran-sition from traditional host-centric network to information-centic network architecture by name-based routing and in-network caching etc. Because of the salient features of ICN and SDN, there are increasing attempts to integrate ICN and SDN into a Software Defined Information-Centric Network (SD-ICN). which offers the opportunity to logi-cally centralized control of the ICN.In SD-ICN, the control plane can programmatically make complex logic decisions, e.g., name based routing and interoperability decisions, while the hardware-concerned data plane only needs to execute some simple operations according to the decisions made by the control plane, such as forwarding packets and caching content objects.In the SDN paradigm, controller manages the data plane via standardized Open-Flow protocol, and flow tables are issued to determine the behaviors of switches. Flow tables of the OpenFlow swicthes are implemented by TCAM, which can provide fast lookup. But TCAM resources are expensive because of their cost and power usage, thus the switch flow tables are limited. Some existing ICN architectures(e.g. NDN) can hardly provide centrilized control of the network contents, which causes inefficient content lookup and unreliable name-based routing.Whats’s more, Traditional on-path caching strategies(LCE etc.) make popular objects to be massively cloned in the net-work and cached redundantly. In this paper, we propose routing and caching strategies accordingly to solve the above problems. And the main contributions are listed as fol-lows:(1)To obtain balanced utilization of flow table and bandwidth resources, we pro-pose the Batch Routing algorithm called BRP-SA in SDN Networks with resource pref-erence consideration.The limited flow table resources confine the number of flows that can pass through the OpenFlow switches, and the limited bandwidth resources confine the data traffic that can pass through the network links. In BRP-SA, flows are classi-fied into elephant flows and mice flows, and flow table resource preference degree and bandwidth resource preference degree are defined based on flow characterisitcs respec-tively. SDN allows a central controller to globally plan the path of the flows according to the operator’s objective. Thus, batch routing is proposed to process multiple flow re-quests simultaneously.Simulation results show that BRP-SA can effectively balance the utilization of both flow table and bandwidth resources, so that the network can accept more flow requests.(2)We propose a new SD-ICN architecture which integrates ICN and SDN, and OpenFlow protocol is extended to support content-related operations.SD-ICN provides centralized control of ICN, and simplifies the procedures of processing an ICN re-quest.Then we propose a novel Preference-aware Reactive Neighbor Caching(PRNC) algorithm with consideration of content resource preference based on the proposed SD-ICN architecture.PRNC considers the temporal and spatial characteristics of the content flow requests. And it will cache the popular contents within content requesters’neigh-bor nodes without repetition, so as to save Internet traffic and reduce time-delay for latter content requests.Simulation results show that PRNC can get coordinated utiliza-tion of both node storage and link bandwidth resources, thus reducing cache reduncy, saving netework traffic and alleviating network congestion.