Modelling the Replication Management in Information Systems

Abstract

IntroductionThis paper presents the development stages of a model of a distributed system in which the primordial abstract elements are the process and the connection (link) between processes. This paper deals with the problem of communication between the processes within the distributed system, with the concept of message exchange between processes, and also presents the problem of distributed agreement.This paper presents the importance of two concepts: the logical time of a distributed system and the causal order of messages. All of these concepts are necessary to define a model of a distributed system in which some processes have the role of management of the operations carried out by other processes which only have an execution role.After presenting of the processes of the system components, there is an abstraction of the exchange of messages between the processes of the distributed system. In the next step, an abstract model of a distributed system is proposed and analyzed, in which there are modelled several worker-processes controlled by a manager-process, which runs on different host than the worker's. The model described here forms the basis for the development of a distributed system tolerant of faults, which uses the replication of operations.Based on the abstract model of the distributed system previously presented, a replication model is proposed, for operations and data based on the structure of a distributed system containing several worker-processes. After the presentation of a list of objectives for the fault-tolerant system, a logical separation of management operations from the execution operations and data storage is proposed.Web services are increasingly used in distributed systems, applications and services of outmost importance. For distributed systems, the architecture oriented on services, SOA - Service Oriented Architecture is already accepted as being the architecture able to interconnect applications running on different operating systems and facilitates the complex interactions between autonomous systems and heterogeneous ones, either within organizations or between organizations in relations of B2B (business-to-business) type. Web services allow software applications found in different organizations (company) to interact with each other, even if those organizations use different hardware systems, or operating systems and even different programming languages. Web services are able to standardize and improve the effectiveness of business activities on the Internet by automatically invoking operations which otherwise should be invoked manually by a human operator. So, Web services allow direct interactions between computers located in different organizations.Based on Web services it is possible to build technologies to support interoperability with Decision Support Systems. Important aspects for healthcare systems are described in [1], where the proposed technology is meant to raise the interoperability degree between different medical information systems is described. Web Services ensure communication between medical units automatically with minimum human intervention, which is an essential requirement when designing applications for users in the medical domain.The design of applications based on Web services can take into account important aspects in creation of new levels of abstraction, giving a special importance to scalability and adaptability. Due to a growing importance given to resource-oriented Web services, the authors must implement the principle of organizing the algorithmic resources in multi-level hierarchy and the principle of uniform access to resources. In order to implement these principles, an architectural model of Web applications is promoted in [2].2Availability and Reliability of Distributed SystemsThe failures and damages of web applications may lead to incorrect processing or even to the system failure in the business of e-commerce type, e-banking or other systems based on transactions. …