Distributed Virtual Private Disk for Cloud Computing

In recent years a new technological paradigm has emerged in the computer science research community. This paradigm is referred to as Cloud Computing, a new computing paradigm whose aims are to allow users to utilize computing infrastructure over the network, supplied as an on-demand service. There are already many commercial solutions based on this approach, as well as many academic research projects that are especially focused on resource management of an infrastructure that comprises computational power, storage space and communication networks. In particular, the demand of storage capacity is increasing, determined by, for example, scientific (e.g. physics experiments) and commercial (e.g. e-commerce sites, search engines) applications, that produce huge quantity of data. Building a data storage service that is pervasive, available, scalable and that can handle massive quantities of data has always been a priority for every distributed system paradigm and infrastructure. Cloud Computing has different goals and characteristics and thus poses new challenges in this area of research. Cloud Computing applications, especially data intensive ones, typically need storage services that provide large amount of storage capacity and the ability of retrieving stored data at any time and in any condition (e.g., infrastructure component failures). Moreover, services are hosted and data are located on third-party resources, a condition that poses a threat to data confidentiality. In many cases, applications or system-specific functions require the availability of raw block devices, for instance when a specific file system is necessary, or when the application needs to directly access physical storage (e.g., in the case of a DBMS). In this Thesis we propose ENIGMA, a distributed infrastructure that provides virtual disks by abstracting the storage resources provided by a set of physical nodes and exposing to Cloud Computing users, applications, and Virtual Machines a set of virtual block storage devices, that can be used exactly as standard physical disks. The important aspect of the work in this thesis is that we would like to make maximum use of resources while