Abstract:The topological properties of overlay networks are critical for the performance of peer-to-peer (P2P) systems. The size of routing table and the diameter are among the most important parameters which measure the autonomy, efficiency, robustness and load balancing of P2Ps, and the Moore bound sets the optimal tradeoff between diameter and degree for any graph. In order to improve the four features for structured P2P networks, researchers have proposed many designs based on the interconnect networks which can approximately achieve the Moore bound. However, the performances of these systems in dynamic environments are far worse than their corresponding interconnect networks in the static environment. Moore bound cannot give a good description for existing P2P systems due to their dynamic features. In this study, therefore, we first prove a new lower bound of the network diameter and average query distance in a highly dynamic environment. The routing latency of the existing systems cannot be more better than the new bound. This is because the existing systems require a fixed number of peers known a priori. P2P systems on the other hand, typically are dynamic, meaning that peers frequently join or leave. To solve the issue, we propose a dynamic multi-way trie tree structure, based on which any interconnect network can be adopted to construct a P2P system. The unified construction method also consists of a series of optimal schemes for P2P systems. DeBruijn and butterfly representing the single and hybrid protocols based on the dynamic multi-way trie tree structure are used to construct two new P2P systems (DPhoenix, BPhoenix) whose performances can exceed the new bound. Also, DPhoenix and BPhoenix can be easily applied to other interconnection networks after minimal modifications, such as, hypercube, kautz, shuffle-exchange and CCC.

P2i-Torus: A Hybrid Artchitecutre for Direct Interconnection

Evaluation of Various Networks Configurated by Adding Bypass or Torus Links.

Using Direct Interconnection Networks For Load-Balanced Architecture

A Cost-Effective Interconnect Architecture For Interconnection Network

STorus: A New Topology for Optical Network-on-chip.

Fully adaptive routing in torus networks based on center distance

Hierarchical and Reconfigurable Optical/electrical Interconnection Network for High-Performance Computing

Research on Next-Generation Scalable Routers Implemented with H-Torus Topology

Performance Metrics Analysis of Torus Embedded Hypercube Interconnection Network

P-Torus: wavelength-based switching in packet granularity for intra-data-center networks

On Fully Reconfigurable Optical Torus-Based Networks-On-Chip Architecture

A Fully Adaptive Routing Algorithm in Torus Switching Fabrics

TM: a New and Simple Topology for Interconnection Networks

A New Optical Network-on-Chip Architecture for Chip Multiprocessor

A Compressed 3D-Torus Architecture for Scalable Router

Dragonfly-of-Torus: A Reconfigurable Network Topology for High-Performance Computing

Network Performance of Pruned Hierarchical Torus Network

A Unified Construction Method of P 2 P Networks with Lower Bound to Support Complex Queries

An Asymptotically Minimal Node-Degree Topology for Load-Balanced Architectures

Implementation of Scalable Routers with Direct Networks

Beyond the Lower Bound: A Unified and Optimal P2P Construction Method