A framework for evaluating storage system security

TLDR

Data security in storage systems is achieved by encrypting data on the wire or on disk, yet no common parameters exist to compare these approaches. The paper proposes a framework to evaluate both encrypt‑on‑wire and encrypt‑on‑disk systems along security and performance axes. The framework is applied using a trace from a time‑sharing UNIX server to quantify the costs associated with each secure storage system. All existing systems trade off along a single continuum of security primitives, and encrypt‑on‑disk systems provide both increased security and improved performance over encrypt‑on‑wire in the traced environment.

Abstract

There are a variety of ways to ensure the security of data and the integrity of data transfer, depending on the set of anticipated attacks, the level of security desired by data owners, and the level of inconvenience users are willing to tolerate. Current storage systems secure data either by encrypting data on the wire, or by encrypting data on the disk. These systems seem very different, and currently there are no common parameters for comparing them. In this paper we propose a framework in which both types of systems can be evaluated along the security and performance axes. In particular, we show that all of the existing systems merely make different trade-offs along a single continuum and among a set of related security primitives. We use a trace from a time-sharing UNIX server used by a medium-sized workgroup to quantify the costs associated with each of these secure storage systems. We show that encrypt-on-disk systems offer both increased security and improved performance over encrypt-on-wire in the traced environment.