GenAI workloads are unpredictable and must be high-performance and congestion-free. To achieve this, monitoring of the environment is critical and must be part of the solution.
With Augtera network AI, observability spans end-to-end from the GPU to the network fabric. Augtera is a light footprint application deployed within the environment, and it brings visibility into the environment and leverages AI to:
- Prevent and lower incidents and fabric degradation
- Maximize utilization while minimizing congestion
- Automate corrective actions through Ansible playbooks
![An image illustrating BeyondEdge Verity fabric view](/static/media/9198938f-8c47-5a0e-82d9-6db6a62cd3f7/DAM-22964249-e70a-45cc-9286-5efcd5f848d4/out/Picture21=GUID-FACC97EB-1483-43DE-BEC5-B3144D2E7963=1=en-us=Low.png)
In addition to providing visibility, Augtera's network AI also provides root cause, remediation, and impact analysis as a comprehensive offering.
![An image illustrating generation AI's real-time ECN and PFC observability heat maps](/static/media/9198938f-8c47-5a0e-82d9-6db6a62cd3f7/DAM-22964249-e70a-45cc-9286-5efcd5f848d4/out/Picture22=GUID-9D8CA2DB-2AAD-45D5-A0C3-5C3AAE067B82=1=en-us=Low.png)
Figure 20 shows the type of observability Augtera’s network AI can provide to an operator while implementing a GenAI workload achieving optimum performance.
The integration of Dell Enterprise SONiC with Augtera AI adds an intelligence layer as part of a solution across the different elements that make up the Ethernet powered infrastructure portfolio for GenAI workloads by Dell Technologies.