Researches In Resource Allocation Algorithm In Networks Virtualization

Internet facilitates achieving and exchanging information, however, thedevelopment of internet faces challenge. The current Internet is plagued with anumber of many fundamental limitations, it is hard to radical change the architecturedue to its multi-provider nature, and as a result, the internet is in ossification.Through Network Virtualization, Internet can effective overcome the ossification andmake the architecture evolutionary. Network Virtualization diversifies the FutureInternet architecture into separate Virtual Networks (VN). The VNs can simultaneoussupport multiple network architectures, experiments and services over a sharedsubstrate network. A fundamental problem in instantiation of Virtual Networks (VNs)is an optimal allocation/mapping of resources. Effective resource allocation algorithmcan improve the resource utilization of infrastructure provider and reduce the cost. Soresearch on load balance schemes has great significanceFirstly, this paper analyzed central resource allocation algorithm and distributedresource allocation algorithm. Aims to resolve the problem of topology sparsity inresource allocation algorithm based on VN decomposition, this paper proposed anadvanced Resource Allocation Algorithm (ARMA) to maintain the compactness ofvirtual network topology in substrate network. The algorithm base on VNdecomposition and obtain the result that use least bandwidth via cooperation ofsubstrates and Central-coordinator. ARMA consists of three algorithms: Star MappingAlgorithm (SMA), Central-coordinator Receiving Data Algorithm (CRDA) andSubstrate-nodes Receiving Data Algorithm (SRDA). SMA allocates resource forcertain star virtual network, and then sends the result and the amount of substratelinks consumed in to Central–coordinator. Through CRDA, Central–coordinatorcompares and chooses the result that consumes the least substrate links as finallyresult, then send the finally result to all substrates nodes. In SRDA, the substrate nodeupdates free resource information about other substrate, and then ready to allocateresource for next star virtual network. Finally, simulation results show that thealgorithm can obtain better delay performance, few message volume and substratelink consumptions.To maintain load balance in substrate nodes, this paper propose an activedynamically resource allocation algorithm based on virtual node live migration. Thealgorithm estimates the load distribution in all substrate nodes via the average loaddivergence degree, and compares it with the tolerant divergence degree of substrate network to determine whether adjust resource allocation. During adjusting resourceallocation, the algorithm shift load from busy nodes to free nodes by virtual node livemigration to maintain load balance, and the impact of migration on bandwidth insubstrate links and delay in virtual links is also taken into account. Simulation resultshows the algorithm can effective balance load in substrate nodes, and impact onbandwidth and delay is slight.