Data Consistency as a Service (DCaaS)

Abstract

Ensuring data consistency over partitioned distributed database systems is a classical problem. Classical solutions proposed to solve this problem are mainly adopting locking or blocking techniques to ensure data correctness. These techniques are not suitable for cloud environments as they produce terrible response times due to the long latency and faultiness of Wide Area Network (WAN) connections among cloud datacenters. To overcome this problem, this paper proposes an inventory-like approach for ensuring data consistency over WAN connections that minimizes the number of exchanged messages over the WAN to enhance response times. As maintaining data consistency is a costly process, we propose to use different levels of data consistency for data objects, as not all data objects have the same importance. Hence, strong consistency will be used only for data objects that are crucial to the correctness of application operations. To save application developers from the hassles of maintaining data consistency, we propose to have a new platform service (i.e. Data Consistency as a Service (DCaaS)) that developers invoke to handle their data access requests fulfilling their different consistency requirements. Experiments show that proposed data consistency approach realized by the DCaaS service provides much better response time when compared with classical locking and blocking techniques.