The Re:GRIDiT Protocol: Correctness of Distributed Concurrency Control in the Data Grid in the Presence of Replication

In recent years, advances in high performance distributed computing have brought forward novel types of applications which access and share resources across heterogeneous, distributed environments. However, at the same time, it has to be noted that current approaches to ensure the correctness of the execution of applications in such distributed environments in the presence of concurrent updates do not scale and are not able to cope with the dynamic nature of these environments, particularly when the replication of huge amounts of data has to be considered. Due to the particular characteristics of such infrastructures (such as data Grids or peer-to-peer networks), especially due to the absence of a global coordinator, new protocols for the synchronization of updates and their subsequent propagation are urgently needed. Despite the considerable work which has been done in the context of distributed transaction management and replication management, there is no protocol which can be seamlessly applied to such environments without impacting correctness and/or overall performance. In this paper we propose Re:GRIDiT, a novel and flexible distributed transactional model, that provides seamless support of replicated data and provably correct transactional execution guarantees without any global component.