As the core infrastructure of cloud computing,a large scale of the Data Center Network(DCN),which consists of millions of servers with high capacity,suffers from node failure such that the reliability is deteriorated.In this thesis,we study how to improve the reliability of DCNs.In fact,any network can be mapped to a topology graph,the reliability of which depends on its fault tolerance such that the design for either efficient diagnosis or reliable routing is required.It is worth to mention that as a deformation of a hypercube,the balanced hypercube is well-known by its strong connectivity,regularity and relatively small diameter,each of which makes a balanced hypercube a trustworthy model to deal with information traffic and provide a certain degree of fault tolerance as well.In this thesis,we propose balanced hypercube strategies for reliability analysis and improvement in DCNs.The details are given as follows.First,from the perspective of fault tolerance abilities,both of restricted con-nectivity and restricted edge-connectivity are key measurements.Be more specific,g-restricted(edge)connectivity can ensure every remaining component to be dis-connected,where the number of neighbors of node is no less than g.We prove{1,2}-restricted edge connectivity and the {1,2}-restricted connectivity of balanced hypercubes to show advantages on balanced hypercube DCNs.Second,from the perspective of diagnostic performance,(t,k)-diagnosability can identify and replace at least k faulty nodes in each iteration until all faulty nodes are found.We prove the(t,k)-diagnosability of balanced hypercube under the PMC model and give the corresponding diagnosis algorithm of low complexity.Third,from the perspective of fault-tolerant routing,the safety level based scheme is integrated.Be more specific,referring to different safety level of spanning subcubes that contains the source and destination nodes,the shortest feasible paths are located such that the reliable transmission is achieved. |