SRv6 uSID w/ TI-LFA

Load top1.vpnv4v6.srv6.usid.configured.cfg

#IOS-XR
configure
load top1.vpnv4v6.srv6.usid.configured.cfg
commit replace
y

#IOS-XE
config replace flash:top1.vpnv4v6.srv6.usid.configured.cfg

Increase the XR2-XR4 link (vlan 3024) metric to 100.

#XR2, XR4
router isis 1
 interface GigabitEthernet0/0/0/0.3024
  address-family ipv6 unicast
   metric 100

Enable TI-LFA so that a link-down event for vlan 3013 on XR1 causes fast-reroute using vlan 3012.

Answer

#XR1
router isis 1
 interface GigabitEthernet0/0/0/0.3013
  address-family ipv6 unicast
   fast-reroute per-prefix
   fast-reroute per-prefix ti-lfa

Explanation

TI-LFA for SRv6 works the same as with SR-MPLS. Instead of an MPLS label stack that is pushed, an SRH is pushed ontop of the packet.

We enable TI-LFA for SRv6 the same way as SR-MPLS. We simply use two commands: fast-reroute per-prefix and fast-reroute per-prefix ti-lfa under the interface under the AFI.

#XR1
router isis 1
 interface GigabitEthernet0/0/0/0.3013
  address-family ipv6 unicast
   fast-reroute per-prefix
   fast-reroute per-prefix ti-lfa

Let’s examine the backup path for XR7’s locator prefix, which is fd00:0:7::/48. Under normal circumstances, this would be an LFA via XR2. However, we’ve increased the metric of the XR2-XR4 link, so a separate P and Q node exists:

The RIB shows us that the SID list for the backup path is XR2’s Adj-SID (uA) for XR4. XR4 will then release the packet to XR7.

If we shut down Gi0/0/0/0.3013 and take a pcap on Gi0/0/0/0.3012, we see that XR1 inserts the SRH into the existing IPv6 header in order to provide fast-reroute protection. The behavior is H.Insert.Red, not H.Encaps.Red, so the existing IPv6 header is retained. (The SRH is inserted, instead of encapsulating the packet with a brand new IPv6 header).

XR2 should perform PSP, placing the fd00:0:7:e001:: SID in the destination, removing the SRH, and forwarding out the interface towards XR4.