System Architecture and Data Allocation/Coordination Technologies of Autonomous Decentralized Database System for High Assurance

Abstract

During the last few years there have been important changes in the way that consumers use the Internet and e-commerce. This exceptional expansion, combined with dynamic interactions with users and providers, is boosting the development of new electronic business models, such as block chain, fintech, bitcoin, Uber, Airbnb, etc. The information systems that support such business processes, usually depend on distributed database systems. These approaches have mostly opted to exploit two important characteristics of this kind of applications: at first, they contain a set of commodity types with a limited inventory. On the other hand, incremental updates are the kind of operations which these items typically involve. In this sense, it is possible to achieve system distribution by using tokens which represent the commodities? instances for sales in e-business. There exist two mainly different mechanisms for distributing tokens: replication and partitioning. Token replication mechanism demands expensive synchronization protocols to provide the consistency of the data. On the other hand, token partitioning, depends on token redistribution strategies which allow dynamic migration of tokens to those servers in which it is needed. Such strategies are developed from the standpoint that the total system can locally be known, and therefore, they fail to address the important problem of how to allocate and coordinate the data to make it adaptive to rapidly changing situations. In this paper, we present Autonomous Decentralized Database System (ADDS) concept and architecture in order to extend the properties of autonomy and decentralization to wide-area distributed database systems. ADDS fundamental technologies confers autonomy and loosely coupling to all the components of the system (databases) and provides a background coordination to adapt the system to evolving situations. Moreover, we present two data allocation and coordination technologies for high response time based on specific situations. Finally, in order to show the effectiveness of our proposal we compare it with conventional partitioning strategies.