The University Research Facility in Big Data Analytics (UBDA) is a university-level research facility at the Hong Kong Polytechnic University, established to provide advanced computing infrastructure for academic research and innovation in data-intensive disciplines. UBDA supports a wide array of research activities across diverse fields – including engineering, health sciences, business, finance, social sciences, and design, by offering high-tier computing resources and technical services that are specifically configured to meet the demands of data-driven investigation, modelling, and simulation.
UBDA operates two distinct computing infrastructure, each designed to serve a wide range of research needs.
- A traditional High-Performance Computing (HPC) platform, suited for parallel simulations, large-scale data processing, and compute-heavy workloads across multiple nodes. Users operate under Linux environments and manage job execution through scheduling system such ass PBS and SLURM, with software installed at the user-level.
- A GPU-enabled Virtual Machine (GVM), which offers remote desktop-style environments equipped with up to 8 NVIDIA GPU cards and 168 virtual CPU cores for AI model development, deep learning and flexible application deployment. These environment provide root-level access and user-friendly interface, allowing researcher to customize workflows and software stacks
This dual offering allows PolyU researcher and students to choose the infrastructure that best matches their backgrounds, technical skills, computational needs and workflows preferences, whether they need the capability of distributed batch processing or the flexibility of customizable virtual environment for iterative model, analysis and application deployments.
UBDA Services
UBDA offers a comprehensive suite of advanced computing and support services to meet the evolving needs of PolyU staff and research students:
- GPU- and CPU-accelerated computing clusters for distributed batch processing across multiple nodes
- GPU-enabled virtual machines for AI model training, deep learning, and simulation workloads, with full root-level access
- Bulk NVME and HDD data storage, including NFS volumes backed by a resilient, self-healing file system for secure handling of large-scale research datasets
- Advanced networking resources, including PolyU IP allocation and TCP port accesses, enabling API deployment and remote application hosting
- 24/7 secure access to all computing resources via Campus network and supported protocols
- Rackmount server and GPU card hosting, with administrative and maintenance support for project's Principal Investigators (PIs)
- Technical consultancy and user support, assisting research teams in optimizing performance, configurating environments, and troubleshooting workloads
UBDA also provides Software-as-a-Service (SaaS) platforms, including JupyterLab Online, which offer browser-based, ready-to-use environments for data analysis, machine learning, and visualization. These platforms are preconfigured with popular frameworks such as TensorFlow and PyTorch and enable rapid prototyping and experimentation without requiring local setup - similar to commercial services like Google Colab but specifically optimized for use within the UDBA computing environment.
UBDA In Context
Compared to traditional University HPC Clusters such as HKU’s HPC2021, CityU’s Burgendy, and industry offerings like HK Cyberport or HK Science Park which primarily feature DGX SuperPODs (DGX H800 Cluster) optimized for large-scale AI, UBDA provides greater flexibility and accessibility for academic researchers and project staff. With two distinct infrastructures, users can choose between high-throughput HPC clusters for simulation-heavy tasks or GPU-enabled virtual machines (GVMs) for rapid experiment in cloud-style environments. UBDA's GVMs support customized workflows, API deployments, and application hosting, all within a remote-desktop interface and with root-level access, making them especially suitable for iterative modelling, data analysis, and research application development.
Mission and Impact
UBDA was established to address the growing demand for advanced computing resources and data analytics expertise across diverse academic disciplines. Through Its infrastructure and services, UBDA empowers researchers to uncover hidden patterns, refine predictive models, and enable evidence-based decision-making, advancing innovation and insight across fields such as engineering, health sciences, business, finance, social sciences and design.
UBDA also plays a key role in fostering cross-disciplinary research, teaching, and industry collaboration by providing a centralized platform where researchers, educators, and external collaborators from different domains can converge around shared data challenges and computational needs. By enabling access to scalable resources and common tools, UBDA facilitates the knowledge exchange, support joint research initiative, and encourages methodological innovation across the PolyU research community and beyond.
