Raijin Supercomputer
In this video from the PBS Works User Group 2017, Rodger Edberg from the National Computational Infrastructure (NCI) presents: PBS Professional at NCI.
“As Australia’s national research computing service, the National Computational Infrastructure (NCI) provides world-class, high-end services to Australian researchers, including access to advanced computational and data-intensive methods, support, and high-performance infrastructure. NCI supports computationally-based research with a focus on the environment, climate and earth system science in particular. Since 2007, NCI’s infrastructure investments, which exceed $80M, have been provided by the Australian Government under its National Collaborative Research Infrastructure Strategy (NCRIS) and Super Science Initiatives.”
In production since June 2013, Raijin is the largest supercomputer in Australia and is among the top 30 systems in the world. One of the distinguishing requirements of this system is the wide spectrum of job types, including jobs that vary both in scale (from very large-scale jobs using a substantial portion of the cluster, to memory-intensive jobs with high I/O requirements) and in completion time (from long jobs on the order of several days, to very short jobs requiring fast turn-around). Overall system balance is important to ensure strong, scalable performance and deliver a high-quality user experience. Architecture and subsystems (such as memory, interconnect and filesystem) must scale as the software and hardware scale out, to protect application performance. In short, all elements of the system – application, hardware, subsystems and middleware – need to scale without impacting each other. Thus, for Raijin’s workload management system, NCI needed a highly scalable, flexible and reliable product that could handle both the size and complexity of its computing requirements.
