Event - Aviz Networks | Aviz Networks
AI-Ready Networks: The Open Networking Advantage
Discover How Open Networking is Reshaping AI Infrastructure
Hyperscalers, GPU clouds, and forward-leaning enterprises are moving to open, disaggregated network architectures to meet AI’s unique demands. This expert panel breaks down the what, why, and how from SONiC-based fabrics to reference architectures so you can chart a practical path to AI-ready networking without vendor lock-in.
Setting the Stage Why Open, Why Now
- Where open networking fits for training vs. inference (front-end vs. back-end paths, QoS classes).
- How to balance capex/opex with performance, reliability, and scale-out agility.
- Operational patterns for Day 0–2 orchestration, observability, and incident response.
- AI pressure on networks: East-west heavy flows, bursty fan-in/out, strict SLOs.
- Disaggregation 101: Hardware choice + SONiC/abstraction layers + independent lifecycle.
- Lock-in vs. leverage: Procurement flexibility and upgrade velocity as strategic advantages.
- Open networking fundamentals: Flexibility, cost efficiency, multi-vendor resilience.
- Control-plane choices: BGP EVPN/VXLAN, fabric policy, and intent-based workflows.
- Lossless considerations: PFC/ECN tuning, queueing strategies, and traffic isolation for AI jobs.
- Hyperscalers & GPU farms: POD sizing, failure domains, oversubscription targets.
- Carriers & enterprises: Brownfield integration, migration waves, and risk controls.
- Day 0–2 playbook: Bring-up, tenant onboarding, SLO monitoring, and proactive troubleshooting.
- SONiC & abstraction layers: Platform independence and hardware enablement.
- Orchestration & CI/CD: Golden configs, validation pipelines, and change windows.
- Observability: Correlating fabric telemetry with GPU/NIC health and job-level metrics.
- Assisted operations: Ticket summarization, anomaly triage, guided remediation.
- Automation with guardrails: Policy-aware actions, RBAC, and audit trails.
- Measuring ROI: MTTD/MTTR reduction, change success rates, and utilization gains.
- Network architects and engineers designing AI fabrics.
- Platform/SRE leaders responsible for Day 0–2 operations.
- IT decision-makers evaluating open vs. vertically integrated stacks.
- A decision framework for open networking in AI environments.
- Best-practice guardrails for lossless fabrics and QoS.
- A roadmap to introduce AI-assisted NetOps with measurable ROI.