Infortrend Launches Its Most Advanced U.2 NVMe SSD for AI Workloads

Why It Matters

The GS 5024U delivers unprecedented speed and scalability for AI training and inference, giving enterprises a competitive edge in time‑critical data processing and reducing total cost of ownership.

Key Takeaways

• 125 GB/s throughput, 2.4 M IOPS via PCIe 5.0
• Supports GPUDirect Storage for optimal GPU utilization
• Scales to 20 PB storage with HDD expansion
• Dual redundant controllers ensure uninterrupted AI workloads
• Automated tiering moves data to cost‑effective QLC SSD

Pulse Analysis

The launch of Infortrend’s EonStor GS 5024U marks a pivotal shift in enterprise storage for artificial‑intelligence applications. By leveraging a high‑performance Intel Xeon 6 processor and PCIe 5.0 connectivity, the system pushes raw throughput to 125 GB/s, dramatically shortening model‑training cycles. This performance leap, combined with 2.4 million IOPS, positions the GS 5024U as a direct competitor to traditional all‑flash arrays, while its support for NVMe‑oF and 200 GbE networking ensures low‑latency data movement across clustered environments.

Beyond raw speed, the GS 5024U’s architecture addresses the broader AI workflow through GPUDirect Storage and native Lustre compatibility. These features enable GPUs to access storage without CPU mediation, maximizing compute efficiency and reducing bottlenecks during inference and real‑time analytics. For organizations running high‑performance computing (HPC) or media‑rendering pipelines, the ability to deliver hundreds of gigabytes per second to parallel file systems translates into faster simulation results and higher‑quality output.

Scalability and cost management are equally critical. With up to 1.4 PB onboard and the option to expand to 5.6 PB on NVMe SSDs or 20 PB via HDDs, the GS 5024U accommodates growing data lakes without sacrificing performance. Automated tiering intelligently migrates completed AI models to lower‑cost QLC SSD or HDD tiers, preserving high‑speed storage for active workloads while optimizing total cost of ownership. This blend of speed, reliability, and economic flexibility makes the GS 5024U a compelling choice for enterprises seeking to future‑proof their AI infrastructure.