Personal tools
You are here: Home Research Scientific Computing on the Cloud

Scientific Computing on the Cloud

The research goal is to study the migration of scientific computing applications to the cloud and to reduce these applications and algorithms to cloud computing frameworks like the MapReduce.

The research goal is to study the migration of scientific computing applications to the cloud. To adapt resource-intensive scientific computing applications for the cloud, the applications must be reduced to frameworks that can successfully exploit the cloud resources. Generally, cloud infrastructure is based on commodity computers, which are cost effective, however are bound to fail regularly. This causes a serious problem as, the software has to adapt to failures and the best solution is to replicate the data and computation. One such framework that is built based on this idea is the MapReduce framework, which has gained popularity as a cloud computing framework on which one can perform automatically scalable distributed applications. MapReduce is well suited for simple, often embarrassingly parallel problems. MapReduce is used for a wide variety of problems like large-scale indexing, graph computations, machine learning and extracting specific data from a huge set of indexed web pages. MapReduce has also been tested for scientific problems. It performed well for simple problems like Marsaglia polar method for generating random variables and integer sort while it has significant problems with more complex iterative algorithms. This gives scope for further research, and we tried to reduce the iterative algorithms to better frameworks or to optimizations of MapReduce.

One of the main goals of this study is to adapt iterative and non-embarrassingly parallel algorithms to MapReduce model, while retaining their efficiency. Our research has shown that scientific computing algorithms can be divided into certain classes, based on how MapReduce model can be applied to them. The hypothesis is: if a whole algorithm cannot be represented as an execution of a single MapReduce model, could a combination of different parts of the algorithm implemented in different MapReduce model based frameworks (like Apache Hadoop, Haloop, Twister etc.) or other distributed computing models like for example Bulk Synchronous Parallel model, lead towards an efficient implementation on the cloud? The second goal of the study is to design a cloud computing framework as a platform for solving scientific computing problems, if the earlier hypothesis is true. The framework should support both large data processing and complex iterative algorithms with an emphasis on heavy calculations instead of the size of the input data. Recently we named the project “Stratus” and are in the process of defining the features of the framework.

Projects

People

Publications

2012

  • P. Jakovits, S. N. Srirama, I. Kromonov: Stratus: A distributed computing framework for scientific simulations on the cloud, The Fifth International Symposium on Advances Of High Performance Computing And Networking (AHPCN-2012), In conjunction with The 14th IEEE International Conference on High Performance Computing and Communications (HPCC-2012), 25-27 June, 2012, pp. 1053-1059. IEEE.
  • S. N. Srirama, C. Willmore, V. Ivanistsev, P. Jakovits et al.: Desktop to Cloud Migration of Scientific Experiments, 2nd International Workshop on Cloud Computing and Scientific Applications (CCSA) @ 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2012), May 13-16, 2012.
  • P. Jakovits, S. N. Srirama: Stratus: Scientific simulations in the cloud, The 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 13-16, 2012.
  • S. N. Srirama, P. Jakovits, E. Vainikko: Adapting Scientific Computing Problems to Clouds using MapReduce, Future Generation Computer Systems Journal, 28(1):184-192, 2012. Elsevier press. DOI 10.1016/j.future.2011.05.025

2011

  • S. N. Srirama, O. Batrashev, P. Jakovits, E. Vainikko: Scalability of Parallel Scientific Applications on the Cloud, Scientific Programming Journal, Special Issue on Science-driven Cloud Computing, 19(2-3):91-105, 2011. IOS Press. DOI 10.3233/SPR-2011-0320.
  • P. Jakovits, I. Kromonov, S. N. Srirama: Monte Carlo linear system solver using MapReduce, 4th IEEE/ACM International Conference on Utility and Cloud Computing (UCC 2011), December 5-7, 2011, pp.293-299. IEEE.
  • P. Jakovits, S. N. Srirama, E. Vainikko: MapReduce for Scientific Computing - Viability for non-embarrassingly parallel algorithms, The 14th International Parallel Computing conference (ParCo 2011), August 30-September 2, 2011. Published in: Advances in Parallel Computing book series, Volume 22: Applications, Tools and Techniques on the Road to Exascale Computing, Edited by: K. Bosschere, E. D'Hollander, G. Joubert, D. Padua, F. Peters, M. Sawyer, 2012, ISSN 0927-5452, pp. 117-124. IOS Press.
  • O. Batrashev, S. N. Srirama, E. Vainikko: Benchmarking DOUG on the Cloud, The 2011 International Conference on High Performance Computing & Simulation (HPCS 2011), July 4-8, 2011, pp. 677-685. IEEE.

2010

  • S. N. Srirama, P. Jakovits, E. Vainikko: Adapting Scientific Computing Problems to Clouds using MapReduce, International Conference on Utility and Cloud Computing (UCC 2010) in conjunction with the International Conference on Advanced Computing (ICoAC 2010), December 14-16, 2010. (Extended in FGCS journal)
  • S. N. Srirama, O. Batrashev, E. Vainikko: SciCloud: Scientific Computing on the Cloud, 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2010), May 17-20, 2010, pp. 579. IEEE Computer Society.