Congestion-Aware Embedding Of Oversubscribed Virtual Data Centers

Predictable network performance is critical for cloud applications and can be achieved by providing tenants a dedicated virtual data center(VDC)with bandwidth guarantee.Recently,the extended Hose model was applied to the VDC abstraction to characterize the tradeoff between cost and network performance.The acceptability determination problem of a VDC with heterogeneous bandwidth demand was proved to be NP-complete,even in the simple tree topology.The embedding problem of oversubscribed data center in physical data center network was studied.Compared with the general virtual data center,the traffic pattern between the virtual machines in the oversubscription virtual data was more complicated.Therefore,the flow model and the embedding problem were described first in mathematics.A heuristic algorithm with lower time complexity was proposed for the embedding problem of virtual machine-packet perturbation algorithm.The perturbation algorithm is network congestion-aware as it detects the bandwidth bottlenecks in the placement process and then selectively relocates some assigned VMs to eliminate congestion.Finally,the simulation algorithm was used to compare the packet perturbation algorithm with the algorithm proposed in the previous work and the famous first-fit.The experiment shows that the proposed algorithm improved the embedding success rate while reducing the complexity of the algorithm.