The Institute of High Energy Physics held the first China Lustre workshop on November 26, 2011. 57 representatives of Lustre users from CAS institutes, foreign enterprises and Chinese universities attended this workshop, and over 100 people joined in the on-line meeting.
Scientific researchers and engineers shared their knowledge and experiences in current status and roadmap of Lustre, requirements and feedbacks, test results and comparison of Lustre file system, third-party monitoring tools, tutorials for Lustre newbies, etc.
In the last decade, with its excellent I/O performance, good scalability and persistent features of open source, hardware neutral and POSIX compatibility, Lustre has become the most popular distributed file system in the HPC domain. According to the latest authoritative statistic (Top500, June 2011), Lustre is used on at least 8 of Top 10 supercomputers, and 70% of Top 100 supercomputers today deploy Lustre.
In recent years, with the rapid development of HPC in China, more and more companies and institutes have joined the Lustre user groups , which include NASA, ORNL, LLNL, CEA, Supercomputing Center in Beijing, Tianjin and Shenzhen, etc.
As an indispensable part of IHEP’s cluster computing environment, Lustre file system supported a disc storage capacity of nearly 2 PB and a bandwidth of almost 24 GB/S for BESIII Experiment, DayaBay Neutrino Experiment, Yangbajin Cosmic Ray Observatory, ATLAS and CMS Experiment.Dr. Eric Barton, the designer of Lustre file system joined this workshop and made inquiries about Chinese users’ needs for Lustre and its application form with IHEP scientists. Both sides reached agreement in Lustre system R&D and future promotion in China.