ИСТИНА

In this paper we present an approach for solving data envelopment analysis (DEA) models using parallel and distributed computing. DEA is an approach for the estimation of production units’ behavior. It is not only a tool for technical-efficiency analysis but also an increasingly popular performance management framework that provides a new approach to traditional cost-performance analysis, frontier estimation, decision making, and deriving insights from best practices. Application of DEA on various sets of problems generates a large number of independent optimization problems for every decision making unit. For computations we use Everest, a cloud platform for publishing, execution and composition of computing applications in a distributed environment. Everest follows the PaaS model and can be used remotely by multiple users to seamlessly run computations. The presented approach is demonstrated by running real-world applications on ad-hoc infrastructures consisting of resources of different types.