Brief Announcement: Modeling MapReduce for Optimal Execution in the Cloud

Date Added: Jul 2010
Format: PDF

Cloud services are being increasingly used to outsource computations. But as cloud services become more popular, the variety of services that are offered is becoming over-whelming. In this paper, the authors aim at providing a model based on linear programming for cloud computations that enables customers to make an optimal choice of which re-sources to allocate, e.g.., one that minimizes monetary costs. This can then be integrated into a system that automates resource allocation. This paper focuses on modelling a restricted type of computations, namely MapReduce jobs. This is not only a relevant programming paradigm that is increasingly used for large-scale computational jobs, but will also offer one's a starting point from which they can generalize their techniques.