The National Cheng Kung University (NCKU) is a public research university located in Tainan, Taiwan. NCKU has consistently been ranked as one of the top universities and is best known for its schools of engineering, computer science, medicine, and planning and design in Asia.
NCKU’s Computer and Network Centre of National Cheng Kung University was tasked with both the deployment and operation of its school’s network and the Taiwan Academic Network (TANet) in Tainan.
The centre was also designated as a ‘gigabit point of presence”’(GigaPOP) for the Taiwan Advanced Research & Education Network (TWAREN). In this capacity, it had to provide NCKU’s faculty and students, as well as other schools and research institutes in Tainan, access to the university’s academic and research network resources.
When building the centre, NCKU aimed to create an AI-enabled, cloud-based virtual learning platform that would provide AI courses for primary and secondary school students.
“In the future, AI applications will be as prevalent as PCs are today. This cloud-based AI classroom will be beneficial to the development of the AI industry and students planning their future careers,” said Professor Chiang Jung-Hsien, Director of the Centre.
“While the platform was intended mainly for primary and secondary school students, it is also the first AI learning resource available for students undertaking their basic education in Taiwan. It aims to inspire students and encourages them to learn about innovative technologies. We hope to inspire students to engage in the research and study of AI, and make proper use of the value that this platform offers,” he added.
The challenge
Designing a series of AI courses required first establishing a solid digital infrastructure for hands-on programming and model training. The content also needed to be enriched and available for both theory and practical class lessons.
The current AI curricula were designed for higher education; making it more challenging to establish an AI education environment for young students at the primary and secondary levels. On top of that, there was also a lack of models for course planning and teacher training. Besides this learning platform, the project team had to spend additional effort on curricula and teaching materials; and with the support of experts, teacher training was designed to ensure that courses more closely reflect real industry applications.
When it came to implementation, NCKU faced challenges from the lack of skilled staff and knowledge. The university needed a professional team that could not only build a platform for high-performance computing in AI applications, but who also possessed the requisite knowledge and expertise in the basics and higher theories of AI and machine learning.
The solution
Logicalis Taiwan was able to plug these gaps. Logicalis Taiwan staff had the in-depth knowledge of algorithms and the computing library necessary to conduct a detailed assessment of the school’s needs. They provided a proof of concept to assure NCKU that the solution met the requirements of the students, faculty, and staff.
Beyond designing and delivering an integrated high-performance computing infrastructure with a cloud VDI (Virtual Desktop Infrastructure), Logicalis Taiwan developed a tailor-made AI course and provided NCKU with the practical know-how to inspire students to pursue advanced degrees and research opportunities in the AI field.
Logicalis Taiwan was also able to utilise its strong relationship with NVIDIA and VMware to identify NCKU’s unique needs, and leveraged its global experience to meet the POC standards.
The Logicalis team was able to advise on the teaching model, the class size, and the number of teachers needed, ensuring that students receive the highest teaching quality over remote connections on the NCKU AI learning platform.
Logicalis Taiwan further suggested consolidating NVIDIA’s vGPU and VMware’s virtualisation solutions to build a containerised computing infrastructure and demonstrate the value of the solution.
The all-flash SSD from Pure Storage for datasets and modelling used in the solution can achieve more than 200,000 I/O operations per second (IOPS), meaning that computing resources can be evenly distributed while maintaining resources for backup and recovery.
Built on the reliability and flexibility of the Cisco Unified Computing System (UCS) server, the AI-enabled, cloud-based virtual learning platform delivers AI model computing, training, and inferencing workloads while supporting logins for up to 50 teachers and students, unhindered by any limitations in distance or geographic location.
Logicalis Taiwan also provided one-on-one training for faculty and staff on how best to maximise the use of AI technology and drafted support documentation for the NCKU IT team. Leveraging its expertise, IT design, and execution capability in higher education, Logicalis Taiwan made it possible for the centre to implement solutions that exceeded NCKU’s requirements for an AI-learning, remotely connected campus.
