IGP Flexible Algorithms (Flex-Algorithm) In IP Networks

An IGP Flexible Algorithm (Flex-Algorithm) allows IGPs to compute constraint-based paths. The base IGP Flex-Algorithm specification describes how it is used with Segment Routing (SR) data planes - SR MPLS and SRv6. An IGP Flex-Algorithm as specified in computes a constraint-based path to: All Flex-Algorithm specific Prefix Segment Identifiers (SIDs) . All Flex-Algorithm specific SRv6 Locators . Therefore, Flex-Algorithm cannot be deployed in the absence of SR or SRv6. This document extends Flex-Algorithm, allowing it to compute paths to IPv4 and IPv6 prefixes.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

The System Architecture for the 5G System describes the N3 interface between gNodeB and UPF (User Plane Function). Mobile networks are becoming more and more IP centric. Each end-user session from a gNodeB can be destined to a specific UPFs (User Plane Function) based on the session requirements. For example, some sessions require high bandwidth, others need to be routed along the lowest latency path. Each UPF is assigned a unique IP address. As a result, traffic for different sessions is destined to a different destination IP address. The IP address allocated to the UPF can be associated with an algorithm. The mobile user traffic is then forwarded along the path based on the algorithm-specific metric and constraints. As a result, traffic can be sent over a path that is optimized for minimal latency or highest bandwidth. This mechanism is used to achieve SLA (Service Level Agreement) appropriate for a user session.

To guarantee loop-free forwarding, all routers that participate in a Flex-Algorithm MUST agree on the Flex-Algorithm Definition (FAD). Selected nodes within the IGP domain MUST advertise FADs as described in Sections 5, 6, and 7 of .

A node may use various algorithms when calculating paths to nodes and prefixes. Algorithm values are defined in the IGP Algorithm Type Registry . A node MUST participate in a Flex-Algorithm to be: Able to compute path for such Flex-Algorithm Part of the topology for such Flex-Algorithm Flex-Algorithm participation MUST be advertised for each Flex-Algorithm data-plane independently, as specified in . Using Flex-Algorithm for regular IPv4 and IPv6 prefixes represents an independent Flex-Algorithm data-plane, and as such, the Flex-Algorithm participation for the IP Flex-Algorithm data-plane MUST be signalled independently of any other Flex-Algorithm data-plane (e.g., SR). All routers in an IGP domain participate in default algorithm 0. Advertisement of participation in IP Flex-Algorithm does not impact the router participation in default algorithm 0. Advertisement of participation in IP Flex-Algorithm does not impact the router participation signaled for other data-planes. For example, it is possible that a router participates in a particular flex-algo for the IP data-plane but does not participate in the same flex-algo for the SR data-plane. The following sections describe how the IP Flex-Algorithm participation is advertised in IGP protocols.

The IS-IS IP Algorithm Sub-TLV is a sub-TLV of the IS-IS Router Capability TLV and has the following format:

Type: IP Algorithm Sub-TLV (Value 29) Length: Variable Algorithm (1 octet): Value from 128 to 255. The IP Algorithm Sub-TLV MUST be propagated throughout the level and MUST NOT be advertised across level boundaries. Therefore, the S bit in the Router Capability TLV, in which the IP Algorithm Sub-TLV is advertised, MUST NOT be set. The IP Algorithm Sub-TLV is optional. It MUST NOT be advertised more than once at a given level. A router receiving multiple IP Algorithm sub-TLVs from the same originator MUST select the first advertisement in the lowest-numbered LSP and subsequent instances of the IP Algorithm Sub-TLV MUST be ignored. The use of the IP Algorithm Sub-TLV to advertise support for algorithms outside the Flex-Algorithm range (128-255) is outside the scope of this document. The IP Flex-Algorithm participation advertised in the IS-IS IP Algorithm Sub-TLV is topology independent. When a router advertises participation in the IS-IS IP Algorithm Sub-TLV, the participation applies to all topologies in which the advertising node participates.

The OSPF IP Algorithm TLV is a top-level TLV of the Router Information Opaque LSA and has the following format:

Type: IP Algorithm TLV (Value TBD1 by IANA) Length: Variable Algorithm (1 octet): Value from 128 to 255. The IP Algorithm TLV is optional. It MUST only be advertised once in the Router Information LSA. When multiple IP Algorithm TLVs are received from a given router, the receiver MUST use the first occurrence of the TLV in the Router Information LSA. If the IP Algorithm TLV appears in multiple Router Information LSAs that have different flooding scopes, the IP Algorithm TLV in the Router Information LSA with the area-scoped flooding scope MUST be used. If the IP Algorithm TLV appears in multiple Router Information LSAs that have the same flooding scope, the IP Algorithm TLV in the Router Information LSA with the numerically smallest Instance ID (Opaque ID for OSPFv2 or Link State ID for OSPFv3) MUST be used and subsequent instances of the IP Algorithm TLV MUST be ignored. The Router Information LSA can be advertised at any of the defined flooding scopes (link, area, or Autonomous System (AS)). For the purpose of IP Algorithm TLV advertisement, area or Autonomous System scoped flooding is REQUIRED. The AS flooding scope SHOULD NOT be used unless local configuration policy on the originating router indicates domain-wide flooding. The IP Flex-Algorithm participation advertised in the OSPF IP Algorithm TLV is topology independent. When a router advertises participation in OSPF IP Algorithm TLV, the participation applies to all topologies in which the advertising node participates.

To be able to associate the prefix with the Flex-Algorithm, the existing prefix reachability advertisements cannot be used, because they advertise the prefix reachability in default algorithm 0. Instead, new IP Flex-Algorithm reachability advertisements are defined in IS-IS and OSPF. The M-flag in the FAD is not applicable to IP Algorithm Prefixes. Any IP Algorithm Prefix advertisement includes the Algorithm and Metric fields. When an IP Algorithm Prefix is advertised between areas or domains, the metric field in the IP Algorithm Prefix advertisement MUST be used irrespective of the M-flag in the FAD advertisement.

The IPv4 Algorithm Prefix Reachability top-level TLV is defined for advertising IPv4 Flex-Algorithm Prefix Reachability in IS-IS. This new TLV shares the sub-TLV space defined for TLVs Advertising Prefix Reachability. The IS-IS IPv4 Algorithm Prefix Reachability TLV has the following format:

Type: IPv4 Algorithm Prefix Reachability TLV (Value 126). Length: Variable based on number of prefix entries encoded Rsvd (4 bits): Reserved for future use. They MUST be set to zero on transmission and MUST be ignored on receipt. MTID (12 bits): Multitopology Identifier as defined in [RFC5120]. Note that the value 0 is legal. Followed by one or more prefix entries of the form:

Metric (4 octets): Metric information as defined in . Flags (1 octet):

D-flag: When the Prefix is leaked from level-2 to level-1, the D bit MUST be set. Otherwise, this bit MUST be clear. Prefixes with the D bit set MUST NOT be leaked from level-1 to level-2. This is to prevent looping. Algorithm (1 octet): Associated Algorithm from 128 to 255. Prefix Len (1 octet): Prefix length measured in bits. Prefix (variable length): Prefix mapped to Flex-Algorithm. Optional Sub-TLV-length (1 octet): Number of octets used by sub-TLVs Optional sub-TLVs (variable length). If a router receives multiple IPv4 Algorithm Prefix Reachability advertisements for the same prefix from the same originator, it MUST select the first advertisement in the lowest-numbered LSP and ignore any subsequent IPv4 Algorithm Prefix Reachability advertisements for the same prefix. If a router receives multiple IPv4 Algorithm Prefix Reachability advertisements for the same prefix, from different originators, where all of them do not advertise the same algorithm, it MUST ignore all of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error. In cases where a prefix advertisement is received in both a IPv4 Prefix Reachability TLV (, ) and an IPv4 Algorithm Prefix Reachability TLV, the IPv4 Prefix Reachability advertisement MUST be preferred when installing entries in the forwarding plane.

The IS-IS IPv6 Algorithm Prefix Reachability TLV is identical to the IS-IS IPv4 Algorithm Prefix Reachability TLV, except that it has a distinct type. The type is 127. If a router receives multiple IPv6 Algorithm Prefix Reachability advertisements for the same prefix from the same originator, it MUST select the first advertisement in the lowest-numbered LSP and ignore any subsequent IPv6 Algorithm Prefix Reachability advertisements for the same prefix. If a router receives multiple IPv6 Algorithm Prefix Reachability advertisements for the same prefix, from different originators, where all of them do not advertise the same algorithm, it MUST ignore all of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error. In cases where a prefix advertisement is received in both an IPv6 Prefix Reachability TLV (, ) and an IPv6 Algorithm Prefix Reachability TLV, the IPv6 Prefix Reachability advertisement MUST be preferred when installing entries in the forwarding plane. In cases where a prefix advertisement is received in both an IS-IS SRv6 Locator TLV and in IS-IS IPv6 Algorithm Prefix Reachability TLV, the receiver MUST ignore both of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error.

A new Sub-TLV of the OSPFv2 Extended Prefix TLV is defined for advertising IP Algorithm Prefix Reachability in OSPFv2, the OSPFv2 IP Algorithm Prefix Reachability Sub-TLV. The OSPFv2 IP Algorithm Prefix Reachability Sub-TLV has the following format:

Type (2 octets) : The value is TBD2. Length (2 octet): 8 MT-ID (1 octet): Multi-Topology ID as defined in Algorithm (1 octet): Associated Algorithm from 128 to 255. Flags: (1 octet): The following flags are defined:

align="center"> 0 1 2 3 4 5 6 7 8 +-+-+-+-+-+-+-+-+-+ |E| Reserved | +-+-+-+-+-+-+-+-+-+ where: bit E: Same as bit E defined in section A.4.5 of . The remaining bits, are reserved for future use. They MUST be set to zero on transmission and MUST be ignored on receipt. Reserved: (1 octets). SHOULD be set to 0 on transmission and MUST be ignored on reception. Metric (4 octets): The algorithm-specific metric value. The metric value of 0XFFFFFFFF MUST be considered as unreachable. An OSPFv2 router receiving multiple OSPFv2 IP Algorithm Prefix Reachability Sub-TLVs in the same OSPFv2 Extended Prefix TLV, MUST select the first advertisement of this Sub-TLV and MUST ignore all remaining occurences of this Sub-TLV in the OSPFv2 Extended Prefix TLV. An OSPFv2 router receiving multiple OSPFv2 IP Algorithm Prefix Reachability TLVs for the same prefix, from different originators, where all of them do not advertise the same algorithm, MUST ignore all of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error. In cases where a prefix advertisement is received in any of the LSAs advertising the prefix reachability for algorithm 0 and in an OSPFv2 IP Algorithm Prefix Reachability Sub-TLV, only the prefix reachability advertisement for algorithm 0 MUST be used and all occurences of the OSPFv2 IP Algorithm Prefix Reachability Sub-TLV MUST be ignored. When computing the IP Algorithm Prefix reachability in OSPFv2, only information present in the OSPFv2 Extended Prefix TLV MUST be used. There will not be any information advertised for the IP Algorithm Prefix in any of the OSPFv2 LSAs that advertise prefix reachability for algorithm 0. For the IP Algorithm Prefix the OSPFv2 Extended Prefix TLV is used to advertise the prefix reachability, unlike for algorithm 0 prefixes, where the OSPFv2 Extended Prefix TLV is only used to advertise additional attributes, but not the reachability itself.

A new Sub-TLV of the OSPFv2 Extended Prefix TLV is defined for advertising IP Forwarding Address, the OSPFv2 IP Forwarding Address Sub-TLV. The OSPFv2 IP Forwarding Address Sub-TLV has the following format:

Type (2 octets) : The value is TBD4. Length (2 octet): 4 Forwarding Address: Same as defined in section A.4.5 of . The OSPFv2 IP Forwarding Address Sub-TLV MUST NOT be used for computing algorithm 0 prefix reachability and MUST be ignored for algorithm 0 prefixes. The OSPFv2 IP Forwarding Address Sub-TLV is optional. If it is not present, the forwarding address for computing the IP Algorithm Prefix reachability is assumed to be equal to 0.0.0.0. The OSPFv2 IP Forwarding Address Sub-TLV is only applicable to Autonomous System (AS) External and Not-So-Stubby Area (NSSA) External route types. If the OSPFv2 IP Forwarding Address Sub-TLV is advertised in the OSPFv2 Extended Prefix TLV that has the Route Type field set to any other type, the OSPFv2 IP Forwarding Address Sub-TLV MUST be ignored.

The OSPFv3 IP Algorithm Prefix Reachability Sub-TLV is defined for advertisement of the IP Algorithm Prefix Reachability in OSPFv3. The OSPFv3 IP Algorithm Prefix Reachability Sub-TLV is a sub-TLV of the following OSPFv3 TLVs defined in : Intra-Area-Prefix TLV Inter-Area-Prefix TLV External-Prefix TLV The format of OSPFv3 IP Algorithm Prefix Reachability Sub-TLV is shown below:

Where: Type (2 octets): The value is TBD3. Length (2 octets): 8. Algorithm (1 octet): Associated Algorithm from 128 to 255. Reserved: (3 octets). SHOULD be set to 0 on transmission and MUST be ignored on reception. Metric (4 octets): The algorithm-specific metric value. The metric value of 0XFFFFFFFF MUST be considered as unreachable. When the OSPFv3 IP Algorithm Prefix Reachability Sub-TLV is present, the NU-bit in the PrefixOptions field of the parent TLV MUST be set. This is needed to prevent the OSPFv3 IP Algorithm Prefix Reachability advertisement from contributing to the base algorithm reachability. If the NU-bit in the PrefixOptions field of the parent TLV is not set, the OSPFv3 IP Algorithm Prefix Sub-TLV MUST be ignored by the receiver. The metric value in the parent TLV is RECOMMENDED to be set to LSInfinity . This recommendation is provided as a network troubleshooting convenience; if it is not followed the protocol will still function correctly. An OSPFv3 router receiving multiple OSPFv3 IP Algorithm Prefix Reachability Sub-TLVs in the same parent TLV, MUST select the first advertisement of this Sub-TLV and MUST ignore all remaining occurences of this Sub-TLV in the parent TLV. An OSPFv3 router receiving multiple OSPFv3 IP Algorithm Prefix Reachability TLVs for the same prefix, from different originators, where all of them do not advertise the same algorithm, MUST ignore all of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error. In cases where a prefix advertisement is received in any of the LSAs advertising the prefix reachability for algorithm 0 and in an OSPFv3 OSPFv3 IP Algorithm Prefix Reachability Sub-TLV, only the prefix reachability advertisement for algorithm 0 MUST be used and all occurences of the OSPFv3 IP Algorithm Prefix Reachability Sub-TLV MUST be ignored. In cases where a prefix advertisement is received in both an OSPFv3 SRv6 Locator TLV and in an OSPFv3 IP Algorithm Prefix Reachability Sub-TLV, the receiver MUST ignore both of them and MUST NOT install any forwarding entries based on these advertisements. This situation SHOULD be logged as an error.

defines the OSPF Flexible Algorithm ASBR Metric Sub-TLV (FAAM) that is used by an OSPFv2 or an OSPFv3 ABR to advertise a Flex-Algorithm specific metric associated with the corresponding ASBR LSA. As described in each data-plane signals its participation independently. IP Flex-Algorithm participation is signaled independent of Segment Routing (SR) Flex-Algorithm participation. As a result, the calculated topologies for SR and IP Flex-Algorithm could be different. Such difference prevents the usage of FAAM for the purpose of the IP Flex-Algorithm. The OSPF IP Flexible Algorithm ASBR Metric (IPFAAM) Sub-TLV is defined for the advertisement of the IP Flex-Algorithm specific metric associated with an ASBR by the ABR. The IPFAAM Sub-TLV is a Sub-TLV of the: - OSPFv2 Extended Inter-Area ASBR TLV as defined in - OSPFv3 Inter-Area-Router TLV defined in The OSPF IPFAAM Sub-TLV has the following format:

Type (2 octets): 2 (allocated by IANA) for OSPFv2, TBD5 for OSPFv3. Length (2 octets): 8. Algorithm (1 octet): Associated Algorithm from 128 to 255. Reserved: (3 octets). SHOULD be set to 0 on transmission and MUST be ignored on reception. Metric (4 octets): The algorithm-specific metric value. The usage of the IPFAAM Sub-TLV is similar to the usage of the FAAM Sub-TLV defined in , but it is used to advertise IP Flex-Algorithm metric. An OSPF ABR MUST include the OSPF IPFAAM Sub-TLVs as part of the ASBR reachability advertisement between areas for every IP Flex-Algorithm in which it participates and the ASBR is reachable in. The FAAM Sub-TLV as defined in MUST NOT be used during IP Flex-Algorithm path calculation, the IPFAAM Sub-TLV MUST be used instead.

The IP Flex-Algorithm is considered as yet another data-plane of the Flex-Algorithm as described in . Participation in the IP Flex-Algorithm is signalled as described in and is specific to the IP Flex-Algorithm data-plane. Calculation of IP Flex-Algorithm paths follows what is described in . This computation uses the IP Flex-Algorithm data-plane participation and is independent of the Flex-Algorithm calculation done for any other Flex-Algorithm data-plane (e.g., SR, SRv6). The IP Flex-Algorithm data-plane only considers participating nodes during the Flex-Algorithm calculation. When computing paths for a given Flex-Algorithm, all nodes that do not advertise participation for the IP Flex-Algorithm, as described in , MUST be pruned from the topology.

The IP Algorithm Prefix Reachability advertisement as described in includes the MTID value that associates the prefix with a specific topology. Algorithm Prefix Reachability advertisement also includes an Algorithm value that explicitly associates the prefix with a specific Flex-Algorithm. The paths to the prefix MUST be calculated using the specified Flex-Algorithm in the associated topology. Forwarding entries for the IP Flex-Algorithm prefixes advertised in IGPs MUST be installed in the forwarding plane of the receiving IP Flex-Algorithm prefix capable routers when they participate in the associated topology and algorithm. Forwarding entries for IP Flex-Algorithm prefixes associated with Flex-Algorithms in which the node is not participating MUST NOT be installed in the forwarding plane.

IGP Flex-Algorithm can be used by many data-planes. The original specification was done for SR and SRv6, this specification adds IP as another data-plane that can use IGP Flex-Algorithm. Other data-planes may be defined in the future. This section provides some details about the coexistence of the various data-planes of an IGP Flex-Algorithm. Flex-Algorithm definition (FAD), as described in , is data-plane independent and is used by all Flex-Algorithm data-planes. Participation in the Flex-Algorithm, as described in , is data-plane specific. Calculation of the flex-algo paths is data-plane specific and uses data-plane specific participation advertisements. Data-plane specific participation and calculation guarantee that the forwarding of the traffic over the Flex-Algorithm data-plane specific paths is consistent between all nodes that apply the IGP Flex-Algorithm to the data-plane. Multiple data-planes can use the same Flex-Algorithm value at the same time and, and as such, share the FAD for it. For example, SR-MPLS and IP can both use a common Flex-Algorithm. Traffic for SR-MPLS will be forwarded based on Flex-algorithm specific SR SIDs. Traffic for IP Flex-Algorithm will be forwarded based on Flex-Algorithm specific prefix reachability advertisements. Note that for a particular Flex-Algorithm, for a particular IP prefix, there will only be path(s) calculated and installed for a single data-plane.

In many networks where IGP Flexible Algorithms are deployed, IGP restoration will be fast and additional protection mechanisms will not be required. IGP restoration may be enhanced by Equal Cost Multipath (ECMP). In other networks, operators can deploy additional protection mechanisms. The following are examples: Loop Free Alternates (LFA) Remote Loop Free Alternates (R-LFA) LFA and R-LFA computations MUST be restricted to the flex-algo topology and the computed backup nexthops should be programmed for the IP flex-algo prefixes.

This specification updates the OSPF Router Information (RI) TLVs Registry as follows: Value TLV Name Reference TBD1 IP Algorithm TLV This Document This document also updates the IS-IS "Sub-TLVs for TLV 242" registry as follows: Value TLV Name Reference 29 IP Algorithm Sub-TLV This Document This document also updates the "IS-IS TLV Codepoints Registry" registry as follows: Value TLV Name IIH LSP SNP Purge Reference 126 IPv4 Algorithm Prefix Reachability TLV N Y N N This document, 127 IPv6 Algorithm Prefix Reachability TLV N Y N N This document, Since the above TLVs share the sub-TLV space managed in the "IS-IS Sub-TLVs for TLVs Advertising Prefix Reachability" registry, IANA is requested to add "IPv4 Algorithm Prefix Reachability TLV (126)" and "IPv6 Algorithm Prefix Reachability TLV (127)" to the list of TLVs in the description of that registry. In addition, columns headed '126' and '127' are added to that registry, as follows:

This document updates the "OSPFv2 Extended Prefix TLV Sub-TLVs" registry as follows: Value TLV Name Reference TBD2 OSPFv2 IP Algorithm Prefix Reachability Sub-TLV This Document, TBD4 OSPFv2 IP Forwarding Address Sub-TLV This Document, This document creates a new registry under "Open Shortest Path First v2 (OSPFv2) Parameters" registry, called "IP Algorithm Prefix Reachability Sub-TLV Flags". The new registry defines the bits in the 8-bit Flags field in the OSPFv2 IP Algorithm Prefix Reachability Sub-TLV (). New bits can be allocated via IETF Review or IESG Approval Bit # Name Reference 0 bit E This Document, 1-7 Reserved This Document, This document updates the "OSPFv3 Extended-LSA Sub-TLVs" registry as follows: Value TLV Name Reference TBD3 OSPFv3 IP Algorithm Prefix Reachability Sub-TLV This Document, TBD5 OSPFv3 IP Flexible Algorithm ASBR Metric Sub-TLV This Document, This document updates the "OSPFv2 Extended Inter-Area ASBR Sub-TLVs" registry as follows: Value TLV Name Reference 2 OSPF IP Flexible Algorithm ASBR Metric Sub-TLV This Document,

This document inherits security considerations from . This document adds one new way to disrupt IGP networks that are using Flex-Algorithm: an attacker can suppress reachability for a given prefix whose reachability is advertised by a legitimate node for a particular IP Flex-Algorithm X, by advertising the same prefix in Flex-Algorithm Y from another, malicious node. (To see why this is, consider, for example, the rule given in the second-last paragraph of ). This attack can be addressed by the existing security extensions, as described in and for IS-IS, in and for OSPFv2, and in and for OSPFv3. If a node that is authenticated is taken over by an attacker, such a rogue node can perform the attack described above. Such an attack is not preventable through authentication, and it is not different from advertising any other incorrect information through IS-IS or OSPF.

Thanks to Bruno Decraene for his contributions to this document. Special thanks to Petr Bonbon Adamec of Cesnet for supporting interoperability testing.