Help us improve your experience.

Let us know what you think.

Do you have time for a two-minute survey?

 
ON THIS PAGE
 

Anchor PFEs and Redundancy in Junos Multi-Access User Plane

Understanding the Anchor PFE

An anchor PFE is the Packet Forwarding Engine (PFE) on a standard line card that has no direct interface connections, but rather provides the core processing of data traffic by doing the following:

  • Encoding/decoding of GTP-U packets. The anchor PFE decodes GTP-U packets from eNodeBs and forwards them to the core network and encodes IPv4 packets from the core network and forwards them to eNodeBs.

  • Enforces class of service and firewall filter rules on subscriber sessions.

  • Collects statistics on data usage for charging/accounting purpose.

Following are important points to consider when setting up anchor PFEs:

  • You must configure at least one anchor PFE line card. We recommend at least two with 1:1 hot-standby redundancy.

  • Each anchor PFE requires a defined pfe- interface of the form pfe-x/y/z.

Configuring No Redundancy for the Anchor PFEs

When no redundancy is required, all anchor PFE interfaces are equally available. The SAEGW-U uses all anchor PFE logical interfaces to anchor sessions/bearers. The routing engine (RE) of the SAEGW-U steers the GTP-U traffic for sessions and bearers to each of the anchor PFEs. The GTP processing for sessions and bearers and filter processing happens on the respective anchor PFE. The charging data is also maintained, collected and reported from the anchor for each session and its bearer.

When there is no redundancy configured, a failure of the anchor PFE line card is catastrophic for the SAEGW-U sessions/bearers in that control plane sessions corresponding to the failed anchor PFE and its data plane are lost. If supported by the SAEGW-C, the SAEGW-U can send an Sx Session Set Deletion Request for the lost sessions through the Sx interface, and the sessions are flushed in the SAEGW-C. All charging and other accounting data is lost for the sessions and bearers. New sessions can come up on the failed anchor PFE interface only when all sessions are flushed in the SAEGW-C for the failed anchor PFE, even if the anchor PFE comes up sooner. If other anchor PFE interfaces are available, new sessions can come up instantly on those anchor PFE interfaces.

Following is a typical configuration for two anchor PFEs with no redundancy.

  1. Configure slot 1 and slot 2 for anchor processing.
  2. Configure interfaces in slot 1 and slot 2 for PFCP processing.

Configuring 1:1 Hot-standby Redundancy for the Anchor PFEs

To have 1:1 PFE redundancy, an aggregated anchor PFE group can be formed as below using exactly two PFE logical interfaces from different slots:

  • Aggregated Anchor PFE group 1 – pfe-0/0/0 (primary), pfe-1/0/0 (secondary)

  • Aggregated Anchor PFE group 2 – pfe-0/1/0 (primary), pfe-1/1/0 (secondary)

You cannot have primary and secondary anchor PFEs on the same line card. For example, the following combination is not supported:

  • Aggregated anchor PFE group 1 – pfe-0/0/0 (primary), pfe-1/1/0 (secondary)

  • Aggregated anchor PFE group 2 – pfe-1/0/0 (primary), pfe-0/1/0 (secondary)

We also do not recommended configuring anchor PFEs on two separate line cards with their secondary anchor PFEs on just one line card. For examle:

  • Aggregated anchor PFE group 1 – pfe-0/0/0 (primary), pfe-2/0/0 (secondary)

  • Aggregated anchor PFE group 2 – pfe-1/0/0 (primary), pfe-2/1/0 (secondary)

When aggregated anchor PFE configuration is used, both the primary anchor PFE and secondary anchor PFE have the session state. But the routing engine (RE) steers the GTP-U traffic for sessions and bearers only to the primary anchor PFE. The GTP processing for sessions and bearers and filter processing happens on the primary anchor PFE. The charging data is also maintained, collected and reported from the primary anchor PFE. The secondary is in hot-standby mode and is ready for takeover only in the event of primary anchor PFE failure.

Given the considerable load that a single anchor PFE linecard can need to handle, a single anchor PFE linecard is limited to a maximum of two redundancy groups. You can configure a single anchor PFE for one of the following roles:

  • Dedicated primary for one redundancy group

  • Dedicated secondary for one redundancy group

  • Primary for two redundancy groups

  • Secondary for two redundancy groups

When 1:1 redundancy is operational, the redundancy interface process monitors the health of the primary and secondary anchor PFEs.

A secondary anchor PFE failure results in zero data plane traffic loss on the primary anchor PFE. All active sessions remain unaffected. New sessions can come up without any latency. When the secondary anchor PFE is restored, there is a catchup phase to program the already active sessions and bearers in the secondary anchor PFE. After this is completed, new sessions are programmed in the secondary anchor PFE in parallel to the primary anchor PFE. From this point forward, the secondary anchor PFE can take over anytime.

If the primary anchor PFE fails, the secondary anchor PFE starts handling traffic. It might take a few seconds to detect the failure of the primary anchor PFE and for the RE to re-route the GTP-U traffic to the secondary PFE. This delay results in traffic loss during the anchor PFE switchover. Additionally, there is a loss of any charging data not reported by the primary anchor PFE before it failed. Anchor PFE switchover does not affect active sessions/bearers. In-flight changes to sessions and bearers as well as new sessions being created during anchor PFE switchover are rolled back. If supported by the SAEGW-C, the SAEGW-U can send an Sx Session Set Deletion Request for the lost sessions through the Sx interface, and the sessions are flushed in the SAEGW-C. After the anchor PFE switchover, the configured primary anchor PFE can be restored, starting as a secondary anchor PFE and going through catch-up similar to secondary APFE failure and restoration described above.

To configure redundancy for two anchor PFE line cards:

  1. Configure PFE interfaces in each anchor PFE line card slot in aggregated anchor PFE configuration. For example:
  2. Reference the aggregated anchor PFE interfaces in the SAEGW-U configuration. For example:

When the configured primary anchor PFE fails, the secondary anchor PFE takes over. When the failed primary anchor PFE recovers, it does not automatically resume primary status. It is now in secondary status until the configured secondary anchor PFE fails.

  1. However, you can force the two anchor PFEs to revert to their configured state by setting a revert-time, in hours, under the [edit interfaces aggregated-pfe-group anchoring-options] hierarchy. For example:

Related Documentation