Supermicro Petascale All-Flash Storage SSG-122B-NE316R (E3.S)

The Supermicro SSG-122B-NE316R is a purpose-built 1U all-flash storage platform engineered around the E3.S NVMe form factor, delivering petascale raw capacity in a space-optimized rack footprint. The system accommodates 16 hot-swap E3.S NVMe drives connected via a PCIe Gen5 backplane, enabling aggregate sequential throughput and IOPS levels that satisfy the most demanding AI training, inference serving, and high-performance computing workloads. By adopting the E3.S form factor—designed specifically for data-center-class NVMe SSDs—the platform achieves superior thermal management and higher areal density compared to legacy U.2 or AIC configurations, making it a natural fit for dense storage clusters and composable disaggregated architectures.

At its core, the SSG-122B-NE316R is built on a dual-processor motherboard platform supporting 4th or 5th Generation Intel Xeon Scalable processors, providing sufficient compute headroom for inline compression, encryption, and storage protocol offload without external processing nodes. High-bandwidth DDR5 memory channels feed the NVMe subsystem with minimal latency, while the PCIe Gen5 fabric ensures each E3.S slot operates at full line rate. Dual high-speed OCP 3.0 network mezzanine slots accept 100GbE or 200GbE adapters, enabling direct NVMe-oF (NVMe over Fabrics) connectivity for fabric-attached storage disaggregation across large GPU clusters.

The platform is particularly notable for its role in AI data pipeline infrastructure, where parallel ingestion of multi-terabyte training datasets, checkpoint storage for large language models, and real-time feature store access require sustained throughput that spinning-disk or hybrid arrays cannot provide. Its 1U form factor allows operators to deploy petabytes of raw all-flash capacity within a single rack, dramatically reducing the physical footprint of storage tiers in GPU-dense AI factories and scientific computing environments. Supermicro's IPMI-based BMC with Redfish API support simplifies lifecycle management at scale.

Ideal for

• High-throughput AI/ML training dataset ingestion for GPU clusters requiring sustained multi-hundred GB/s read bandwidth to eliminate storage bottlenecks
• Large language model checkpoint storage and rapid restore, enabling fast fault recovery and iterative fine-tuning cycles without I/O stalls
• NVMe-oF fabric-attached storage disaggregation in composable HPC and AI factory architectures connecting hundreds of compute nodes
• High-frequency financial analytics and real-time risk modeling platforms demanding sub-millisecond latency at millions of IOPS
• Genomics and life-sciences sequence processing pipelines that require dense parallel random reads across petabyte-scale reference datasets
• Video rendering and media production storage backends for 8K multi-stream ingest, transcoding, and delivery workflows requiring deterministic low-latency access

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.