HIGH-EFFICIENCY COMPUTING SYSTEMS

Optimization of data flows in genomics data center using Lustre distributed file system

R. N. Arifulov^a, N. V. Popova^a, S. A. Naumenko<sup>ab

a</sup> M. V. Lomonosov Moscow State University, Department of Bioengineering and Bioinformatics, Moscow
^b Institute for Information Transmission Problems of the Russian Academy of Sciences (Kharkevich Institute), Moscow

Abstract: We present the architecture of the datacenter for processing of high throughput sequencing data. To improve the I/O performance we use specialized storage systems with hardware RAID controllers, distributed filesystem Lustre for load balancing among several data servers and cluster filesystem OCFS2 for dedicated volumes built on Fiber Channel infrastructure for high-load tasks.

Keywords: sequencing, cluster supercomputer, treating genomic data, Lustre.