پردازش معامله در شبکه پایگاه داده نقطه به نقطه

This paper investigates a transaction processing mechanism in a peer to peer database network. A peer to peer database network is a collection of autonomous data sources, called peers, where each peer augments a conventional database management system with an inter-operability layer (i.e. mappings) for sharing data. In this network, each peer independently manages its database and executes queries as well as updates over the related data in other peers. In this paper, we consider a peer to peer database network where mappings between peers are established through data-level mappings for sharing data and resolving data heterogeneity. With regards to transaction processing in a peer to peer database network, we mainly focus on how to maintain a consistent execution view of concurrent transactions in peers without a global transaction coordinator. Since there is no global transaction coordinator and each peer executes concurrent transactions independently, different peers may produce different execution views for the same set of transactions. For this purpose, we investigate potential problems that arise when maintaining a consistent execution of concurrent transactions. In order to guarantee consistent execution, we introduce a correctness criteria and propose two approaches, namely Merged Transactions and OTM based propagation. We assume that one single peer initiates the concurrent transactions. We also present a solution for ensuring the consistent execution view of concurrent transactions considering the failures of transactions.

A peer to peer database network (P2PDBN) is a network of peers where each peer, as an independent database, participates in the network to share data with other peers. The local databases on peers are called peer databases. In a P2PDBN, each peer chooses its own database schema and maintains data independently. Although peer databases are created independently, data in one peer may semantically relate to data in another peer. Therefore, each peer specifies pair-wise mappings with other peers for sharing and exchanging related data. From the view of design perspective, a P2PDBN is similar to a conventional federated database system (FDBS) and a multidatabase system (MDBS), since the systems are generally designed in a bottom-up fashion. The conventional systems are built from given a set of pre-existing, independently created sources. The sources in these systems are tightly coupled and the data in these sources are homogeneous, i.e. data vocabularies are same. In order to access data from distributed sources, a globally integrated schema is created to represent the information of the sources. The creation of such schemas from a set of source schemas, known as schema integration, is an interesting problem which has been studied extensively in the literatures. For creating a global database, each local source defines an export schema, which describes the data it is willing to share with others [35]. The global database is the union of all the export schemas. Authors in Ref. [34] proposed a five-layer schema architecture for a loosely-coupled federated database systems. In the architecture there is no federated schema or central administrator. The owners of the independent database systems are responsible for creating and maintaining their own federated schema(s) [5]. In this architecture, the lowest layer (first layer) consists of the local source schemas. The second layer consists of component schemas, which are translations of the source schemas into a common data model. The third layer is comprised of export schemas. Through the export schema, a source describes which part of its data it is willing to share with others. The federated schema integrates multiple export schemas and its information is, in turn, filtered through the external schemas. Each application accesses the global database by the definition of an external schema. It is also assumed that in a FDBS sources are to be stable and unchanging throughout the system's lifetime. However, in a P2PDBN, sources are loosely-coupled and data vocabularies of peers may be different but may represent the same real-world entities. Making a virtual global schema from the schemas of peers is not possible due to the pair-wise mappings between peers and the dynamic behavior of peers. From the point of view of users, the way data are retrieved by a query in a peer-to-peer system is similar to a federated database system. In both systems a user accesses the entire system through its local peer or source. However, in a federated system, a user querying the federated schema of a source can only retrieve results from the sources that participate in this local federated schema. On the other hand, the corresponding operation in a peer-to-peer system retrieves data from all the sources, both the local and remote sources. In a P2PDBN, a query posed in a peer not only retrieves data from its acquainted peers but also retrieves data from their acquainted peers and their acquainted peers and so on. Retrieving data from sources that a user was previously unaware of is one of the distinguishing characteristics of peer-to-peer systems and it is one of the main benefits of the peer-to-peer integration methodology.

In this paper, we introduce a transaction execution model for a peer to peer database network where sources are heterogeneous and instance-level mappings are used to associate data from different sources. Our approach is scalable because a peer doesn't need any global knowledge of the system, and there is no global coordinator. Transactions are processed by each peer independently and consistency is maintained recursively through acquaintances. A peer only ensures the serializability of its immediate acquaintances by ensuring acquaintance-level serializability. Mainly, we contribute the following: • We analyze the execution of transactions in a peer to peer database network. • We introduce a correctness criteria to ensure the consistency of a peer to peer database network during the concurrent execution of transactions initiated from a single peer. • We propose two approaches ensuring a global consistent execution of transactions without violating the autonomy of LDBSs. A future goal is to investigate the transaction processing when global transactions are initiated from many peers that need to be executed concurrently in the system, and to analyze the correctness criteria for such executions. Note that the proposed approach does not consider the fact that translation and propagation of transactions may involve in a cycle and can generate new data at each cycle round until a fixpoint is reached [48]. Therefore, as such, the approach does not guarantee completeness. Solving this issue, too, is our future goal.