Mobile cloud computing: A survey
Future Generation Computer Systems
| Hi-index | 0.00 |
This paper addresses difficulties in mapping blocks onto cloud storage and proposes a novel cost-aware log-structured block I/O management over one of the commercial cloud storage, Amazon S3. The proposed scheme is imbedded in the virtual USB drive architecture that replaces the NAND flash of the USB memory with the capacity-free network storage. The key of the proposed scheme is to perform onto the cloud storage log-structured writes with the optimal number of data blocks that adaptively changes with I/O characteristics (the number of I/O operations and storage size) and cloud storage pricing policy. The proposed scheme also efficiently manages associate metadata by using well-organized data structures and layouts both in the memory and on the cloud storage. Performance analysis shows that the proposed scheme can reduce the total I/O costs significantly up to 54%, as compared with a simple one-to-one mapping scheme.