Abstract

Free-of-charge metacomputing aims at integrating surplus computing resources and utilizing their inter-connected computing power to fulfil computational demands at virtually no cost. The existing efforts on free-of-charge metacomputing can be observed in grid computing, parasitic computing and volunteer computing. As extensively discussed in the literature, these three metacomputing forms all have their respective challenges and shortcomings, ranging from sophisticated enabling technologies to possible frequent interruptions, not to mention the potential ethical and legal issues in parasitic computing. Based on our observation on the growing marketing strategy of offering cloud service samples (free quotas), we argue that it is also possible to follow a metacomputing approach to take advantage of free resources in the public cloud market. By applying this idea to our educational work, we gradually developed an implementation framework to facilitate exploiting free quotas of cloud user accounts. The relatively unique features and characteristics of cloud resource exploitation eventually turn our effort into a distinctive metacomputing form, and we name it bonus computing. Guided by the implementation framework, we initially verified bonus computing's effectiveness and efficiency by implementing a proof-of-concept (PoC) system over multiple cloud vendors. Then, we justified bonus computing's applicability by extending the PoC system to a Monte Carlo solution to a real-world problem in Astronomy. Based on our existing practices and the recent SLURM cluster experiments, we have tried to comprehensively analyze bonus computing's advantages and disadvantages against the other comparable metacomputing forms, which in turn strengthens our confidence in this work's contribution especially to the educational community.