Backing Up Your Data to the Cloud: Want to Pay Less?

In this paper we present CAB, a Cost-Aware cloud Backup system, designed for users owning multiple connected devices. Putting client-side monetary cost reduction first, CAB allows users to access their data via multiple devices and applies file updates to the cloud backup with low time delay. We describe some of the optimizations necessary to achieve cost reduction, including data de-duplication, a log-structured data layout, garbage collection, and novel metadata management. Using two one-month long real-world traces, evaluation of a prototype implementation of CAB shows it reduces cost by 16 - 44% compared to a cost-oblivious baseline backup system. Our current prototype implementation stores data in Amazon S3, but is portable to different cloud providers that support REST protocol.