Formal Modelling of Resilient Data Storage in Cloud

Abstract

Abstract. Reliable and highly performant handling of large data stores constitutes one of the major challenges of cloud computing. In this paper, we propose a formalisation of a cloud solution implemented by F-Secure -a provider of secure data storage services. The solution is based on massive replication and the write-ahead logging mechanism. To achieve high performance, the company has abandoned a transactional model. We formally derive a model of the proposed architectural solution and verify data integrity and consistency properties under possible failure scenarios. The proposed approach allows the designers to formally define and verify essential characteristics of architectures for handling large data stores.

Towards Integrated Modelling of Dynamic Access Control with UML and Event-B

“…Verification of data integrity and data consistency properties in Event-B framework has also been investigated in [15] in the context of cloud data base. Moreover, in [16,17] we verified by proofs correctness and safety of consistent updates of patient data records.…”

Performance challenge to resilience: A formal view

“…However, in a pursue of high performance, cloud solutions rarely rely on the transactional model and hence deliver weaker guarantees regarding data integrity. In my talk, I give a formal outlook on the resilience properties that can be guarantee with highly-performant solution [3].…”

Cost and Performance Modeling for Earth System Data Management and Beyond

“…Various studies model resilience based on the distribution of data across storage devices and strategy; for example, in [21], resilience depending on RAID levels is investigated and visualized in 2D heatmaps depending on error rates of memory and storage. In [19], a formal method of investigating resilience depending on data replication strategy and hardware is described and explored on several use cases.…”