Formalizing and Checking Multilevel Consistency
Developers of distributed data-stores must trade consistency for performance and availability. Such systems may in fact implement weak consistency models, e.g., causal consistency or eventual consistency, corresponding to different costs and guarantees to the clients. We consider the case of distrib...
Saved in:
Published in | Verification, Model Checking, and Abstract Interpretation Vol. 11990; pp. 379 - 400 |
---|---|
Main Authors | , , , , |
Format | Book Chapter |
Language | English |
Published |
Switzerland
Springer International Publishing AG
2020
Springer International Publishing |
Series | Lecture Notes in Computer Science |
Online Access | Get full text |
Cover
Loading…
Summary: | Developers of distributed data-stores must trade consistency for performance and availability. Such systems may in fact implement weak consistency models, e.g., causal consistency or eventual consistency, corresponding to different costs and guarantees to the clients. We consider the case of distributed systems that offer not just one level of consistency but multiple levels of consistency to the clients. This corresponds to many practical situations. For instance, popular data-stores such as Amazon DynamoDB and Apache’s Cassandra allow applications to tag each query within the same session with a separate consistency level. In this paper, we provide a formal framework for the specification of multilevel consistency, and we address the problem of checking the conformance of a computation to such a specification. We provide a principled algorithmic approach to this problem and apply it to several instances of models with multilevel consistency. |
---|---|
Bibliography: | Partially supported by CEFIPRA DST-Inria-CNRS Project 2014-1, AVeCSo. |
ISBN: | 9783030393212 3030393216 |
ISSN: | 0302-9743 1611-3349 |
DOI: | 10.1007/978-3-030-39322-9_18 |