IPAM

This section describes IP Address Management (IPAM) at the rack/site level.

IPAM

During UPF registration, active IPAM instance reserves four address-ranges per UPF per DNN.

  • Range-1: Active cluster, nodemgr-1

  • Range-2: Active cluster, nodemgr-2

  • Range-3: Standby cluster, nodemgr-1

  • Range-4: Standby cluster, nodemgr-2

During normal operation, Rack-1/Site-1 handles UPF-register/release, address-allocate/release for subscribers coming up in GR-instance-1.

If Rack-2/Site-2 goes down, Rack-1/Site-1 gets role-change trigger for GR-Instance-2.

  • IPAM in Rack-1/Site-1, restores the content of GR-Instance-2 from local-cache-pod (which was already synced)

  • IPAM in Rack-1/Site-1 handles UPF-Register/Release and address-allocate/release for subscribers coming up with GR-Instance-2 using the restored content in addition to handling GR-Instance-1.

Each IPAM pool is associated to a GR-Instance, with the following:

  • Pool name is unique across all the instances.

  • Address-ranges are unique within VRF and across all the instances.

The same pool configuration must be configured in both the active and standby cnSGW-CSMF clusters of a particular instance.

During address-allocation, active instance assign free-IP from reserved address-range for the UPF.

Incase new address-ranges is not available, change ownership of standby’s address-range to current active instance and continue assigning address-ranges from it.