Cluster Computing Analysis Based on Beowulf Architecture
Purnawansyah Purnawansyah, Ramdan Sastra · International Journal of Computing · 2016
The need for computing speed in data processing is in high demand indicated by the increasing data processing in many companies. Aviation companies, for instance, show the growing number of passengers daily. The data processing for a big number of customers requires supercomputer technology at the cost of exponential fund. Therefore, the technology able to process big data with low cost is necessary. This research has improved a cluster computer which can process data with a big capacity processing that is generally cannot be processed using one computer. Cluster computer is built using Beowulf cluster architecture using Linux Debian operating system and SSH, NFS, MPI library and GANGLIA services. The result of this research is a cluster computer prototype which is able to process 38.000x38.000 matrix calculation data. In the first testing, the computer was able to work for a maximum of 30.000x30.000 matrix calculation whereas cluster computer can process up to 38.000x38.000 matrix size. This result indicates that the designed cluster computer is successful to calculate big data with a low computer specification. The further improvement can be applied in a more complicated computer calculation process and bigger data.