UMD Researchers Create Low-Cost, High-Performing Cloud Storage Solution

Cloud data storage applications are very popular and are an increasingly important element of online services. As such, there is a growing demand for cloud data storage that is reliable, secure, and retrievable.

While the advantages of cloud storage and computing are very well known, there is one major limitation: data storage and updating are prohibitively expensive.

University of Maryland (UMD) researchers have developed a practical, low-cost, low-bandwidth Proof of Retrievability (PoR) scheme that guarantees customers better data security and authenticity, and allows clients to retrieve their data, along with proof that it has not been corrupted or manipulated.

PoR schemes allow customers to store files on a cloud server and enable them to access it and monitor them from time to time, but this service comes at the expense of using network resources. There are many existing PoRs that allow data retrieval, but the PoR scheme proposed by UMD researchers costs less and outperforms existing ones.

The scheme was developed by Department of Electrical and Computer Engineering Assistant Professor Charalampos Papamanthou and Associate Professor of Computer Science Elaine Shi, both jointly appointed in the UMD Institute for Advanced Computer Studies (UMIACS), and the late Emil Stefanov, a former Ph.D. candidate in the Department of Electrical Engineering and Computer Sciences at the University of California at Berkeley.

“[Using our cloud data scheme,] one can store data in the cloud and check that this data has not changed since the time of the upload, in a very efficient way. For example, to check that a 4 GB movie is stored in the cloud, only a few seconds are required,” he said. “Unlike prior work [where data would be uploaded once and never change], one can update data efficiently. This reflects better real-world applications.”

The new dynamic PoR requires 300 times less bandwidth than existing technologies and provides better data security.

The researchers call it the “light weight” dynamic PoR. While conventional schemes need 400 plus blocks to be transferred between the client and server for a single data access, the new scheme needs only 1.05 to 1.35 blocks.

Applications are huge, especially with storage outsourcing, Papamanthou said. “One can integrate this with any current client-side application that accesses cloud storage, like Amazon S3 or Google Drive. The new version of the software will essentially provide a ‘check’ button that will inform the client about the state of his data.”

According to Papamanthou, the client base is “anyone using cloud storage to store very important information and cannot afford to lose it or have it corrupted.”

Research on this new initiative continues and the next step is to optimize the scheme further by reducing the bandwidth, he said. It was quite a challenge to maintain an efficient check procedure while allowing the client to update their data, he added.

“The Office of Technology Commercialization [OTC] really helped us better articulate the contributions of our invention so that they could be understood by non-tech-savvy people,” Papamanthou said.

OTC conferred its Invention of the Year award to this invention last year.

Published December 14, 2015