Adapting Scientific Applications to Cloud by Using Distributed Computing Frameworks

Scientific computing is a field that applies computer science to solve scientific problems from domains like genetics, biology, material science, chemistry etc. It is strongly associated with high performance computing (HPC) and parallel programming fields as scientific computing typically utilizes large scale computer modeling and simulation and thus requires large amounts of computer resources. Public clouds seem to be very suitable for solving scientific computing problems, but they are often built on commodity hardware and it´s not simple to design applications that can efficiently utilize large amounts of computing resources. This paper gives an overview of a study that researches the use of distributed computing frameworks like MapReduce to greatly simplify solving scientific computing problems in the cloud and compares how well the results measure up to the current de facto standard practices of the distributed computing field.