Search in Unstructured Peerto-Peer Networks based on Dynamic Topology Adaptation

摘要

Peer-to-Peer networks (P2P) have gained great attention and popularity. One key challenge aspect in Unstructured P2P resource sharing environments is efficient searching algorithm. Flooding and Random Walk (RW) are two typical search algorithms. Flooding searches aggressively and covers most nodes but it generates large amount of query messages. RW searches conservatively and at each hop it generates a fixed amount of query messages but it takes longer search time. This paper proposes the Dynamic Search Algorithm (DS), a generalization of flooding and RW. It resembles flooding for short-term search and RW for long-term search. This system describes a multi-database system as a 4 tiered Client-Server DBMS architecture. The proposed system provides remarkable changes in database servers which process thousands of concurrent queries. The load balancing mechanism is implemented by using a four tier model. The first component of this system is a web based interface or Graphical User Interface. The second component is a client application program running in an application server. The third component is a Global Database Management System which manages multiple database servers. The GDMS receives all queries, divides and forwards to respective databases. The fourth component is a remote homogenous local component database system server. More than one database component depends on the requirements of the application. . Transaction submitted from the client interface to a multi-database system server through an application server will be decomposed into a set of sub queries and will be executed at various remote homogenous local component database servers and also, in case of information retrieval, all queries will be composed and will get back result to the end users.