Storage networks have become critical components of computing environments and are evolving into complex, networked and distributed storage models. With the growth in the number of organizations resorting to electronic data and on-line access, there has been a large increase in the amount of sensitive data stored on-line. Furthermore, this data has to be shared, replicated, and kept on-line. As a result, storage systems are becoming more vulnerable to security breaches, which can result in damaging losses. While regulatory requirements mandate data security, many recent incidences of data theft clearly show that that existing theoretical solutions are often not deployed and deployed systems are not secure. Currently, our research on storage security focuses on security and trustworthiness of SSP (Storage Service Provider), temporal access control, and secure, flexible and efficient cross-domain global file sharing.
Security and Trustworthiness of SSP Recent trend of outsourcing data storage and management operations to third party SSP has raised many new privacy concerns: while the SSPs can be trusted to store and backup the data, they cannot necessarily be trusted to keep the data confidential, or correctly charge the consumers for their utilized service. A simple solution for ensuring confidentiality is to encrypt data using a key known only to the client, thus shifting key management and accompanying responsibilities completely to the client. The problem becomes significantly harder when the client must share the data with other clients. Our research on group key management answers some of these questions (such as file sharing problem). In order to prevent inflation of consumer usage by SSPs or deflation of usage by consumers, and to enhance trust between them, we have designed a novel system called Saksha, a secure accounting system that enables automated and verifiable metering of the resources utilized by the consumers. A provider that includes Saksha as a part of its storage service can prove to its customers the amount of resources utilized by them.
Timed-Release Cryptography One of the most important aspects of data security is the notion of access control. Many different types of access control paradigms have been discussed in literature, each with their own properties and contributions. This project investigates the “temporal access control model” for storage systems. Adding time restrictions in storage systems gives access control mechanisms new rich capabilities: 1) in a military or covert operation setting, certain orders should be released only at specific time, 2) it can be used for patients to allow medical doctors to access their medical records only for certain periods of time. As a preliminary work on temporal access control, we have investigated a problem called timed-release cryptography, which allows a sender to encrypt a message so that only the intended recipient can read it only after a specified time. This work formalizes the concept of a secure timed-release public-key cryptosystem and showes that, if a third party is relied upon to guarantee decryption after the specified date, this concept is equivalent to identity-based (ID) encryption; this explains the observation that all known constructions use ID-based encryption to achieve timed-release security. We have provided several provably-secure constructions of timed-release encryption: a generic scheme based on any identity-based encryption scheme, and two more-efficient schemes based on the existence of cryptographically admissible bilinear mappings.
Secure, Flexible and Efficient Cross-Domain Global File Sharing There is a rising trend of collaboration and global sharing of information across multiple domains. For example, consider a group of scientists who want to share files generated by their simulation applications with scientists from a different organization to allow them to analyze their data and share the knowledge gathered through those results. In this example, local users need to freely share data and collaborate with remote users. In addition, users need high performance data access. Existing cross-domain file sharing systems require administrative interference which introduces delay and reduces the flexibility of the system. Along with one of my students, I have designed and implemented SGFS - a secure global file sharing system satisfying various design requirements: efficiency for high performance data access, flexibility of cross-domain file sharing without administrative interference, support for flexible policies and off-the-shelf policy managers, ability to be deployed in diverse environments, ease of management and low administrative overheads. Unlike existing systems satisfying only some of these requirements, SGFS is designed to satisfy all of these requirements. Contrary to the common belief, the SGFS revocation mechanism shows that maintaining revocation lists will not impose significant overhead.
coreFS Existing network file systems such as NFS or SFS often involve steep learning curve, which makes it difficult for security researchers or college students to learn network file system. To address this need, we are building coreFS. The goal of this file system is to give file system developers some form of basic distributed file system, which can be later modified as per the implementor’s requirement. Most of our storage security projects are being built on top of coreFS. Since it was released open-source on Mar. 2007, it has been downloaded more than 90 times
from 21 countries over the world.
DISC Our research on storage security is closely associated with DISC (DTC Intelligent Storage Consortium), where leading companies in storage industry are involved with research and development of new storage architectures. I maintain close collaboration with DISC faculty as well as storage experts from industry, which helps us evaluate and enhance the practicality and efficiency of our solutions. For example, we have provided an open-source reference implementation of the ANSI T10 object based storage interface standard.