This section describes the NetQ integration with NVLink4. This integration supports the following:
- Domain management: create and manage multiple NVLink4 domains. After you create and configure a domain, run Global Fabric Manager (GFM) to collect telemetry data which can be visualized in the UI.
- Inventory management: manage your inventory of NVLink4 switches and GPU nodes, and view statistics and data for each device.
- API reference: access a Swagger instance to view NVLink4 API options.
- Fluentd message reference: example NVLink4 messages in JSON format for fluentd collectors.
Each GPU node and NVSwitch has a designated telemetry agent embedded in NVOS. This agent fetches telemetry data and streams it to a Fluentd data collector that integrates with NetQ or a third-party client.
Additionally, NetQ maintains GFM processes with high availability. If the GFM process stops unexpectedly, NetQ quickly and automatically remediates issues.
To get started, install NetQ.
Refer to the glossary for additional reference materials.