Founded in 1955, Nanyang Technological University (NTU) is a world-leading research university in Singapore. NTU’s School of Computer Science and Engineering (SCSE), ranked 1st in Asia by the Academic Ranking of World Universities (2017) for Computer Science and Engineering, is a leader in the field. To bolster its leadership and turbocharge multidisciplinary research, the school replaced its isolated data storage systems with a Huawei storage system, making it easier for professors and researchers in different areas to access and share information while boosting productivity and efficiency in performance of instructional tasks.
Technology challenges for cutting-edge cross-functional research
A clear pioneer in the field of computer science and engineering education, the SCSE was the first school in Singapore to offer a Computer Engineering program. To this day, SCSE is the only school in Singapore that trains students in both computer science and computer engineering fundamentals.
That explains why professors at the SCSE, together with their doctorate students and research staff, carry out advanced research in a wide spectrum of topics. They work in various research groups, for instance Hardware and Embedded Systems, Cyber Security and Forensics, Data Management and Analytics, and Computational Intelligence, just to name a few.
But the school’s legacy system only supported two types of clusters, which could not handle the kind of multidisciplinary research that the researchers needed. The limited computing resources also meant that there were consistently long pending job queues, hindering productivity.
To ensure researchers are able to efficiently and effectively leverage technology to access research materials, it is necessary that all documents are stored in a unified resource pool to enable information sharing and synergies across research disciplines.
Moreover, the school has complex and diverse research directions. Some of its programs require fat nodes (large memory per node) and high peak computing performance.
Given the collaborative nature of researchers in wide array of disciplines at the school, the researchers carry out tasks in vastly varying workload sizes at peak and off-peak times, with rapidly changing resource volumes. To accomplish these goals, the school requires a cloud-based Information Technology (IT) platform that supports flexible partitioning and rapid configuration.
Replacing many systems with one solution
The school began looking for a better way of doing things and decided to replace its multiple legacy systems with a single integrated solution that would offer a better experience for both professors and researchers.
This system must provide sufficient capacity and have the ability to be expanded at any time. In addition, the computing node must be able to support the running of large-memory serial programs, OPENMP programs, and Message Passing Interface (MPI) programs. Lastly, the new solution must be able to provide higher density, while occupying less physical space and be more energy efficient at the same time.
Huawei came onboard as the SCSE’s technology partner on this transformation journey and deployed the KunLun 9016 and OceanStor Storage system to form the basis of the hardware platform. Through implementing both physical and virtual partitioning to deploy different High-Performance Computing (HPC) applications, multiple groups could simultaneously use the system, thus conducting multidisciplinary research to maximum advantages.
Empowering a centralized resource pool
The school chose Huawei OceanStor all-flash storage and the KunLun platform over other offerings in the market to build a research resource pool. More and more research projects have begun to pool their computing and storage resources, allowing them the convenience of continually sharing and developing their research technologies.
Huawei’s KunLun platform provided one of the best solutions for the school’s needs, which can well support the running of large-memory serial programs, OPENMP programs, and MPI programs. KunLun is ideal for core databases, service application consolidation, in-memory computing, HPC fat nodes, and more scenarios.
Each KunLun server also runs multiple HPC services, including high-performing databases, machine learning, cloud computing, and big data services. The flexible combination of physical and virtual partitioning can adapt to small and large applications of different computing types, allowing for NTU’s easy and efficient use of a centralized resource pool.
The SCSE agrees the comprehensive sharing of research resources will become a cornerstone of its future talent development and scientific research. The improved efficiency of teaching and scientific research is poised to take the university’s research excellence to new heights.