Research On Dynamic Reconfiguration Algorithm Of Virtual Network Resources

With the fast development of cloud computing technology and the number and scale of Internet business of unprecedented growthin in recent years,the traditional structure of the Internet that followed by a series of problems,the Internet ossification problem is particularly prominent.Accepting the past way that simply increasing Internet architecture might face two mainly difficults: on the one hand the new added devices would produce high price cost since these new devices would change the current infrastruction,it made the hardware like routers even hard changed radically.On the other hand,many modern Internet businesses is mutil-provider,so the old framework could not be changed unless the original Internet Service Providers(ISPs)consistently agreed.Due to the above factors,the feasibility of constructing a brand new becomes quite low.To solve the confronting problems of Internet,some researchers raised Network Virtualization technology(NV).NV also becomes a viral and viable way to solute network rigid problem gradually.Virtual Network Embedding(VNE)is an important side in the study of Network Virtualization technology;VNE means that Virtual Network Requests(VNR)are embeded on the Substrate Networks(SN)and these VNRs shares the SN resource together.How to embed the VNRs on the SNs effectively is a hot topic past five years.Only researching VNE is not enough nowadays,since when the already mapped virtual requests' life circle run out,they would generate many fragmentation resources.This phenomenon could lead some extra problems like low acceptance rate,unbalance load,high power consumption and so on.It is necessary to reconfigure the embeded virtual nodes or virtual links purposefully to optimize the referred indexes.The follow is the main work of this paper.(1)Designing and achieving the heuristic energy-aware reconfiguration algorithm(HEARA).Its main aim is saving energy,this algorithm migrate the virtual node or virtual link on the physical device which load is lowest to another physical device whose rest resources are more a threshold value,then we can close the zero-load physical devise to carry energy-saving point.Because of we do some work on selecting moving candidate,the index acceptance rate and load balance are taken into account,besides,we add path-splitting algorithm in this paper.The experiment shows that HEARA is known to have been one of the few comprehensive energy-aware algorithms which not only increase the acceptance rate but also consider other performance like load balance and so on.(2)Considering the low acceptance rate and unbalanced load issue in virtual network embedding,designing and achieving the Dynamic Substrate Network Reconfiguration Algorithm(DSNRA),the general process of DSNRA is as follows: a new physical node and related links will increase when a virtual request was rejected due to bottleneck substrate devices.Then DSNRA found the bottleneck node and other high load rate physical nodes within the distance constraint and put them in a corresponding set.Next,comparing the revenue with the migration cost of the already mapped virtual nodes that were in the above-mentioned set,the decision will make whether to relocate them and their related virtual links.Finally,adding another new physical node and related links and repeat the before steps continuously until the node load rate of the set is under the pre-set index.Experimental results demonstrated that DSNRA could improve acceptance rate and balance the load.(3)This paper combined HEARA with DSNRA,when SN lack of resources,it preferred finding sleeping physical nodes and links reactivate them then made related migration to adding new substrate devices.To do this,the mix algorithm has lower energy consumption than DSNRA and more balance load than HEARA.It is an comprehensive optimization considering several indicators.