NVIDIA Unveils Minerva to Bring End-to-End Observability for AI Factories
NVIDIA has launched Minerva, an AIOps platform meant to help teams find problems earlier and keep large AI data centers, which it calls AI factories, productive.
At that scale, a single degraded part, such as a GPU running hotter than its peers or an NVMe drive slowing under load, can slow an entire training job. Traditional tools split the view between networking, compute, reliability and machine learning teams.
Minerva connects signals across NVIDIA GPUs, networking, storage, schedulers and workloads into a dynamic knowledge graph and applies AI models to turn that data into actionable insight. It focuses on AI factory yield, measured through GPU hours, goodput, cluster uptime and token production.
Summary written by Readthrough from the original article. The full story, with photos and any updates, is at NVIDIA Newsroom.
