NVIDIA brings AI solution to minimize InfiniBand data center downtime

NVIDIA has recently announced a solution for minimizing downtime at the InfiniBand supercomputing data centers. Named the NVIDIA® Mellanox® UFM® Cyber-AI platform, it utilizes artificial intelligence based analytics for detecting any security threats as well as operational problems and predict network failures.

The company’s latest platform will provide alerts in case of abnormal behavior of the system and applications, possible threats to the system and failures, while also undertaking corrective measures. The solution also aims to deliver security alerts when there are attempts to hack the system for hosting undesired applications, including crypto mining.

The platform targets a reduction in data center downtime, an issue which, as per the research by ITIC, could lead to costs of over $300,000 per hour.

Reportedly, it is an extension of the UFM platform portfolio, being used to manage InfiniBand systems since almost a decade. It leverages AI for learning the network workload and operational cadence of a data center, using real-time as well as historic workload and telemetry data. Keeping this as a baseline, the platform can track the health and network modifications systemwide, detecting performance degradations, usage and changes to profiles.

Also Read: NVIDIA acquires Mellanox to enhance data center computing technologies

Companies which have been using the UFM platform have seemingly shown considerable interest in NVIDIA’s latest solution.

According to NVIDIA’s Senior VP of marketing for Mellanox networking, Gilad Shainer, the UFM Cyber-AI platform offered by the company detects the unique vital signs of a data center and leverages them for identifying degradation in performance, component failure and abnormal usage trends. It enables system administrators to determine and respond to potential security threats faster and address any upcoming failures, which will save cost and ensure consistent service is offered to customers, he noted.

The associate director of services and technology at NCI Australia, Allan Williams said that the institute plays a key role in the national research landscape. Its supercomputing infrastructure caters to around 5,000 researchers who are using it for critical national as well as global activities. He added that UFM allows NCI to manage its supercomputers effectively and optimize performance.

 

Source Credits: https://nvidianews.nvidia.com/news/nvidia-unveils-ai-platform-to-minimize-downtime-in-supercomputing-data-centers