Osaka University and NEC Construct RED-ONION Data Transfer Platform

Osaka University's D3 Center, the Institute of Scientific and Industrial Research, and NEC Corporation have partnered to build "RED-ONION" (Research EnhanceD Osaka university Next-generation Infrastructure for Open research and innovatioN). The dedicated platform enables secure, bidirectional transmission of large-scale, sensitive research datasets—such as genomic sequences and clinical data—between generating laboratories and central computing infrastructure on the university's Suita Campus, with trial operations beginning in October 2026.

Dedicated Network and Supercomputing Integration
RED-ONION connects experimental instruments directly to the university data center over a dedicated 100-gigabit-per-second (Gbps) optical fiber line. The infrastructure interfaces directly with high-performance computing (HPC) clusters, including the SQUID and OCTOPUS supercomputers and the mdxII cloud computing platform located at the D3 Center, allowing rapid AI processing and simulation analysis with immediate return of processed results to researchers.

The system will support high-throughput data pipelines, including the "nanogap biomolecular sequencer" development led by Professor Masaki Taniguchi under the Cabinet Office's Economic Security Important Technology Development Program (K Program).

Parallel Transfer Engine Architecture
In standard network transfers, server-side data processing bottlenecks frequently limit the utilization of high-bandwidth optical connections. RED-ONION addresses this by implementing the NEC Ultra-high-speed Data Transfer System (NEC UDTS), which coordinates network and server data handling through a dedicated transfer engine that executes concurrent input/output processes in parallel.

Verification benchmarks confirm the architecture can complete a 1-terabyte (TB) disk-to-disk transfer in approximately 90 seconds over a 100 Gbps dedicated link.

Trial operations scheduled for October 2026 will evaluate operational parameters, computing resource allocations, and supported data profiles ahead of potential wider deployment across additional academic and medical research institutions under NEC's BluStellar framework.

Additional Context
This section details technical specifications not included in the original news release.

Conventional Wide Area Network (WAN) and local disk-to-disk data transfers often suffer performance degradation due to TCP window scaling limits, kernel context-switching overhead, and non-optimized storage I/O subsystems. High-throughput data engines overcome these limitations by utilizing zero-copy data pathing, multi-threaded asynchronous I/O, and Remote Direct Memory Access (RDMA) over Converged Ethernet (RoCE) or InfiniBand protocols, allowing data blocks to bypass intermediate operating system buffers during transit.

At 100 Gbps line rates, transferring one terabyte in roughly 90 seconds corresponds to an effective sustained throughput of approximately 88.8 Gbps (around 11.1 gigabytes per second), which closely approaches the physical line rate limit after deducting Ethernet framing, IP packet header overhead, and storage controller latency. Storage endpoints in these HPC workflows typically leverage parallel file systems—such as Lustre, IBM Spectrum Scale (GPFS), or BeeGFS—interfacing with non-volatile memory express (NVMe) solid-state drive arrays to prevent read/write bottlenecks during sustained multi-gigabyte-per-second ingestion cycles.

Edited by Romila DSilva, Induportals Editor, with AI assistance.