Research On Virtualization Technologies In Optical Data Center Networks

Data centers(DCs)have the advantages of high-computing and large-scale storage.And the optical transmission technologies such as wavelength division multiplexing(WDM)and orthogonal frequency division multiplexing(OFDM)have the advantage of high bandwidth,low latency,fine granularity and low power consumption.The DC and the optical transmission technologies joint together to comform the optical data center networks(ODCNs),which can satisfy the requirement of mass storage and high computing ability from network demands.The technologies of cloud computing and virtualization are widely adopted in DCs and the applications in DCs,such as cloud storage,VOD,3D cloud animation rendering etc.are in forms of virtual machines(VMs).The DC failure will lead to the huge loss of data and even the paralysis of many on-line services.Thus here is the problem#1,popular for the service providers:in ODCNs,how to apply VM migration to realize the fast recoveries of both failed DC and downtime services in a limited time span?Meanwhile,the fact that DCs have high energy consumption,high operation overhead,high carbon emission and low server utilization makes the power consumption of DCNs become a big challenge.However,there are few articles concerning on the entire power consumption from both IT equipment and switching devices in DCs.And the long-term,long-haul and big-size communications between the emerging huge-size VMs would take a remarkable amount of power consumption.Thus,here is the problem#2 for the service providers:how to achieve the joint objective of energy saving from both network switching and routing devices and IT equipments?Besides,in ODCNs,the infrastructure as a service(IaaS)technology facilitates the customers to rent the computing and network resources,according to their own computing requirements,and the virtual networks(VNs)are constructed.Virtual network embedding(VNE)has the natural advantage of substrate resource sharing and draws a lot of attentions.And the server consolidation and the virtual concatenation-based spectrum defragmentation are respectively the core technologies in the cloud DCNs and the elastic optical networks(EONs).Thereby,here is the challenging problem#3:how to combine the server consolidation and the virtual concatenation together and integrate them into the VNE solution in the elastic optical data center networks(EODCNs)?On the other hand,the blooming development of novel network demands such as mobile terminals,internet of vehicles and remote telemedicine forces ODCNs to reform its supporting technologies.And the spotlight next-generation 5G networks keep expanding and changing the ODCN shape-configuration and service levels.So here is the problem#4,the new threshold to be crossed by the service providers:how to perform an on-demand and high-utilized-resource VNE in the heterogeneous and multi-domain 5G networks?(1)This paper,in the first chapter,introduces the background and the researching significance on ODCNs,and summarizes the major researching problems in the key technologies of ODCNs as well as their researching status;then,respectively on the network virtualization and VNE technology,this chapter provides the introduction of both development histories and oriented challenges,and through the summarized search for their current situation,their urgent needs and the problem solutions come out;finally this chapter summarizes the contributions and innovative structures.(2)Parallel Virtual Machine Migration in WDM Optical Data Center NetworksPeople can run multiple applications independently at a single location using virtualization in DCs,and also can move VMs to a suitable place.Since DC carries plentiful applications,its failure would result in huge data loss,and the VM migration from a failed DC to the target backup ones should draw extensive attentions.Thus for mitigating the service paralysis caused by a DC failure,this chapter proposes an effective parallel inter-DC VM migration framework with the consideration on VM grading,load balancing and elastic bandwidth allocation.(3)Virtual Network Embedding for Power Savings of Servers and Switches in Elastic Data Center NetworksIn EODCNs,it becomes very essential to achieve the virtual optical network(VON)embedding along with two main components:VM mapping and VL mapping.In VM mapping,this chapter tries to find appropriate servers to hold VMs.While in VL mapping,the optimal substrate path should be determined for each virtual lightpath.The power efficiency is becoming a significant concern of the VON embedding in EODCNs,and some VON embedding solutions have been proposed to improve the power efficiency through sleeping light-duty servers.However,aside from servers,the switches in EODCNs are also power contributors.Thus,with the comprehensive consideration of load balancing and power-efficient VON embedding,this chapter formulates the problem and designs a novel heuristic for EODCNs,with the objective of performing the integrated power savings for servers and switching devices in a load-balancing manner.Also,the server clustering strategy is utilized to dynamically update intra-cluster server lists managed by the upper traffic controllers.(4)Novel Elastic Optical Network Embedding Using Re-optimized Virtual Concatenation(VC)Framework Accompanied by Hitless PPSM FunctionEON provides fine-grained and orthogonal frequency slots for the connection establishment that will realize the high-rate data transmission among DCs.With the implementation of cloud DCs and EON virtualization,it becomes essential to achieve the EON embedding(EONE).Especially,the link mapping must satisfy some rigid constraints in terms of spectrum continuity,consistency and non-overlapping,which may create spectrum fragments over fiber links.Thus,this chapter re-optimizes the VC framework accompanied by hitless push-pull spectrum movement function when the common link mapping fails.The proposed design framework integrates spectrum movement with inverse multiplexing,so that the more EONE requests can be satisfied via performing effective spectrum defragmentation.This chapter also optimizes the number of sub-bands with the constrained buffer size to compensate the differential delay at the receiving end.The problem is formulated using the relaxed integer linear programming(ILP)model,and the corresponding heuristic tailored to four different service-sorting strategies is also designed.(5)Virtual 5G Network Embedding in a Heterogeneous and Multi-Domain Network InfrastructureThe emergence of the fifth generation mobile communication(5G)integrating wireless and wired domains,can be qualified for the complex virtual network operations oriented to the cross-domain requirement.This chapter focuses on the multi-domain virtual network embedding in a heterogeneous 5G network infrastructure,which facilitates the resource sharing for diverse function demands from fixed/mobile end users.This chapter proposes the mathematical ILP model for this problem.And based on the layered-substrate-resource auxiliary graph and an effective six-quadrant service-type-judgement method,5G embedding demands can be classified accurately to match different user access densities.A collection of novel heuristic algorithms of virtual 5G network embedding are proposed.(6)The sixth chapter is the summary of the full-text researching work and shows the prospects of the future researching plans.