Many possible constraints may be used to compute a path over a network. Some networks are deployed with multiple planes. A simple form of constraint may be to use a particular plane. A more sophisticated form of constraint can include some extended metric, like delay, as described in [RFC7810]. Even more advanced case could be to restrict the path and avoid links with certain affinities. Combinations of these are also possible. To provide a maximum flexibility, the mapping between the algorithm value and its meaning can be defined by the user. When all the routers in the domain have the common understanding what the particular algorithm value represents, the computation for such algorithm is consistent and the traffic is not subject to looping. Here, since the meaning of the algorithm is not defined by any standard, but is defined by the user, it is called a Flexible Algorithm.

An algorithm defines how the best path is computed by IGP. Routers advertise the support for the algorithm as a node capability. Prefix-SIDs are also advertised with an algorithm value and are tightly coupled with the algorithm itself.

An algorithm is a one octet value. Values from 128 to 255 are reserved for user defined values and are used for Flexible Algorithm representation.

To guarantee the loop free forwarding for paths computed for a particular Flexible Algorithm, all routers in the network must share the same definition of the Flexible Algorithm. This is achieved by dedicated router(s) advertising the definition of each Flexible Algorithm. Such advertisement is associated with the priority to make sure that all routers will agree on a single and consistent definition for each Flexible Algorithm.

Definition of Flexible Algorithm includes:

• Metric type

• Affinity constraints

To enable the router to advertise the definition for the particular Flexible Algorithm, advertise-definition command is used. At least one router in the area, preferably two for redundancy, must advertise the Flexible Algorithm definition. Without the valid definition being advertised, the Flexible Algorithm will not be functional.

Various link attributes may be used during the Flexible Algorithm path calculation. For example, include or exclude rules based on link affinities can be part of the Flexible Algorithm definition, as defined in IETF draft draft-ietf-lsr-flex-algo.

Link attribute advertisements used during Flexible Algorithm calculation must use the Application-Specific Link Attribute (ASLA) advertisements, as defined in RFC8919 (IS-IS) and RFC8920 (OSPF). In the case of IS-IS, if the L-Flag is set in the ASLA advertisement, then legacy advertisements (IS-IS Extended Reachability TLV) are used instead.

The mandatory use of ASLA advertisements applies to the following link attributes:

• Minimum Unidirectional Link Delay

• TE Default Metric

• Administrative Group

• Extended Administrative Group

• Shared Risk Link Group

To be able to forward traffic on a Flexible Algorithm specific path, all routers participating in the Flexible Algorithm will install a MPLS labeled path for the Flexible Algorithm specific SID that is advertised for the prefix. Only prefixes for which the Flexible Algorithm specific Prefix-SID is advertised is subject to Flexible Algorithm specific forwarding.

A router may compute path for multiple Flexible Algorithms. A router must be configured to support particular Flexible Algorithm before it can compute any path for such Flexible Algorithm. A router must have a valid definition of the Flexible Algorithm before Flexible Algorithm is used.

The router uses the following rules to prune links from the topology during the Flexible Algorithm computation:

• All nodes that don't advertise support for Flexible Algorithm are pruned from the topology.

• Affinities:

  • Check if any exclude affinity rule is part of the Flexible Algorithm Definition. If such exclude rule exists, check if any color that is part of the exclude rule is also set on the link. If such a color is set, the link must be pruned from the computation.

  • Check if any include-any affinity rule is part of the Flexible Algorithm Definition. If such include-any rule exists, check if any color that is part of the include-any rule is also set on the link. If no such color is set, the link must be pruned from the computation.

  • Check if any include-all affinity rule is part of the Flexible Algorithm Definition. If such include-all rule exists, check if all colors that are part of the include-all rule are also set on the link. If all such colors are not set on the link, the link must be pruned from the computation

  Note	See Flexible Algorithm Affinity Constraint.

• Router uses the metric that is part of the Flexible Algorithm definition. If the metric isn't advertised for the particular link, the link is pruned from the topology.

router isis instance flex-algo algo microloop avoidance disable

router ospf process flex-algo algo microloop avoidance disable

router isis instance flex-algo algo fast-reroute disable

router ospf process flex-algo algo fast-reroute disable

Flexible Algorithm path to any prefix must be installed in the forwarding using the Prefix-SID that was advertised for such Flexible Algorithm. If the Prefix-SID for Flexible Algorithm is not known, such Flexible Algorithm path is not installed in forwarding for such prefix..

Only MPLS to MPLS entries are installed for a Flexible Algorithm path. No IP to IP or IP to MPLS entries are installed. These follow the native IPG paths computed based on the default algorithm and regular IGP metrics.

Prefix redistribution from IS-IS to another IS-IS instance or protocol was limited to SR algorithm 0 (regular SPF) prefix SIDs; SR algorithm 1 (Strict SPF) and SR algorithms 128-255 (Flexible Algorithm) prefix SIDs were not redistributed along with the prefix. The Segment Routing IS-IS Flexible Algorithm Prefix SID Redistribution feature allows redistribution of strict and flexible algorithms prefix SIDs from IS-IS to another IS-IS instance or protocols. This feature is enabled automatically when you configure redistribution of IS-IS Routes with strict or flexible algorithm SIDs.

A limitation of the existing Flexible Algorithm functionality in IS-IS is the inability to compute the best path to a prefix in a remote area or remote IGP domain. Prefixes are advertised between IS-IS areas or between protocol domains, but the existing prefix metric does not reflect any of the constraints used for Flexible Algorithm path. Although the best Flexible Algorithm path can be computed to the inter-area or redistributed prefix inside the area, the path may not represent the overall best path through multiple areas or IGP domains.

The Flexible Algorithm Prefix Metric feature introduces a Flexible Algorithm-specific prefix-metric in the IS-IS prefix advertisement. The prefix-metric provides a way to compute the best end-to-end Flexible Algorithm optimized paths across multiple areas or domains.

Note	The Flexible Algorithm definition must be consistent between domains or areas. Refer to section 8 in IETF draft https://datatracker.ietf.org/doc/draft-ietf-lsr-flex-algo/.

The following tables explain the behaviors for advertising (transmitting) and processing (receiving) Flexible Algorithm link attributes.

Use the following command to configure the IOS XR IS-IS Flex Algo implementation to strictly use the link delay metric value received in the IS-IS ASLA:

router isis instance-id receive application flex-algo delay app-only

Use the following command to configure the Flexible Algorithm-specific TE metric value under IS-IS, where metric_value is from 1 to 16777214:

• router isis instance interface type interface-path-id address-family { ipv4 | ipv6} [unicast] te-metric flex-algo metric_value [level {1 | 2}]

The following example shows how to configure the IS-IS Flexible Algorithm-specific TE metric value to 50:

Router(config)# router isis 1
Router(config-isis)# interface HundredGigE 0/0/0/2
Router(config-isis-if)# address-family ipv4 unicast
Router(config-isis-if-af)# te-metric flex-algo 50

SR-TE On Demand Next-Hop (ODN) feature can be used to steer the BGP traffic towards the Flexible Algorithm paths.

The following example configuration shows how to setup BGP steering local policy, assuming two router: R1 (2.2.2.2) and R2 (4.4.4.4), in the topology.

Configuration on router R1:

vrf Test
address-family ipv4 unicast
  import route-target
   1:150
  !
  export route-policy SET_COLOR_RED_HI_BW
  export route-target
   1:150
  !
!
!
interface Loopback0
ipv4 address 2.2.2.2 255.255.255.255
!
interface Loopback150
vrf Test
ipv4 address 2.2.2.222 255.255.255.255
!
interface TenGigE0/1/0/3/0
description exr1 to cxr1
ipv4 address 10.0.20.2 255.255.255.0
!
extcommunity-set opaque color129-red-igp
  129
end-set
!
route-policy PASS
  pass
end-policy
!
route-policy SET_COLOR_RED_HI_BW
  set extcommunity color color129-red-igp
  pass
end-policy
!
router isis 1
is-type level-2-only
net 49.0001.0000.0000.0002.00
log adjacency changes
affinity-map RED bit-position 28
flex-algo 128
  priority 228
!
address-family ipv4 unicast
  metric-style wide
  advertise link attributes
  router-id 2.2.2.2
  segment-routing mpls
!
interface Loopback0
  address-family ipv4 unicast
   prefix-sid index 2
   prefix-sid algorithm 128 index 282
  !
!
interface TenGigE0/1/0/3/0
  point-to-point
  address-family ipv4 unicast
  !
!
!
router bgp 65000
bgp router-id 2.2.2.2
address-family ipv4 unicast
!
address-family vpnv4 unicast
  retain route-target all
!
neighbor-group RR-services-group
  remote-as 65000
  update-source Loopback0
  address-family ipv4 unicast
  !
  address-family vpnv4 unicast
  !
!
neighbor 4.4.4.4
  use neighbor-group RR-services-group
!
vrf Test
  rd auto
  address-family ipv4 unicast
   redistribute connected
  !
segment-routing
traffic-eng
  logging
   policy status
  !
  segment-list sl-cxr1
   index 10 mpls label 16294
  !
  policy pol-foo
   color 129 end-point ipv4 4.4.4.4
   candidate-paths
    preference 100
     explicit segment-list sl-cxr1
     !
    !
   !
  !
!
!

Configuration on router R2:

vrf Test
address-family ipv4 unicast
  import route-target
   1:150
  !
  export route-policy SET_COLOR_RED_HI_BW
  export route-target
   1:150
  !
!
!
interface TenGigE0/1/0/1
description cxr1 to exr1
ipv4 address 10.0.20.1 255.255.255.0
!
extcommunity-set opaque color129-red-igp
  129
end-set
!
route-policy PASS
  pass
end-policy
!
route-policy SET_COLOR_RED_HI_BW
  set extcommunity color color129-red-igp
  pass
end-policy
!
router isis 1
is-type level-2-only
net 49.0001.0000.0000.0004.00
log adjacency changes
affinity-map RED bit-position 28
affinity-map BLUE bit-position 29
affinity-map GREEN bit-position 30
flex-algo 128
  priority 228
!
flex-algo 129
  priority 229
!
flex-algo 130
  priority 230
!
address-family ipv4 unicast
  metric-style wide
  advertise link attributes
  router-id 4.4.4.4
  segment-routing mpls
!
interface Loopback0
  address-family ipv4 unicast
   prefix-sid index 4
   prefix-sid algorithm 128 index 284
   prefix-sid algorithm 129 index 294
   prefix-sid algorithm 130 index 304
  !
!
interface GigabitEthernet0/0/0/0
  point-to-point
  address-family ipv4 unicast
  !
!
interface TenGigE0/1/0/1
  point-to-point
  address-family ipv4 unicast
  !
!
router bgp 65000
bgp router-id 4.4.4.4
address-family ipv4 unicast
!
address-family vpnv4 unicast
!
neighbor-group RR-services-group
  remote-as 65000
  update-source Loopback0
  address-family ipv4 unicast
  !
  address-family vpnv4 unicast
  !
!
neighbor 10.1.1.1
  use neighbor-group RR-services-group
!
neighbor 2.2.2.2
  use neighbor-group RR-services-group
!
vrf Test
  rd auto
  address-family ipv4 unicast
   redistribute connected
  !
  neighbor 25.1.1.2
   remote-as 4
   address-family ipv4 unicast
    route-policy PASS in
    route-policy PASS out
   !
  !
!
!
segment-routing
!
end