AI infrastructure requires efficient, low-latency, scaled up and scaled out manageability solutions, ensuring reliability and performance across critical systems like compute and storage nodes, networking devices, rack managers, CDUs, PDUs, etc.
Currently, these components are managed using firmware and software solutions with fragmented designs and vendor-specific implementations, which poses challenges for integration, security and limits the full potential of a comprehensive AI infrastructure management stack.
This session proposes a unified, open-source firmware-based approach that leverages OpenBMC across various management controllers to standardize interfaces/protocols, extend device/component support, aggregate management data, and facilitate efficiency/interoperability in infrastructure management.
Tuesday August 5, 2025 3:30pm - 3:50pm GMT+07 TaiNEX2 - 701 H
