DOI: 10.5176/978-981-08-5837-7 181
Authors: J.Eccles and G.Loizou
Given the current definitions of cloud computing, it is proposed that in order to arrive at the current deployment of abstraction classes in a cloud architecture, it is required to redesign the cloud such that the model of the cloud architecture consists of sets of abstraction classes. These classes are integrated on a loosely-coupled basis. It then becomes essential that the modelling of the cloud architecture and, where applicable its simulation, are implemented on a consistent basis. The meaning of each abstraction class is enhanced through the use of operational semantics via sets of specific XML like modules (metadata) derived from the relevant layers in the cloud architecture model. This is realised via the associated use of a predicate logic paradigm. As aconsequence, the integration between such abstraction classes is modelled by dataflows and implemented via metadata which provides a level of control through policy-based interaction. It is then shown how a model of the cloud architecture that gives the currently-recognised characteristics of cloud computing can be produced using this method of design, rather than have a cloud based purely on size, number of resources, degree of processing or complexity. Thus the functional paradigms of a cloud can be modelled via abstraction classes without the requirement for a deep understanding of how, or necessarily where, those functional paradigms may be implemented.
Additionally, this paper shows how a cloud model architecture that employs abstraction classes throughout the system can enable the integration of new areas of computation, such as ubiquitous computing, into the paradigm of cloud computing. In principle this is done through the abstraction of the new area and its subsequent integration into the cloud compute model by employing the correct predicate logic paradigm.