Meet Aviz at ONUG AI Networking Summit London | Aviz Networks
Registration Closed
AI Networking Summit 2024 – London – ONUG
December 10, 2024
Evaluate AI for Your Network — Networks Made for AI & AI Made for Networks
AI is reshaping how modern networks are designed, deployed, and operated. This session walks through a pragmatic evaluation framework across two dimensions: Networks made for AI (building AI-ready fabrics) and AI made for Networks (applying AI/ML to network operations). You’ll leave with concrete steps to pilot, measure ROI, and scale—while maintaining control over data, models, and innovation velocity.
Key Outcomes
- AI Fabric Design Basics: Separate front-end / back-end paths, differentiated QoS classes, lossless transport (PFC/ECN), and GPU/NIC awareness in the topology.
- Open, Vendor-Agnostic Building Blocks: Choosing interoperable ecosystems across GPUs, switches, DPUs, and front-end infrastructure to avoid lock-in and enable faster upgrades.
- Reference Architectures vs. Open Networking: When to adopt a prescriptive RA (e.g., Spectrum-X) and when a multi-vendor SONiC-based fabric provides better flexibility.
- Capacity & Failure Domains: POD sizing, oversubscription targets, traffic isolation, and blast-radius containment for training vs. inference workloads.
- Observability Requirements: Correlating fabric telemetry with GPU/NIC health, flow-level analytics for job SLOs, and intent-drift detection as the environment scales.
- Start Small, Prove ROI: Narrow, high-impact use cases—ticket summarization, config assistance, anomaly triage, and guided troubleshooting—before expanding to closed-loop automation.
- Model Strategy & Data Control: Options for private, open-source LLMs; retrieval-augmented generation (RAG) with network context; guardrails, RBAC, and audit trails for safe automation.
- MLOps for NetOps: Versioning prompts/policies, evaluating model changes, and measuring outcomes such as MTTD/MTTR reduction and change success rate improvements.
- Scale-Out Adoption: From pilot to production—governance checklists, success metrics, and a roadmap to expand AI capabilities without sacrificing reliability.
- Hands-On Walkthrough: See how an AI-ready fabric is orchestrated, and how AI assistants can accelerate operations from Day 0 bring-up through Day 2 troubleshooting.
- 1:1 Deep Dives: Visit the booth for design reviews, migration plans, and tailored ROI scenarios.
- A clear framework to evaluate AI readiness across design, operations, and governance.
- An actionable pilot plan with measurable milestones and expected ROI.
- Guidelines to retain control of data, prompts, and models while scaling AI adoption.
- A decision path between reference architectures and open, multi-vendor fabrics.