Environment for the simulation of different ACP and error recovery in Distributed Data Bases

Abstract

This paper presents an evolution of a simulation environment where situations in which a DDB should maintain data integrity are modeled and implemented. These situations include failures taking place during transactions, and the focus is mainly on the utilization and later comparison of different atomic commit protocols (ACP). The implementation was made on Java due to different aspects such as ease of work, portability, etc. The environment is mainly based on failure recovery of transactions in a distributed data environment, and it allows to choose for the simulation development between two-phase, three-phase, optimistic and pessimistic protocols. The task scheduler defined and implemented to carry out each simulation, based on an execution design establishing each specific problem to be solved, is also specified.