1. LSP fragmentation expansion
Principle overview:
Generally, a router uses LSP to describe all its link state information. If the link state information is too large, the router must generate multiple LSP fragments to carry all link state information. It can be seen from the foregoing that the LSPID of each LSP is composed of the System-ID of the source router that generated the LSP and the ID of the pseudo node (the value is 0 in the ordinary LSP, and the value in the pseudo node LSP is not 0), LSPNumber (LSP points) The fragment number) is combined to uniquely identify. Since the length of the LSPNumber field is 1 Byte, the maximum number of fragments that an IS-IS router can generate is 256, and the amount of information carried is limited. According to RFC3786, IS-IS can configure a virtual System-ID and generate virtual IS-IS LSP messages to carry routing and other information.
The IS-IS LSP fragment extension feature allows IS-IS routers to generate more LSP fragments. By configuring additional virtual systems for the router, each virtual system can generate 256 LSP fragments (up to 50 virtual System), so that the IS-IS router can generate up to 13,056 LSP fragments.
basic concept:
- Original system (Originating System): The original system is a router that actually runs the IS-IS protocol. A single IS-IS process is allowed to advertise LSPs like multiple virtual routers, and "Originating System" refers to the "real" IS-IS process.
- System ID (Normal System-ID): The system ID of the initial system.
- Virtual System: A system identified by an additional system ID, used to generate extended LSP fragments. These fragments carry additional system IDs in their LSP IDs.
- Additional System ID (Additional System-ID): The system ID of the virtual system, which is uniformly assigned by the network administrator. Each additional system ID allows 256 extended LSP fragments to be generated.
LSP fragmentation method 1:
If there are routers in the network that do not support the LSP fragment extension feature, use Mode-1 fragment extension mode.
In this way, the link information to each virtual system is carried in the LSP issued by the initial system; at the same time, the LSP issued by the virtual system also contains the link information to the initial system, and the virtual system also participates in routing SPF Calculation. In this way, the virtual system looks like a real router connected to the initial system.
The LSP of the virtual system contains the same area address and Overload Bit as the original LSP. If there are TLVs with other characteristics, they must also be consistent.
The neighbor information carried by the virtual system points to the initial system, and the metric is the maximum value (the maximum value is 64 in the case of narrow metric) minus 1; the neighbor information carried by the initial system points to the virtual system, and the metric must be 0. This ensures that the virtual system will definitely become a downstream node of the initial system when other routers are performing route calculations.
As shown in the figure, R2 is a router that does not support fragment expansion, R1 is set to fragment expansion of mode-1, R1-1 and R1-2 are virtual systems of R1, and R1 puts part of the routing information into R1-1 and R1-2 is sent out in LSP packets. When R2 receives packets from R1, R1-1, and R1-2, it considers that there are three independent routers on the opposite end and performs normal routing calculations. At the same time, the cost of R1 to R1-1 and R1-2 are all 0, so the cost of the route from R2 to R1 is equal to the cost of the route from R2 to R1-1.
LSP fragmentation method 2:
Mode-2 is used when all routers in the network support the LSP fragmentation extension feature. In this mode, the virtual system does not participate in the routing SPF calculation, and all routers in the network know that the LSP generated by the virtual system actually belongs to the initial system.
If R2 supports fragment expansion and R1 is set to Mode-2 fragment expansion, R1 will put part of the routing information into the LSP packets of R1-1 and R1-2 to send out (R1-1 and R1 -2 is the internal virtual node). When R2 receives the LSPs of R1-1 and R1-2, knows that their initial system is R1 through IS Alias ID TLV, and regards the information released by R1-1 and R1-2 as R1 information, so in R1-1 and R1-2 will not exist in the topology calculated by R2, as shown in the following figure:
2. IS-IS measurement and expansion
ISO 10589 defines the following four metrics for IS-IS protocol:
- Default metric: It is our common metric type, which is generally inversely proportional to the interface bandwidth. By default, all routers in the routing domain must support this metric.
- Delay metric: an optional metric used to indicate the delay of link packet transmission.
- Cost measurement: an optional measurement used to represent the cost of link packet transmission.
- Error metric: An optional metric used to indicate errors in link packet transmission.
Currently, Huawei's VRP system only supports default metrics. Delay, overhead, and error metrics are mainly used to support QoS routing scenarios. According to usual usage and terminology, the metrics or cost values mentioned below refer to the default metrics unless otherwise specified. By default, the default cost of the IS-IS interface of Huawei equipment is 10. The cost can also be automatically calculated based on the interface bandwidth. The cost of a path refers to the total cost of all links that the route passes through from the source to the destination. ISO 10589 specifies that the total cost of a path is 1023, so the cost value should be planned reasonably in the network. The following figure shows the routing prefix and its metric value carried in the IP internal reachability TLV (type 128) of an LSP:
In the original ISO 10589 definition, the default metric field is only 8 bits long, of which the 8th bit is The reserved bit is set to 0; the 7th bit is used to indicate whether the route is from inside or from outside the routing domain, set to 0 to indicate internal routing, and set to 1 to indicate external routing. In this way, only 6 bits are left to represent the metric value, ranging from 0 to 63. This metric is also called Narrow Metric. In the later definition of
RFC1195, the metric definition method of ISO 10589 was directly borrowed . This RFC specifies the metrics carried in the IP reachability TLV integrated with IS-IS. Narrow metrics can be used in the following TLVs:
- IP internal reachability TLV (type 128): used to carry IS-IS routing information in the routing domain.
- IP external reachability TLV (type 130): used to carry IS-IS routing information outside the routing domain.
- IS neighbor TLV (Type 2): used to carry neighbor information.
With the expansion of the network scale and the demands of new applications on the network, the measurement range is too small to meet actual needs. Therefore, a longer metric field is defined in RFC3784. This new metric is used in the following two newly defined TLVs:
- Extended IP reachability TLV (type 135): used to replace the original IP internal or external reachability TLV, carry IS-IS routing information, and can carry sub TLV.
- Extended IP Neighbor TLV (Type 22): It is an extension of Type 2 TLV, used to carry neighbor information.
Compared with the previous 6-bit long metric value field, the type 135 TLV uses a 32-bit metric value field, so the maximum metric value of a route can reach 4261412864; the type 22 TLV uses a 24-bit metric value field, so an IS The maximum metric value of the -IS interface can be extended to 16777215. This type of metric is called Wide-metric.
The Huawei VRP system uses narrow metrics by default. You can use the command cost-style to modify the metric type. The metric can be configured to one of the following types according to the specific situation:
- Compatible (Compatible Metric): The route sent and received by the device can either use narrow metrics or width metrics.
- Narrow (narrow metric): The route sent and received by the device can only be a narrow metric.
- Narrow-compatible (compatible with narrow metric): The route sent by the device uses a narrow metric, and the received route can use a narrow metric or a width.
- Wide: The route sent and received by the device can only be wide.
- Wide-compatible: The route sent by the device uses the width, and the received route can use the narrow metric or the width.
Three, ISIS management mark
As shown in the above figure, RouterA needs to communicate with RouterB, RouterC, and RouterD in other Level-1 areas. To ensure information security, routers in other Level-1 areas cannot receive the message information sent by RouterA.
First, you can configure the same management tag value tag for the IS-IS-enabled interfaces of RouterB, RouterC, and RouterD. Then, when the Level-1-2 router in Area 4 performs route penetration from the Level-2 to the Level-1 area, it should match the specified tag. In this way, it can be satisfied that when RouterA communicates with other Level-1 areas, it only communicates with RouterB, RouterC, and RouterD. At this time, the topology formed on RouterA is shown in the following figure:
Management tag values are associated with certain attributes. When cost-sytle is wide, wide-compatible, or compatible, if the advertised reachable IP address prefix has this attribute, IS-IS will add the management tag to the IP reachability information TLV of the prefix. In this way, the management tag will be distributed to the entire routing domain along with the prefix.
4. IS-IS neighbor flap suppression (same as OSPF)
Cause:
If the status of the interface carrying IS-IS services switches between Up and Down, it will cause frequent oscillations of neighbor status. At this time, IS-IS will quickly send Hello messages to re-establish neighbors, synchronize the database LSDB, and trigger routing calculations, which will cause a large number of message interactions, affect the stability of existing neighbors, and have a greater impact on IS-IS services. It will also affect the normal operation of other services that rely on IS-IS (such as LDP and BGP). In order to solve this problem, IS-IS implements the neighbor oscillation suppression function, that is, when the neighbor frequently oscillates, the oscillation suppression is activated to delay the establishment of the neighbor, or the service traffic delays through the frequently oscillating link to achieve the purpose of suppressing the oscillation.
Basic concepts:
flapping_event: flapping event, the last time the neighbor status on the interface switches from Up to Init or Down, it is called flapping_event. flapping_event is used as the oscillation source input to trigger the oscillation detection mechanism to start work.
flapping_count: current flapping times.
detect-interval: Oscillation detection interval, used to determine whether a valid oscillation event is triggered.
threshold: Oscillation suppression threshold. When the accumulated effective oscillation event triggers exceeds this value, the oscillation suppression phase is entered.
resume-interval: Resume interval. When two consecutive effective oscillation events exceed this value, the oscillation suppression phase is exited.
Implementation principle: The
IS-IS interface starts a flapping_count counter, and the interval between two adjacent flapping_event generation times is within the detect-interval, which is recorded as a valid flapping event. The flapping_count count is increased by 1. When the flapping_count count is greater than the threshold, the system determines that the flapping has occurred and needs to enter the flapping suppression stage. After entering the shock suppression phase, the flapping_count is cleared to 0. Before the flapping_count is greater than the threshold, if the interval between the two flapping_events is greater than the resume-interval, the flapping_count is cleared to 0. Neighbor flapping suppression starts timing from the last time the neighbor status changed to Init or Down.
Users can configure the three key parameters of oscillation detection: detect-interval, threshold, and resume-interval through the command line.
Shock suppression is divided into two modes: Hold-down and Hold-max-cost:
- Hold-down mode: In view of the frequent flooding and topology changes during the neighbor establishment process, the neighbor is prohibited from re-establishing for a period of time to avoid frequent database synchronization and a large number of message interactions.
- Hold-max-cost mode: To address the frequent switching of user service traffic, set the link cost value to the maximum value Max-cost within a period of time (Max-cost in IS-IS Wide mode=16777214, IS-IS Narrow mode Max-cost=63) to prevent the user’s business traffic from passing through frequently oscillating links.
Hold-down mode and Hold-max-cost mode can be superimposed and used. When it takes effect at the same time, enter Hold-down mode first, and then enter Hold-max-cost mode after exiting Hold-down mode.
By default, IS-IS enables the Hold-max-cost mode. Users can modify the oscillation suppression scheme and oscillation suppression period through the command line. After an interface enters the flapping suppression phase, all neighbors on the interface will enter the flapping suppression phase.
5. ISIS and BFD linkage: similar to OSPF
Overview:
Generally, IS-IS sets the interval for sending Hello packets to 10 seconds. Generally, the time for announcing that the neighbor is Down (that is, the neighbor's hold time) is configured to be 3 times the Hello packet interval. If the neighbor does not receive the Hello message from the neighbor within the failure time of the neighboring router, the neighbor will be deleted. It can be seen that the minimum time for the router to detect a neighbor failure is at the second level. This may cause a large number of packets to be lost in a high-speed network environment.
Bidirectional Forwarding Detection (BFD) can provide light-load and fast (millisecond-level) channel failure detection, which solves the problem of insufficient IS-IS detection mechanisms. Using BFD is not to replace the Hello mechanism of the IS-IS protocol itself, but to cooperate with the IS-IS protocol to find out the faults in the adjacent connection more quickly, and promptly notify IS-IS to recalculate related routes to correctly guide the forwarding of packets.
When the link on the primary path fails, BFD can quickly detect the failure and notify the IS-IS protocol. IS-IS Down drops the interface neighbors of the failed link and deletes the corresponding IP protocol type of the neighbor, thereby triggering the topology calculation. Updating the LSP enables other neighbors (such as RouterC) to receive the updated LSP from RouterB in time, thus achieving rapid convergence of the network topology.
IS-IS and BFD linkage includes IS-IS linkage with static BFD and IS-IS linkage with dynamic BFD:
Static linkage experiment:
Environment: R4 and R5 are separated by a switch, and BFD is used to detect non-direct links
Configure on R4:
ISIS part:
isis 1 //Enter the ISIS process
is-level level-1 //Is the level-1 router
network-entity 49.0001.0000.0000.0004.00 //Configure the network identity
interface GigabitEthernet0/0/0 //Enter the interface to enable ISIS
ip address 172.16.0.1 255.255.255.0
isis enable 1
BFD part:
Bfd //Enable bfd
bfd 4to5 globally bind peer-ip 172.16.0.2 interface GigabitEthernet0/0/0 //Bind the IP address of the other party's interface with its own interface
discriminator local 1 //local identifier
discriminator remote 2 / /Remote ID
commit Commit configuration
The configuration on R5 is omitted, mirroring the relationship with R4, and then using a command to check whether the static BFD neighbor relationship is established successfully: Successful
dynamic linkage experiment: The
three devices establish a BFD session connection with each other. Take the configuration on R2 as an example:
Bfd //Enable bfd
isis 1 globally //Enter the isis process
is-level level-1
bfd all-interfaces enable //Enable all interfaces enabled by isis bfd
network-entity 49.0001.0000.0000.0001.00
interface GigabitEthernet0/0/1 //Enable isis on the corresponding interface
ip address 10.13.13.1 255.255.255.0
isis enable 1
interface GigabitEthernet0/0/1
ip address 10.13.13.1 255.255.255.0
isis enable 1
View BFD neighbor relationship: successfully established
Six, ISIS NSR
With the rapid development of networks today, users have increasingly demanded applications such as data, video, and voice, and operators have also put forward higher requirements for the reliability of IP networks. When a node in the network fails, or the main/standby switch is artificially performed during the maintenance process, the equipment may not be able to construct routing information, resulting in traffic loss or even network paralysis. The deployment of NSR (Non-Stop Routing) can solve the above problems and provide users with high reliability guarantee for uninterrupted forwarding of key services.
The IS-IS NSR feature ensures that the standby board can quickly take over the services of the original main control board through the high synchronization between the active and standby IS-IS real-time data, so that the neighbors do not perceive the equipment failure. After the active/standby switchover, the new active main control board can use these real-time data to quickly restore the protocol, so that neighboring devices are not aware of the active/standby switchover. IS-IS NSR is mainly realized by backing up the following data:
• Configuration data: All configurations completed by the user, including neighbor information, timer parameter information, and configuration information under the process.
• Dynamic data: including interface parameters and status, neighbors, LSDB and other information.
七、IS-IS Auto FRR
IS-IS Auto FRR uses the LFA (Loop-Free Alternates) algorithm to pre-calculate the backup link and joins the forwarding table together with the main link. When the network fails, IS-IS Auto FRR can quickly switch traffic to the backup link before the control plane route is converged to ensure that the traffic is not interrupted, thus achieving the purpose of protecting the traffic, thus greatly improving the IS-IS network The reliability.
The basic idea of LFA calculating the backup link is: take the neighbor that can provide the backup link as the root node, and use the SPF algorithm to calculate the shortest distance to the destination node. Then, a loop-free backup link is calculated.
IS-IS Auto FRR supports filtering the backup routes that need to be added to the IP routing table. Only the backup routes that pass the filtering strategy will be added to the IP routing table. Therefore, users can more flexibly control the IS-IS backup routes added to the IP routing table. .
The collated information comes from: "HCIE Routing and Exchange Learning Guide", Huawei hedex document