Flexible Algorithm Technology White Paper

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contents

Overview·· 1

Technical background· 1

Benefits· 1

flexible algorithm implementation· 1

Concepts· 1

flexible algorithm extension TLVs for IS-IS· 2

Overview of TLVs· 2

IS-IS FAD Sub-TLV· 3

Sub-TLVs of IS-IS FAD Sub-TLV· 3

flexible algorithm selection· 5

Generate flexible algorithm topology using the algorithm·· 6

Introduction to algorithm topology· 6

Constraint condition· 6

Algorithm topology generation steps· 7

TI-LFA FRR of flexible algorithm·· 8

Calculation of TI-LFA FRR path for flexible algorithm·· 8

Tangent guard of flexible algorithm against microloops· 9

Application of flexible algorithms in SRv6· 10

Calculate the optimal path for SRv6 forwarding using flexible algorithm·· 10

Working principle of flexible algorithm in calculating the optimal path· 10

Calculate the optimal path for SRv6 BE using flexible algorithm·· 11

Calculate the optimal path for SRv6 TE policy using flexible algorithm·· 12

Flexible algorithm calculates the SID list for SRv6 TE policy candidate path created by ODN· 13

Typical network application· 14

Topology planning based on flexible algorithm·· 14

Optimal path calculation based on flexible algorithm·· 15

Related documentation· 16

 

Overview

Technical background

IGP protocols typically calculate the optimal path to a destination based on the cost of the link. However, in scenarios where flexible routing and traffic engineering are required, the cost of the link cannot be the sole criterion for path quality. For instance, for real-time services, low latency should be the standard for path selection; in networks where the control and forwarding planes are separated, different types of traffic need to be forwarded through different paths. In such scenarios, the paths calculated by traditional IGP routing algorithms cannot meet diverse user needs.

By adopting traffic engineering methods like RSVP and MPLS TE, which calculate paths based on various metrics and topological constraints, including path cost, the issues can be resolved by using the calculated tunnel path instead of the original path from the IGP protocol. However, this approach requires the deployment of new protocols, increasing the complexity of network deployment and maintenance. flexible algorithm technology extends IGP protocols by advertising the algorithm type, metric type, and topological constraints for route calculation through IGP protocol messages, thus enabling the calculation of paths that meet specific conditions based on IGP protocols. Currently, only the IS-IS protocol supports flexible algorithm.

Benefits

Flexible algorithm has the following advantages:

·     Flexible algorithm customization: Users can freely select the metric types for flexible algorithm route computation (IGP link cost, link delay, and MPLS TE metric) and constraint conditions (including or excluding specific links) based on their requirements.

·     Flexible topology planning: According to specific conditions, flexible algorithm can plan a network topology that fully meets the demands of the IGP protocol. For instance, it can split different business traffic among various topologies and calculate the optimal path for the traffic in each respective topology.

·     Loop prevention mechanism: Within the IGP routing domain (RD), all devices configured with the flexible algorithm use the same algorithm definition, and the calculation results are consistent, which will not cause a loop.

·     High availability (HA): The flexible algorithm supports backup path calculation through TI-LFA FRR, enabling rapid switchover to the backup path for continuous forwarding when there is a link or node fault in the network, thereby minimizing traffic loss to the greatest extent.

·     Support for SRv6 VPN application: The flexible algorithm can be applied to the SRv6 VPN network, calculating a public network forwarding path that fulfills specific requirements for the SRv6 VPN. The flexible algorithm can be used for both SRv6 BE path computations and SRv6 TE path computations in the SRv6 VPN.

flexible algorithm implementation

Concepts

The fundamental concepts related to flexible algorithm include the following:

·     FAD—flexible algorithm definition, a composite of algorithm type, metric type, and constraints.

·     Algorithm type—Algorithm used for path calculation. In the current software version, only the SPF algorithm is supported.

·     Optimization goal—Use the flexible algorithm to calculate the path with the least cost to reach the destination, based on a specific metric type.

The flexible algorithm supports the following metric types:

¡     IS-IS link cost.

¡     Link delay.

¡     TE cost.

·     Constraint—Link filter criteria for topology generation. You can use a set of constraints to exclude links with specific affinities from or include them in the flexible algorithm topology. flexible algorithms use the following link attributes:

¡     Affinity attribute—A link attribute, also known as link color. IS-IS uses LSPs that contain extended administrative group sub-TLVs to advertise FAD constraints, including flexible algorithm exclude admin group sub-TLV, flexible algorithm include-any admin group sub-TLV, and flexible algorithm include-all admin group sub-TLV, which are corresponding to the exclude-any, include-any, and include-all rules, respectively.

¡     SRLG—A Shared Risk Link Group (SRLG) is a set of links that share a resource. If one link in the group fails, the other links might also fail. IS-IS uses LSPs that contain flexible algorithm exclude SRLG sub-TLVs to advertise SRLGs that FAD excludes.

·     flexible algorithm ID—The unique identifier for flexible algorithm is a digit between 128 and 255. The IGP Algorithm Types value range allocated by IANA for IGP protocols is 0 to 255, and flexible algorithm utilizes the latter half of this range.

flexible algorithm extension TLVs for IS-IS

This section introduces various extension TLVs used by the IS-IS protocol to advertise FAD information, ensuring loop-free forwarding path calculation for flexible algorithm.

Overview of TLVs

Table 1 flexible algorithm extension TLVs for IS-IS

TLV	Description	Carry Location
IS-IS Flexible Algorithm Definition Sub-TLV	Announcement: Definition of flexible algorithm	As a sub-TLV of IS-IS Router Capability TLV.
Exclude Admin Group Sub-TLV from the IS-IS Flexible Algorithm.	Announcement: The 'Exclude-Any' rule in the affinity properties of flexible algorithm constraint conditions.	As a sub-TLV of the IS-IS Flexible Algorithm Definition Sub-TLV.
The IS-IS Flexible Algorithm Include-Any Admin Group Sub-TLV is included.	Announcement: The Include-Any rule applies to the affinity properties in the constraint conditions of flexible algorithm.	As the Sub-TLV of the IS-IS Flexible Algorithm Definition Sub-TLV.
The IS-IS Flexible Algorithm includes the All Admin Group Sub-TLV.	Announcement: The Include-All rule applies to affinity properties in the constraint conditions of flexible algorithm.	As a sub-TLV of the IS-IS Flexible Algorithm Definition Sub-TLV.
Exclude SRLG Sub-TLV from IS-IS Flexible Algorithm.	Announce the Exclude SRLG rule in the condition constraints of flexible algorithm within the shared risk link group.	As a sub-TLV of the IS-IS Flexible Algorithm Definition Sub-TLV.

 

IS-IS FAD Sub-TLV

The IS-IS Flexible Algorithm Definition Sub-TLV (hereafter referred to as IS-IS FAD Sub-TLV) is used to advertise flexible algorithm definitions, including flexible algorithm identifier (ID), metric type, algorithm type, and priority. The format of the message is shown in Figure 1, and the meanings of the fields included in the message are shown in Table 2.

Figure 1 Message format of IS-IS FAD Sub-TLV

 

Table 2 Field description for IS-IS FAD Sub-TLV

Field name	Length	Description
Type	8 bits	The type of TLV, with a value of 26.
Length	8 bits	Total length of TLV.
Flex-Algorithm	8 bits	The flexible algorithm identifier (ID) has a value range of 128 to 255.
Metric-Type	8 bits	The metric type advertised by FAD, which is the optimization goal of flexible algorithm: ·     0—IGP Metric ·     1—Min Unidirectional Link Delay ·     2—Traffic Engineering Default Metric
Calc-Type	8 bits	The type of algorithm currently only supports the Shortest Path First (SPF), which is represented by the value 0.
Priority	8 bits	The privilege level of the FAD advertisement, with a value range of 0 to 255.
Sub-TLVs	Variable length	Optional Sub-TLV

 

A device can only advertise one type of IS-IS FAD Sub-TLV for the same flexible algorithm. If a device receives multiple different IS-IS FAD Sub-TLVs from the same FAD advertisement source for the same flexible algorithm, it will select the IS-IS FAD Sub-TLV carried by the message with the smallest LSP fragment number.

Sub-TLVs of IS-IS FAD Sub-TLV

In the Sub-TLVs field of the IS-IS FAD Sub-TLV, the carried <cf color=000000>child TLV is used to advertise  the constraint conditions of the FAD. Different types of child TLVs can only appear once in the</cf> IS-IS FAD Sub-TLV. If they appear multiple times, the IS-IS FAD Sub-TLV won't be processed by the receiver.

Exclude Admin Group Sub-TLV from the IS-IS Flexible Algorithm

The IS-IS Flexible Algorithm Exclude Admin Group Sub-TLV is used to advertise the Exclude-Any rule in the affinity properties of flexible algorithm constraints. The format of the message is shown in Figure 2, and the meanings of the fields included in the message are shown in Table 3.

Figure 2 Exclude Admin Group Sub-TLV from IS-IS Flexible Algorithm

 

Table 3 Field description for IS-IS Flexible Algorithm Exclude Admin Group Sub-TLV

Field name	Length	Description
Type	8 bits	The TLV type, with a value of 1.
Length	8 bits	The total length of TLV must be a multiple of 4 bytes.
Extended Administrator Group	The variable length can reach up to a maximum of 256 bits.	The link affinity properties contained in the Exclude-Any rule.

 

Flexible IS-IS Algorithm with Include-Any Admin Group Sub-TLV

The IS-IS Flexible Algorithm Include-Any Admin Group Sub-TLV is used to advertise the Include-Any rule in flexible algorithm constraint conditions. The format of the message is shown in Figure 3, and the meaning of the fields included in the message is shown in Table 4.

Figure 3 IS-IS Flexible Algorithm features an Include-Any Admin Group Sub-TLV

 

Table 4 Field description for IS-IS Flexible Algorithm Include-Any Admin Group Sub-TLV

Field Name	Length	Description
Type	8bits	The TLV type, with a value of 2.
Length	8bits	The total length of the TLV must be an integer multiple of 4 bytes.
Extended Administration Group	Variable length, with a maximum length of 256 bits.	The Include-Any rule contains link affinity properties in the package.

 

IS-IS Flexible Algorithm includes all administrative group Sub-TLVs

The IS-IS Flexible Algorithm Include-All Admin Group Sub-TLV is used to advertise the Include-All rule of affinity properties in the constraint conditions of flexible algorithm. The format of the message is shown in Figure 4, and the meanings of the fields in the message are presented in Table 5.

Figure 4 Flexible IS-IS Algorithm Include-All Admin Group Sub-TLV is incorporated

 

Table 5 IS-IS Flexible Algorithm features an inclusive sub-TLV for the admin group

Field Name	Length	Description
Type	8 bits	The TLV type, with a value of 3.
Length	8 bits	The total length of TLV must be an integer multiple of 4 bytes.
Expanded Administration Group	The length is adjustable, with a maximum length of 256 bits.	The link affinity properties are included in the Include-All rule package.

 

Exclude SRLG Sub-TLV from the IS-IS Flexible Algorithm

The IS-IS Flexible Algorithm Exclude SRLG Sub-TLV is used to advertise the shared risk link group (SRLG) exclusion rule in flexible algorithm constraint conditions. The format of the message is shown in Figure 5, and the meanings of the fields contained in the package are shown in Table 6.

Figure 5 SRLG Sub-TLV is excluded from the IS-IS Flexible Algorithm

 

Table 6 Field description for the IS-IS Flexible Algorithm Exclude SRLG Sub-TLV

Field Name	Length	Description
Type	8 bits	The TLV type with a value of 5.
Length	8 bits	The total length of TLV must be a multiple of 4 bytes.
Value of Shared Risk Link Group	Variable length, with a maximum length of 128 bytes.	Exclude the shared risk link groups contained in the SRLG rules.

 

flexible algorithm selection

After a node is configured with flexible algorithm, it becomes eligible to participate in the path calculation of flexible algorithm. In the IS-IS routing domain (RD), there needs to be at least one node announcing a FAD for devices to use flexible algorithm for path calculations. To avoid loops, nodes running the same flexible algorithm with the same identifier (ID) must select a consistent FAD within the advertisement range of the FAD. The selection mechanism is as follows:

1.     Select the FAD with the highest privilege level from the advertisements in this zone, including those generated locally and received. If no local FAD is advertised, select the FAD with the highest privilege level from the received advertisements.

2.     If there are multiple FADs with the same highest privilege level, the FAD advertised by the device with the largest router identifier is selected. In the IS-IS protocol, the router identifier is the System ID.

Generate flexible algorithm topology using the algorithm

Introduction to algorithm topology

The flexible algorithm topology refers to the scope of the path calculated by flexible algorithm. When flexible algorithm generates the topology, it judges whether the affinity properties and the shared risk link group of the link match the constraints. Only the links that match are included in or excluded from the topology by flexible algorithm.

The topology planning of flexible algorithm can isolate forwarding paths of different types of business traffic, allowing users to flexibly customize the optimal path of IGP protocol traffic forwarding to the greatest extent. This avoids the problem where the path selection cannot be influenced only by metric type.

Constraint condition

Based on the constraint condition of affinity properties

IS-IS supports 256 affinity {properties} {bit} positions, with numbering from 0 to 255. To facilitate use, {naming} can be defined for affinity property bit positions. As shown in Figure 2-6, the first bit position is mapped "blue" and the fifth bit position is mapped"red".

Figure 6 Example of Affinity Properties Bit Position

 

The constraint conditions based on affinity properties include:

·     The Exclude-Any rule implies that if any affinity property in the link matches an affinity property in the Exclude-Any rule, the link will be excluded by flexible algorithm during topology generation. This rule is advertised through the Flexible Algorithm Exclude Admin Group Sub-TLV.

·     Include-any rule: This rule indicates that if a link has at least one affinity property in common with the affinity properties in the Include-any rule, the link will be included in the algorithm topology generation by flexible algorithm. The link will be excluded only if its affinity properties do not contain any of the affinity properties in the Include-any rule. This rule is advertised through the Flexible Algorithm Include-Any Admin Group Sub-TLV.

·     The Include-all rule. This rule signifies that only when the affinity properties of the link include all the affinity properties in the Include-all rule, will the link be included in the topology generation by flexible algorithm, otherwise, the link will be excluded. This rule is advertised through the Flexible Algorithm Include-All Admin Group Sub-TLV.

Based on SRLG constraint conditions

The constraint condition based on SRLG only includes one rule - the Exclude SRLG rule. This rule means that if a link belongs to the shared risk link group described by the Exclude SRLG rule, the link will be excluded by flexible algorithm during topology generation. The Exclude SRLG rule is advertised through the Flexible Algorithm Exclude SRLG Sub-TLV. Excluding links from the shared risk link group in the topology can improve the reliability of the calculated path.

Algorithm topology generation steps

The specific steps for generating a topology by the algorithm are as follows:

1.     If a node is not configured with FAD, it indicates that the node does not participate in flexible algorithm. In that case, the node is deleted by flexible algorithm.

2.     Check if there are Exclude-Any rules with affinity properties in the FAD. If Exclude-Any rules exist, flexible algorithm excludes links that have the same affinity property as the Exclude-Any rule.

3.     Check whether there exists a rule of Exclude SRLG in the FAD shared risk link group. If there is an Exclude SRLG rule, flexible algorithm will exclude links that belong to the shared risk link group described by the Exclude SRLG rule.

4.     Check if there is an Include-any rule for affinity properties in FAD. If an Include-any rule exists, flexible algorithm excludes any link that does not contain at least one affinity property from the Include-any rule.

5.     Check if there are Include-all rules for affinity properties in FAD. If Include-all rules exist, flexible algorithm excludes any links that do not have all the affinity properties of Include-all rules.

6.     If the metric type of FAD is not the link cost of IS-IS, and the link has not advertised the cost value of the metric type used by FAD, then the link is excluded by flexible algorithm.

As shown in Figure 7, without using flexible algorithm, all nodes calculate the optimal path to the destination based on the IS-IS link cost. When calculating the path using flexible algorithm 130, all nodes except Router F are configured with flexible algorithm 130. The FAD of flexible algorithm 130 is:

·     Use the SPF algorithm.

·     Calculate the minimum cost path to the destination based on the IS-IS link cost.

·     Exclude affinity attribute red.

Figure 7 Topology without the use of flexible algorithm

 

The topology of the generated flexible algorithm 130 is as shown in Figure 8. After the topology of flexible algorithm 130 is generated, each node calculates the path based on the algorithm type and optimization goal.

Figure 8 Topology of flexible algorithm 130

 

TI-LFA FRR of flexible algorithm

TI-LFA FRR (Topology-Independent Loop-free Alternate Fast Reroute) can provide link and node protection for path calculation of flexible algorithm. When a fault occurs at a link or node, traffic swiftly switches to the backup path for continued forwarding, thereby minimizing traffic loss to the greatest extent.

The operating mechanism of flexible algorithm TI-LFA FRR is largely the same as that of SRv6 TI-LFA FRR. For detailed information, see SRv6 Technology White Paper.

Calculation of TI-LFA FRR path for flexible algorithm

The method for calculating TI-LFA FRR paths and SRv6 in flexible algorithm is the same. The only difference is that TI-LFA FRR of flexible algorithm is based on FAD, and backup paths are calculated within the topology planned by flexible algorithm.

As shown in Figure 2-9, none of the nodes use flexible algorithm. Router B is the source node and Router A is the destination node, with digits in the link indicating cost values. The traffic path is from Router B to Router A. To avoid the loss of traffic due to link faults between Router B and Router A, the backup path calculated by TI-LFA is from Router B to Router C to Router E to Router F to Router A.

Figure 9 Backup path of TI-LFA that has not utilized flexible algorithm

 

As shown in Figure 10, all nodes are configured with flexible algorithm {algorithm} 130. The FAD for flexible algorithm {algorithm} 130 is:

·     Use the SPF algorithm.

·     Calculate the minimum cost path to reach the destination based on the IS-IS link cost.

·     Exclude links with "red" as their affinity property.

At this time, the TI-LFA calculated backup path is Router B -> Router C -> Router D -> Router E -> Router F -> Router A.

Figure 10 TI-LFA backup path with flexible algorithm

 

Tangent guard of flexible algorithm against microloops

Under normal circumstances, switchover refers to the rerouting of data traffic to the backup path after a network fault occurs. Convergence-induced loops by nodes adjacent to the fault node during switchover, are termed as switchover microloops. flexible algorithm supports the function to prevent microloops during switchover, by delaying the convergence time of the device to avoid loop formation.

As shown in Figure 11, all nodes in the network deploy flexible algorithm's TI-LFA, with FAD as follows:

·     Use the SPF algorithm.

·     Calculate the path with the least cost to the destination based on the IS-IS link cost. Assume that the cost value of all links is 10.

·     Exclude the link with the "red" affinity property.

Under normal circumstances, the path from Router A to Router C is Router A -> Router B -> Router C. When Router B encounters a fault, the traffic to Router C will switch over to the backup path calculated by TI-LFA. After Router A converges, the traffic to Router C will be forwarded along the converged path. At this moment, if Router D (or Router F) has not yet converged and still forwards traffic according to the pre-converged path, a loop will form between Router A and Router D (or Router F).

After enabling the micro-loop prevention function on Router A, it can delay convergence for a period of time. During this period, Router A does not respond to topology changes and still forwards traffic using the TI-LFA backup path, while other nodes in the network converge normally. After Router D and Router F have completed convergence, Router A starts converging. Once Router A convergence is done, the traffic to Router C is forwarded along the post-convergence path, preventing loops from occurring.

Figure 11 Preventing minor loop faults

 

Application of flexible algorithms in SRv6

Calculate the optimal path for SRv6 forwarding using flexible algorithm

Working principle of flexible algorithm in calculating the optimal path

When IS-IS is applied to SRv6 network construction, it uses LSPs with SRv6 Locator TLV to release SRv6 Locator information. Nodes in the SRv6 network will calculate the optimal route to reach the Locator network segment.

The SRv6 Locator TLV contains an algorithm field, which identifies the algorithm used for the Locator through the algorithm ID. By default, the algorithm ID in the SRv6 Locator TLV is 0, indicating that the node uses the SPF algorithm to calculate the optimal path to the SRv6 Locator prefix based on IGP link cost. When a Locator section is associated with a flexible algorithm, the algorithm ID in the SRv6 Locator TLV is the ID of the associated flexible algorithm, indicating that the node will use flexible algorithm to calculate the optimal path to the SRv6 Locator prefix. Different Locator sections can be associated with different flexible algorithms, calculating different forwarding paths or planning different forwarding topologies, making it convenient for users to flexibly plan traffic for various services.

For example, the SRv6 Locator is associated with flexible algorithm 130, and the FAD of this algorithm is as follows:

·     Use the SPF algorithm.

·     Calculate the minimum cost path to the destination based on the IS-IS link cost.

·     Exclude affinity attribute blue.

In flexible algorithm topology, Router A uses SPF to calculate the optimal path to reach the Locator prefix 2001:0:8:6::/64, as shown in Figure 12.

Figure 12 Flexible algorithm calculates the optimal path for SRv6 forwarding

 

Calculate the optimal path for SRv6 BE using flexible algorithm

As shown in Figure 13, L3VPN over SRv6 BE or L2VPN over SRv6 BE is deployed between PE 1 and PE 2 in the IPv6 network. User's second and third layer datagrams are encapsulated through the SRv6 tunnel and forwarded on the optimal path calculated by IGP. By using flexible algorithm, network resources can be fully utilized. For instance, links with high cost but relatively idle can be added to the forwarding path, thus enabling flexible planning of SRv6 BE forwarding path in the IPv6 backbone network.

Figure 13 SRv6 BE based on flexible algorithm

 

When applying the flexible algorithm for optimal path calculation in SRv6 BE, the release process of the private network route between CE 1 and CE 2 is the same as when flexible algorithm is not used. The difference lies in the IGP path calculation method to reach the destination Locator network segment. As shown in Figure 14, taking L3VPN over SRv6 BE as an example, the process to calculate the optimal path to reach the Locator network segment after using flexible algorithm is:

1.     PE 1, PE 2, P 1, P 2, P 3, and P 4 all use flexible algorithm 128. The FAD of this algorithm is:

¡     Use the SPF algorithm.

¡     Calculate the minimum cost path to the destination based on the IS-IS link cost.

¡     Include affinity attribute red.

2.     PE 2 associates this device's Locator section A5:80::/64 with flexible algorithm 128, and uses the IS-IS protocol to release this Locator section.

3.     After receiving the IS-IS route containing Locator section A5:80::/64, PE 1, P 1, P 3 and P 4 calculate the optimal path to the destination network segment A5:80::/64 based on the constraint conditions in flexible algorithm 128. The optimal path calculated is: PE 1 -> P 1 -> P 3 -> PE 2. After the packet from Site 1 to Site 2 is encapsulated with an IPv6 header, it is forwarded to PE 2 via this path by PE 1.

Figure 14 Route calculation process of SRv6 BE based on flexible algorithm

 

Calculate the optimal path for SRv6 TE policy using flexible algorithm

As shown in Figure 15, when SRv6 VPN uses SRv6 TE packet forwarding, the network administrator plans a path from PE 1 to PE 2 that needs to pass through P 4 (P 4’s SID is included in the SID list of possible paths) via SRv6 TE Policy. Without the deployment of flexible algorithm, the IS-IS protocol uses the SPF algorithm to calculate the optimal path based on link cost, which is: PE 1->P 2->P 4->PE 2.

Figure 15 SRv6 TE policy path without using the flexible algorithm

 

As shown in Figure 16, Locator A4:80::/64 on P 4 and Locator A5:80::/64 on PE 2 are both associated with flexible algorithm 128. The devices P 1, P 2, P 3, P 4, PE 1, and PE 2 in the network have all joined flexible algorithm 128. The FAD of flexible algorithm 128 is:

·     Use the SPF algorithm.

·     Calculate the minimum cost path to the destination based on IS-IS link cost.

·     Include affinity attribute red.

On PE 1, in the SRv6 TE Policy for reaching the destination node PE 2, the SID list of the optimal candidate path is <A4:80::1, A5:80::1>. When PE 1 calculates the packet forwarding path for this SID list, the path will no longer pass through P 2, but instead follow the forwarding path calculated based on the constraint conditions of flexible algorithm, which is PE 1 -> P 1 -> P 4 -> P 3 -> PE 2.

Figure 16 SRv6 TE policy path after using flexible algorithm

 

Flexible algorithm calculates the SID list for SRv6 TE policy candidate path created by ODN

After creating an On-Demand Next-Hop (ODN) template, if the Color extended community attribute of the received BGP route matches the Color value of the ODN template, then the next-hop address of this BGP route is used as the destination node address for the SRv6 TE Policy. The Color value of the ODN template is used as the Color property of the SRv6 TE Policy. Thus, an SRv6 TE Policy is automatically created. After the automatic creation of the SRv6 TE Policy by ODN, two candidate paths will be generated.

·     The candidate path has a preference of 200. If the ODN template is associated with flexible algorithm, the SID list under this candidate path is dynamically calculated by flexible algorithm.

·     The candidate path with a preference of 100 requires its SID list to be calculated by PCE.

As shown in Figure 17, on Router A, the ODN template is associated with flexible algorithm 130, with a FAD of:

·     Use the SPF algorithm.

·     Calculate the minimum cost path to the destination based on the IS-IS link cost.

·     Exclude affinity attribute blue.

When Router A receives a BGP route carrying the same Color value as the ODN template transmitted by Router F, it sets the next-hop address (Router F's address) as the destination node's address and automatically creates an SRv6 TE Policy. In the SRv6 TE Policy with a Preference of 200, the SID list calculated using flexible algorithm method is <A3:82::1, A4:82::1, A6:82::1>.

Figure 17 Flexible algorithm computes the SID list for the candidate path of the SRv6 TE Policy in the ODN mode

 

Typical network application

Topology planning based on flexible algorithm

With flexible algorithm, different forwarding topologies can be planned for different services in a network with SRv6 VPN to achieve traffic allocation and isolation, meeting various service requirements of tenants within a site. As shown in Figure 18, each intermediate node in the network joins the corresponding flexible algorithm, and PE 1 and PE 2 join flexible algorithms 198 and 199 at the same time, linking their own Locator sections with flexible algorithm. Different flexible algorithms calculate different network topologies, and flexible algorithms 198 and 199 separately calculate green and red topologies. On PE devices, the forwarding paths to opposite Locator sections are restricted within the corresponding topologies, that is, the topologies planned by flexible algorithm associated with the opposite Locator sections. This achieves the segregation of different business traffic to different network topologies for forwarding.

Figure 18 Topology planning based on flexible algorithm

 

Optimal path calculation based on flexible algorithm

The network described in Figure 18 distinguishes different network topologies for different services. Within each network topology, flexible algorithm can define metric types, calculating optimal traffic forwarding routes for varying business requirements. As shown in Figure 19 and Figure 20, the network topology planned by flexible algorithm 198 supports media services requiring low latency such as voice and video frequency (VF); the network topology planned by flexible algorithm 199 accommodates general data service. flexible algorithm 198 calculates the optimal path to the destination based on IS-IS link latency, while flexible algorithm 199 calculates it based on IS-IS link cost. Both algorithms have calculated different forwarding paths for the traffic they carry, meeting diverse traffic forwarding requirements.

Figure 19 Media business traffic forwarding path

 

Figure 20 Traffic forwarding path for ordinary data service

 

Related documentation

draft-ietf-lsr-flexible algorithm-16