Simulation And Performance Analysis Of A Chord-based Service Identifier Resolution System

As the rapid expansion of the quantity of resources in the current Internet, every kind of services has its own style of management, so there is no universal solution. New kinds of services are restricted by the old network architecture. DNS has a few shortcomings as updating is slow, system configuration is complicated, and mobility is unsustainable.The project "a fundamental research on universal trustworthy network and pervasive service" puts forward a new concept of identifier network. It divides the Internet into two layers named "service layer" and "network layer". Service identifier (SID) resolving system in the "service layer" solves some problems of traditional network well. In the universal network, SID resolving system takes charge in distributing and administering SID, and each service has one unique SID and the SID resolving system is in charge of that. The SID resolving system adopts distributed structure. The SID resolving system adopts distributed structure. It has a few benefits as follows: good performance on describing and explaining services, sustainable mobility, robustness, sustainable load balance, easy to configure. SID resolving system adopts Chord algorithm, so the servers are peer to peer. The management model has performed well in the current limited environment.Firstly, this paper analyzes three kinds of delays of the SID resolving system, and then it finds out the key delay which affects the efficiency of query via simulation. Toward this kind of delay, this paper puts forward a plan of adding cache mechanism in the SID resolving system. The core idea of adding cache mechanism is to improve the efficiency of query by sacrificing part of the system capacity. In the design of the current system, we assume that it has to pass n hops before the user's request gets to the object server according to the algorithm. The improved method is: saving the query result in the cache file of every passed server by the recursive response message. So if the same request arrives at any server above, it will obtain the result directly, and does not have to get to the object server. Countering the practical situation of resources and services requirement distribution, this paper achieves the query request module, simulates hundreds of SID resolving servers, designs a method to gain statistic, and then gradually improve the cache mechanism based on different amount of visiting of services. At last, we analyze the performance of SID resolving system in contrast of adding cache mechanism before and after. The result of simulation shows that the efficiency of query of SID resolving system is improved on a large scale.