Fabric intents

A fabric is a group of switches that are managed as a single logical unit. A single data center can include many complementary and mutually supporting fabrics.

With the Fabric Services System, you create a new fabric (or plan changes to an existing fabric) by designing and deploying a fabric intent: a detailed plan for a fabric's topology, whose node configuration files are deployed as a single transaction.

Each fabric exists within a region, a logical entity that stores configuration data common to the fabrics within it and that manages the orderly deployment of various types of intents. A fabric that is created within one region is not visible within, or available to, other regions.

After you deploy the fabric intent, the resulting fabric serves as the foundation for additional management templates that you can superimpose to manage the distribution of traffic across the fabric.

Note: The Fabric Services System enforces the use of a static management IP address for SR Linux nodes participating in managed fabrics (those consisting of nodes that are managed by the Fabric Services System)
  • This feature is always enabled for managed fabrics, and cannot be disabled.
  • Upon upgrade to Fabric Services System Release 24.12, all SR Linux nodes in managed fabrics will convert to the use of a static management IP address if they were not already so configured.
  • If a new fabric is not configured for automatic deployment, then a newly discovered SR Linux node that enters a Ready state will still use a DHCP management IP address; but upon deployment of the fabric, participating nodes will adopt a static management IP address and a default route.
  • Maintenance intents to upgrade software or replace a node conform to this requirement:
    • For a maintenance intent that upgrades a node's operating system, the existing static management IP address is retained after the upgrade.
    • For a maintenance intent that replaces a node, the static management IP address is released on the old SR Linux node, and the same IP address is then used for the replacement node.

Fabrics consisting of SR Linux nodes that are not managed by the Fabric Services System are not subject to this restriction, and will continue to use their existing, externally-configured setup for their management IP address.