Self-Maintainable Views Group in P2P Environment
Abderrazak Sebaa, Tari A. Kamel · American academic & scholarly research journal · 2014
Abstract. Peer-to-Peer environment is gradually more popular for sharing data for distributed applications. Peer-to-peer makes these applications more reliable, efficient, available and scalable. Also, materialized views which are derived from base relations are often used to speed up query processing and data sharing. In this paper, we are interested in materialized view maintenance in P2P environment; we propose a new approach which uses some characteristics of views in such system. Our maintenance approach consists to create virtual groups of peers. Groups will contain views that sharing maximum of common data sources, then, we will associate at each group a set of auxiliary views which will be stored in a peer chosen by our algorithm, it will be the group center. The auxiliary views set will allow the group to be self-maintainable. Then, we propose a view maintenance algorithm based on two steps, notification of modifications and computing & sending of view updates. Finally simulation experiments implemented with java, show that both maintenance cost and total message number are reduced using our approach, we compared group maintenance cost in various situations in order to find the optimal conditions that our proposal is more efficient.Keywords: view materialized; P2P; view maintenance.(ProQuest: ... denotes formulae omitted.)1 INTRODUCTIONPeer-to-peer (P2P) systems are in full growth since several years. This paradigm allows the design of very large systems with high availability and low cost. Indeed, an important applications class in P2P system for data sharing exists, where many users must be able to work on the same data. Often materialized views are defined over tables and\or others existing views, it's one among mechanism that implement data sharing, but consistency its main issue, when the underlying data sources changes, materialized views contain non updated data and remains until their refreshment. Most of view maintenance techniques require data sources access; these approaches are not always possible because data sources may be distributed across different sites, and sometime they are unavailable, but even if they are available, communication cost of their access may be prohibitive. For these reasons, self- maintainability of materialized views is recommended (CHEN, LIU, & Rundesteiner, 2004; Gupta, Jagadish, & Mumick, 1996; Mohania, & Kambayashi, 2000). Materialized view is said self-maintainable if any view update in response to a change can be computed without using data sources (Samtani, Kumar, & Mohania, 1999). In P2P environment materialized views self maintenance will be the most appropriate solution. In this paper, we are interested in materialized view maintenance in P2P environment. Our objective is to implement the principle of self-maintenance in this environment.The specific contributions of our work are as follows.* Specific steps to decompose materialized views of the system to groups are proposed. Our decomposition offers several advantages, where each group contain views that share maximum of common data sources,* An effective designation of the center of each group, the center will serve as a location for materialized auxiliary views.* Balancing of group sizes for an efficient management of their contents.* Materialized view self-maintenance algorithm is proposed.* Validation of our solution by experiments simulation, we give results and corresponding analysis.The remaining of this paper is organized as follows. Section 2 contains a brief overview of related work. Section 3 provides motivation, describes views decomposition into groups and how materializing auxiliary views. In Section 4, self maintenance algorithm is given. Experiment results are provided in section 5.2 RELATED WORKMaterialized view maintenance has been the subject of much research, this process has definite interest with data warehouses as in (Zhuge, Garcia-Molina, Hammer, & Widom, 1995;Agrawal, El Abbadi, Singh, & Yurek, 1997; Ding, Zhang, & Rundensteiner, 1999; Zhang, Yang, & Wang, 2010; Gupta, Jagadish, & Mumick, 1996; Quass, Gupta, Mumick, & Widom, 1997; Samtani, Kumar, Mohania, 1999; Mohania, & Kambayashi, 2000; Cui, & Windom, 2000; Laurent, Lechtenborger, Spyratos, & Vossen, 2001), other are directed towards the views maintenance in the distributed environment (Mork, 2005; Agrawal, Silberstein, Cooper, Srivastava, & Ramakrishnan, 2009 ; CHEN, LIU, & Rundesteiner, 2004) , bat little work for P2P environment ( Qin, Wang, & Du, 2005; Bellahsene, Cart, & Kadi, 2010; Li, & Ishikawa, 2010). …