Platform Efficiency Challenges and the OCP Opportunity Context Aware AI ML Agents on DC SCM f скачать в хорошем качестве

Platform Efficiency Challenges and the OCP Opportunity Context Aware AI ML Agents on DC SCM f

Raghu Kondapalli CTO - Axiado Corporation, sundaram Arumugasundaram Principal security research Architect - Axiado Corporation  Server-platform efficiency encompasses energy use, resource utilization, security, thermal management, workload distribution, and system health. However, recent platform studies highlight low infrastructure utilization across servers, racks, switches, and pods. GPU utilization is around 60%, with CPUs and networks similarly affected. Since post-deployment efficiency remains underemphasized, OCP is well-positioned to drive standardization. To improve ROI, the industry needs standardized metrics, real-time monitoring, and automated corrective systems. Context-aware resource management and continuous security assessments are essential. AI-ML solutions can synthesize contextual metrics for actionable insights and enable real-time platform adjustments. At 2024 OCP Brisbane, we shared a thermal management AI-ML model running on a DC-SCM, saving up to 50% of cooling power; Now, it predicts GPU energy for ML workloads. We also developed an LLM-based model for live security threat mitigation.