Authors:
A. Jiménez-Yáñez
1
;
J. Navarro
2
;
F. D. Muñoz-Escoí
3
;
I. Arrieta-Salinas
1
and
J. E. Armendáriz-Iñigo
1
Affiliations:
1
Universidad Pública de Navarra, Spain
;
2
Ramon Llull University, Spain
;
3
Univ. PolitÉcnica de València, Spain
Keyword(s):
Scalable Datastore, Epidemic Data Replication, Performance Evaluation, Consistency.
Related
Ontology
Subjects/Areas/Topics:
Cloud Applications
;
Communication Networks and Protocols
;
Distributed and Mobile Software Systems
;
Distributed Architectures
;
Software Engineering
Abstract:
Epidemic data replication protocols are an interesting approach to address the scalability limitations of classic distributed databases. However, devising a system layout that takes full advantage of epidemic replication is a challenging task due to the high number of associated configuration parameters (e.g., replication layers, number of replicas per layer, etc.). The purpose of this paper is to present a Java-based simulation tool that simulates the execution of epidemic data replication protocols on user-defined configurations under different workloads. Conducted experiments show that by using the proposed approach (1) the internal dynamics of epidemic data replication protocols under a specific scenario are better understood, (2) the distributed database system design process is considerably speeded up, and (3) different system configurations can be rapidly prototyped.