Data Transfer Evaluation of Nomadic Data Consistency Model for Large-Scale Mobile Systems

Masahiro KURODA  Ryoji ONO  Yoshiki SHIMOTSUMA  Takashi WATANABE  Tadanori MIZUNO  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E82-D   No.4   pp.822-830
Publication Date: 1999/04/25
Online ISSN: 
DOI: 
Print ISSN: 0916-8532
Type of Manuscript: Special Section PAPER (Special Issue on Internet Technology and Its Applications)
Category: 
Keyword: 
data consistency,  performance evaluation,  data synchronization version vector,  nomadic system,  replication,  scalability,  disconnected operation,  

Full Text: PDF(419.9KB)
>>Buy this Article


Summary: 
The optimistic consistency scheme has been established with respect to data consistency and availability in distributed systems. The nomadic data consistency model using version vectors to support data versioning for data synchronization and concurrent conflict detection is suitable for an optimistic replication system that supports large-scale wireless networks. This paper describes the architecture and its data consistency model using data versioning and its access domain control targeted for nomadic data sharing systems, such as collaborative works using database and messaging, and the data transfer optimizations of the model. We evaluate our data versioning scheme comparing with a traditional data versioning and the data transfer optimization by estimation and measurement assuming a mobile worker's job. We generate arithmetic formulas for data transfer estimation using the optimizing techniques and apply them to large-scale data sharing configurations in which collaboration groups are dynamically formed and data is exchanged in each group. The data versioning with an access domain increases flexibility in data sharing configurations, such as mobile collaboration systems and client/server type mobile systems. We confirmed that the combination of the general optimizations and the access domain configurations based on our data consistency model is applicable for large-scale mobile data sharing systems.