On Cloud Storage and the Cloud of Clouds Approach

Many recently proposed cloud storage architectures build a single virtual cloud storage system by using a combination of diverse commercial cloud storage services - the so called cloud of clouds approach. Thereby, the data to be stored is dispersed among different (independent) cloud storage providers in a redundant way. This is commonly accomplished either by naively replicating the data to several providers (storing an entire copy of a file at each provider) or by dispersing suitably encoded data, i.e., only a certain threshold of file fragments is required for reconstruction of a file. Furthermore, since many vendors of commercial cloud storage services do not provide adequate means to securing the cloud from within the cloud infrastructure, many recently proposed cloud storage architectures (transparently) add relevant security and privacy features from the outside. In doing so, they are mainly trying not to affect the cloud provider's interfaces and inner workings. In this paper we take a closer look at distributed cloud storage systems. We provide an overview of information dispersal strategies to realise reliable distributed cloud storage systems and provide an overview of state-of-the-art cloud storage approaches. Then, we analyse them with respect to security properties. Furthermore, we discuss the lack of privacy features and in particular features to provide access privacy in existing distributed cloud storage systems, which is an important direction for future research on distributed cloud storage.