As Secure as Possible Eventual Consistency

• Type: Article in International Conference Proceedings Book
• Date: 2017

Authors

Shoker, A (Author) Other; Yactine, H (Author) Other; Baquero, C (Author) Other

Conference proceedings International

Title: PROCEEDINGS OF THE 3RD INTERNATIONAL WORKSHOP ON PRINCIPLES AND PRACTICE OF CONSISTENCY FOR DISTRIBUTED DATA (PAPOC 17)

Pages: 5:1-5:5

3rd International Workshop on Principles and Practice of Consistency for Distributed Data (PaPoC)

Belgrade, SERBIA, APR, 2017

Indexing

ISI Web of Knowledge - 0 Citations

Scopus - 4 Citations

Other information

Authenticus ID: P-00M-SVY

DOI: 10.1145/3064889.3064895

Abstract (EN): Eventual consistency (EC) is a relaxed data consistency model that, driven by the CAP theorem, trades prompt consistency for high availability. Although, this model has shown to be promising and greatly adopted by industry, the state of the art only assumes that replicas can crash and recover. However, a Byzantine replica (i.e., arbitrary or malicious) can hamper the eventual convergence of replicas to a global consistent state, thus compromising the entire service. Classical BFT state machine replication protocols cannot solve this problem due to the blocking nature of consensus, something at odd with the availability via replica divergence in the EC model. In this work in progress paper, we introduce a new secure highly available protocol for the EC model that assumes a fraction of replicas and any client can be Byzantine. To respect the essence of EC, the protocol gives priority to high availability, and thus Byzantine detection is performed off the critical path on a consistent data offset. The paper concisely explains the protocol and discusses its feasibility. We aim at presenting a more comprehensive and empirical study in the future.

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 5