Industry Simulation Gateway on a Scalable Cloud

Large scale simulation experimentation typically requires significant computational resources due to an excessive number of simulation runs and replications to be performed. The traditional approach to provide such computational power, both in academic research and industry/business applications, was to use computing clusters or desktop grid resources. However, such resources not only require upfront capital investment but also lack the flexibility and scalability that is required to serve a variable number of clients/users efficiently. This paper presents how SakerGrid, a commercial desktop grid based simulation platform and its associated science gateway have been extended towards a scalable cloud computing solution. The integration of SakerGrid with the MiCADO automated deployment and autoscaling framework supports the execution of multiple simulation experiments by dynamically allocating virtual machines in the cloud in order to complete the experiment by a user-defined deadline.