Forecast-based resource reconfiguration algorithm for network virtualization

Abstract

In order to solve imbalance issues of used resources of substrate network,which are brought by virtual net-work joining and leaving dynamically,forecast-based resource reconfiguration algorithm（FRRA） for network virtualiza-tion environment was proposed.Algorithm FRRA took full advantage of known information to calculate interval of re-configuration and used this interval to replace the periodic time span in previous algorithms.In order to make reconfigu-ration time global optimal,it took two measures:resources were divided into critcal resources and general resources,and different algorithms were used to key resources and common resources respectively;according to the probability that re-source migration was failure,the limit number of reconfiguration requests was deduced.FRRA algorithm with VNA-II algorithm and PMPA algorithm in the experiments was compared.The results show that algorithm FRRA reduces recon-figuration cost 69% than that of VNA-II and 42% than that of PMPA.Algorithm FRRA improves VN request reception rate 29% than that of VNA-II and 52% than that of PMPA.

Keywords

• network virtualization;resource allocation;reconfiguration;forecast