CN1153417C - Routing protocol testing method and system based on punch-through-end network model - Google Patents

Abstract

The routing protocol testing method and system based on the punch-through-end network model belong to the field of Internet large-scale routing technology. The test method is aimed at an actual router under test. Each node of the pass-through-end network topology model represents a router, and each edge represents an actual link between routers. Use requirements, find an appropriate node in the above topology structure to represent the above-mentioned router under test, so that all other nodes can exchange routing information with the above-mentioned router under test, so that the above-mentioned router under test can perform routing in such a large-scale network environment Protocol testing. Compared with the existing technology, it can provide the Internet large-scale routing test function based on the complete network model.

Description

Routing protocol testing method and system based on punch-through-end network model

technical field

The routing protocol testing method and system based on the pass-through-end network model belong to the field of Internet large-scale routing technology, and in particular relate to a testing technology for routing protocol implementation, Internet topology simulation, and large-scale routing interaction.

Background technique

Routing technology is the core of the Internet, and the research work on routing is usually based on theoretical analysis and simulation experiments. For example, Sidhu et al. showed through simulation experiments that when the input queue is limited, in the Open Shortest Path First OSPF (Open Shortest Path First) routing protocol, the election of representative routers and the competition in the flooding process will cause jitter. Under the topological structure of an actual network service provider, Basu et al. studied the stability of the OSPF protocol in detail through simulation experiments. Based on the queue model and online simulation experiments, Ye et al. studied the dynamic optimization method of OSPF metrics. Shaikh et al. used a mathematical model to quantitatively analyze the impact of network congestion on routing stability, and used a tester to verify the analysis results under a simple topology model. In addition, the researchers also designed and implemented a test bed to provide a platform for evaluating and comparing routing algorithms. Before using the test bed, it is necessary to set the routing algorithm and network load used, and configure the nodes and links of the network one by one. Generally speaking, the above-mentioned studies are either based on theoretical analysis or simulation experiments based on models, lacking the measurement and analysis of actual systems. Whether the result is applicable to the actual system still needs to be further verified by practice.

In addition, part of the research is based on measuring the actual Internet. For example, after 2 years of passive measurement and active error injection, Labovitz et al. found that after an Internet inter-domain routing error, it takes tens of minutes to achieve routing consistency again. However, this kind of research method not only needs to invest huge manpower and material resources, but also may cause certain damage to the existing network, so it is not suitable for large-scale and extensive development.

In order to verify the characteristics of the real system, an effective method is to build a test system and interconnect it with the actual system under test for testing. For example, a network emulator provides a means of IP layer network emulation using a fast parallel discrete event simulation core. Similar to this, another emulator can set a fixed delay to interact with the actual system when simulating the transmission of IP packets from source to destination. Based on the Trace Modulation technology, Noble et al. studied the reproduction method of the end-to-end characteristics in the actual wireless network, and ensured that this technology is transparent to the data sent and received by the system under test. The virtual network interconnection test bed is mainly oriented to the network transmission control protocol supported by the network simulator. These studies focus on the forwarding of IP packets in the network, but lack of consideration of routing packets, let alone the interactive characteristics of routing protocols.

Among the many studies above, there is no precedent of combining routing protocol and network topology model for routing protocol testing. Research on routing protocols often starts from the perspective of simulation, while research on network topology models only focuses on end-to-end characteristics and lacks consideration of the routing characteristics of network core equipment. Although some commercial tester products include routing protocol testing functions, they cannot provide perfect Internet routing testing functions due to the lack of support for the Internet topology model and the lack of cooperation between the various ports of the test system.

Contents of the invention

The object of the present invention is to provide a routing protocol testing method and system based on a pass-through-end network model for routing protocol testing.

The method proposed by the invention is characterized in that it is a router routing protocol testing method based on a pass-through-end network topology model with a distributed real-time operating system as a development platform. For an actual router under test, each node of the pass-through-end network topology model represents a router, and each edge represents an actual link between routers, and then according to the number of ports of the above-mentioned router under test and actual use requirements, Find an appropriate node in the above topology structure to represent the above-mentioned router under test, so that all other nodes can exchange routing information with the above-mentioned router under test, so that the above-mentioned router under test can perform routing protocol testing in such a large-scale network environment. It contains the following steps in order:

(1) According to the actual router under test, use the pass-through-end network topology model to configure the network scale to be simulated and the interconnection relationship between each layer;

(2) Simulate and generate a corresponding large-scale network topology model;

(3) according to the port number and the node type of the above-mentioned tested router, locate the layer where the above-mentioned tested router is located;

(4) judge whether there is above-mentioned tested router satisfying condition in above-mentioned network topology model, if not, then return to step (2);

(5) Otherwise, the above-mentioned network topology model is mapped to an Internet structure composed of multiple autonomous systems AS and each autonomous system AS internally uses the Open Shortest Path First (OSPF) protocol;

(6) According to the requirements of the OSPF protocol, the Type 5 link state advertisement LSA generation subsystem converts the link state information outside the autonomous system into a Type 5 link state advertisement, and sends it to the node directly connected to the router under test The corresponding protocol test subsystem;

(7) Select a port of a node connected to the pass-through domain node at the location of the above-mentioned router under test as the current port;

(8) In the AS where the router under test is located, the Type 4 LSA generation subsystem generates Type 4 link state advertisement LSAs for each AS border router, and sends these LSAs to the protocol test sub-system corresponding to the current interface system;

(9) Determine the corresponding other routers in the AS for the current interface, and the LSA generation subsystem of the first to third types of link state advertisements forms the link state advertisements of the first to third types according to the protocol requirements, and sends them to the corresponding protocol test subsystem;

(10) each link state advertisement subsystem sends an LSA description packet to the above-mentioned router under test through its respective protocol testing subsystem;

(11) After the above-mentioned router under test receives the corresponding description packet, it sends the LSA request packet, and after the protocol testing subsystem receives the request packet, it sends the corresponding LSA update packet, and starts timing each link state update process simultaneously;

(12) After the protocol test subsystem receives the corresponding LSA confirmation packet, the timing ends;

(13) process all ports of the above-mentioned tested router by steps (7)～(12);

(14) Statistical average delay time and output;

(15) END.

Wherein the mapping relationship between the pass-through-end domain topology model TS described in step (5) and the Internet structure is as follows:

TS Topology Model - BGP, OSPF-oriented Internet

Entire Topological Model - The Internet

A pass-through domain and all stub domains it connects to - an autonomous system

Links connecting different pass-through domains——BGP interactive links

Passthrough domain——OSPF backbone area

Stub domain—OSPF non-backbone area

Nodes connecting different pass-through domains - AS border routers

Pass through the nodes in the domain - OSPF domain border router

Nodes in the stub domain—OSPF intra-area routers

Links connecting different stub domains——OSPF virtual links.

The system of the routing protocol testing method based on the punch-through-end network model is characterized in that it is a kind of Internet routing testing system IRTS, which contains: a topology generating subsystem; controlling the above topology generating subsystem and being connected to its data output terminal The selected subsystem of the router under test; a total of 5 types of link state advertisement generation subsystems LSA1-LSA5 connected to the data output terminals of the above-mentioned router selection subsystem; respectively connected to the data output terminals of the above-mentioned LSA generation subsystem And N protocol test subsystems that interact with the router under test for routing protocols, where N is the number of ports of the router under test; respectively control the above-mentioned topology generation subsystem, the router under test selection subsystem and the N protocol test subsystems Operate and manage the OAM subsystem.

Experiments prove that it achieves the expected purpose: to realize the test of routing protocol based on the network model.

Description of drawings

Figure 1. Pass-through-end network topology model

In Figure 1: 1 is the punch-through domain; 2 is the terminal domain; 3 is the terminal domain connecting multiple punch-through domains.

Figure 2. Internet topology diagram

Figure 3. Detailed diagram of node a in Figure 2: IF is the interface.

Figure 4. Overall flowchart of Internet routing.

Figure 5. Describing the overall flow chart of Internet routing used in the example.

Figure 6. The overall structure of the Internet routing test system IRTS and its connection diagram with the router under test RUT.

Figure 7. High-dimensional ordered federated data structure diagram.

Figure 8. The simulated network topology diagram for an actual example.

Figure 9. Response latency versus network size.

Figure 10. High-dimensional curvilinear regression in logarithmic scale.

Figure 11. RUT selection algorithm flow chart.

Detailed ways

Before the actual network equipment (router) is put into operation, it is often necessary to carry out an actual test according to the method it uses. However, if it is directly tested in an actual large-scale network environment, once the router does not meet the requirements, it may have a destructive impact on the actual network. In order to solve this problem, based on the idea of combining Internet network topology simulation with routing protocol implementation, we designed and realized the Internet routing testing system IRTS (Internet Routing Testing System), the flow chart of which is shown in Figure 4.

For an actual router under test (Router Under Test abbreviated as RUT), we first simulate a large-scale network topology similar to the actual network, where each node represents a router, and each edge represents a link between routers . Then, we will see how the RUT will be actually applied in the actual environment, including the number of ports of the RUT and the network level where the ports are located. Then configure our system according to the actual application requirements of RUT, so as to find an appropriate node in the simulated network to represent the RUT, and the routing test system simulates all other nodes (each node represents a router) To exchange routing information with RUT. In this way, from the perspective of RUT, RUT itself is in a large-scale network environment, but does not know that this is a simulated environment. Therefore, if the RUT can run normally and stably in this simulated large-scale network environment, it means that the RUT can also run normally in the actual network according to this application method.

Below, we first illustrate how the system works according to the flow chart through a simple example. An example is shown in Figure 8. First configure the network topology to be simulated: the number of pass-through domains Tranist=1; the average number of nodes in each pass-through domain Tnode=1; the average number of stubs connected to each pass-through domain node Stub=2 ; The average number of end domain nodes contained in each end domain Snode=1. In this way, the total number of nodes included in the topology graph is: (Transit*Tnode)*(Stub*Snode+1)=3, that is, the network topology graph shown in FIG. 2 . Then we configure the type of RUT to have two ports belonging to different areas, then we can only select node a as the RUT, and then we simulate the routing behavior of node b and node c respectively, and send routing information to node a, so that Node a is in our simulated network topology to realize network routing test.

As a research prototype system, IRTS currently only implements OSPF (Open Shortest Path First) protocol testing. The system structure is shown in Figure 6, including the Operation And Management (OAM) subsystem, topology generation subsystem, and RUT selection Subsystem, link state advertisement (Link State Advertisement, referred to as LSA) generating subsystem and protocol testing subsystem.

The OAM subsystem controls, coordinates, and configures other subsystems according to user requirements, including parameters of random network topology, RUT parameters, OSPF configuration, etc. Topology generation is based on an appropriate network topology model, simulating a hierarchical random network topology map with practical significance, thus representing the current Internet topology to a certain extent. In order to improve the scalability of the system (such as adopting other topological models in the future), the results of topology generation are stored in GB format. The RUT selection subsystem reads the topology map stored in GB format, and selects the appropriate node in the topology map as the RUT according to the OAM configuration, and configures the IP addresses of each port of all nodes at the same time. If there is no node satisfying the configuration condition of the RUT in the current topology graph, the topology generating subsystem is notified to regenerate the topology graph. The LSA generating subsystem generates a specific LSA for each node directly connected to the RUT according to the specific position of the RUT in the topology map, and sends it to the corresponding protocol testing subsystem. Multiple protocol test subsystems run simultaneously in IRTS, and each protocol test subsystem represents a node. On the basis of realizing routing protocol interaction, it sends the LSA generated by the LSA generation subsystem to RUT. Among them, each protocol test subsystem independently simulates the routing behavior of a router and completes the corresponding routing interaction functions, so it is a router with complete functions in the view of RUT. On the basis of normal routing interaction of each protocol testing subsystem, an upper layer subsystem (LSA generation subsystem) controls sending simulated LSA, so that each protocol testing subsystem simulates a network.

Next, we describe the following two issues in detail:

1. Network topology mapping problem

In order to improve the scalability of routing, the current Internet mainly adopts a hierarchical routing architecture: firstly, the Internet is divided into multiple autonomous systems AS (Autonomous System), and Border Gateway Protocol (BGP for short) is mainly used between AS; Internally use OSPF protocol or Routing Information Protocol (RIP for short). The RIP protocol is applied in a small network. This article mainly discusses the topology corresponding to the BGP and OSPF routing protocols. For the AS adopting the OSPF protocol, it is usually divided into several areas (Area), and the nodes (backbone nodes) in the backbone (Backbone) area are connected to each area of the lower layer. In this way, the Internet is artificially divided into three layers: AS, area, and node. As shown in Figure 2, there are five areas in the autonomous system AS1: A0 to A4, where A0 is the backbone.

It should be pointed out that BGP and OSPF divide the network in different ways: for BGP, each node belongs to an AS; for OSPF, each port of a router belongs to an area. The so-called port of a router is the socket or outlet used by the router to connect to other routers. For example, in Figure 2, node c belongs to AS1; among the four ports IF1, IF2, IF3, and IF4 connecting nodes b, c, d, and e to node a, IF1 and IF2 belong to area A0, while IF3 belongs to A2, and IF4 belongs to A1 , so node a does not belong to a specific area, such a node is called an area border router, that is, multiple ports of the node belong to different areas.

Different topology models lead to different simulation and generated topology graphs, and the performance of routing protocols and algorithms may depend on the applied topology environment. Therefore, in order to test the routing characteristics in the real network environment, it is necessary to select the same topology model as the Internet topology. Currently commonly used network topology models include the following: 1) simple and regular topological structures, such as star connection, ring connection, tree connection, grid structure, etc.; 2) well-known topological structures, such as ARPAnet, NFSnet, MCI backbone domain, etc.; 3) randomly generated topology. As the Internet continues to develop freely, its structure is very different from any specific structure; at the same time, in order to simulate the random dynamics of the Internet, we choose the third type, that is, the random topology model. GT-ITM is a typical random topological graph generation tool, which can generate random topological graphs based on planar random model, N-level model and through-end model. Among them, as shown in Figure 1, the Transit-Stub TS model is more representative of the current Internet structure, so we choose this model.

Each dark ellipse in the figure represents a Transit Domain, and each light circle represents a Stub Domain, thus providing a hierarchical Internet structure. The model can be set with the following parameters: the number T of the pass-through domain; the average number N1 of the pass-through domain nodes contained in each pass-through domain; the average number K of the end domains connected to each pass-through domain node; The average number N2 of domain nodes. In this way, the total number of nodes included in the topology graph is: (T×N1)×(K×N2+1). When the degree of nodes is not too large, the number of nodes is equivalent to the number of subnets in the network.

This model provides a topology map with a three-layer structure, each layer itself can be controlled by a planar random graph control method, and supports the control of various configurations and interrelationships of the pass-through domain and the terminal domain, and has a strong Model and parameter setting capabilities.

According to different levels, the Internet topology can be divided into many types. For example, from the perspective of the data link layer, the Internet has a flat structure, from the perspective of BGP and RIP, the Internet has a two-layer structure, and from the perspective of BGP and OSPF, the Internet has a three-layer structure. Therefore, it is necessary to map the above TS model to the BGP and OSPF-oriented Internet topology, as shown in Table 1. Through such correspondence, and removing the end domains that connect multiple pass-through domains at the same time, we have completed the TS topology model to the Internet topology.

Table 1 Correspondence between TS model and Internet TS topology model Internet facing BGP and OSPF whole topology model Internet A passthrough domain and all terminal domains connected to it an autonomous system Links connecting different passthrough domains BGP interaction link punch-through domain OSPF backbone area terminal domain OSPF non-backbone area Nodes connecting different passthrough domains AS border router Nodes in the punch-through domain OSPF Domain Border Router nodes in end domain OSPF intra-area router Links connecting different terminal domains OSPF virtual link

Flutter structure conversion.

2. The problem of node selection under test

In order to test the routing interaction characteristics of RUT in the Internet, IRTS must be configured according to the actual use of RUT, including the number of RUT ports, the status of each port in the Internet (such as the region it belongs to), and IP address. For example, if the RUT is required to have four ports, and there are two ports in the backbone area and one port in each of the other two areas, then node a in Figure 2 can be selected as the RUT. At this time, IRTS runs four protocol test subsystems to simulate the routing behavior of nodes b, c, d, and e respectively, so as to realize the network topology information exchange of the entire Internet at the routing level.

When performing RUT selection, a high-dimensional ordered linked list data structure as shown in Figure 7 is used. For a certain node N in the topology map, check all its ports and sort them according to the area connected to this node: connect all ports belonging to the same area with a linked list, and then sort the areas according to the number of ports they have (the backbone area is ranked first), thus obtaining the high-dimensional ordered linked list L _N of the node N. The RUT selection algorithm is shown in Figure 11, where CAL(N) is the high-dimensional ordered linked list L _N of computing node N; when the parameter is RUT, the calculation result is the high-dimensional ordered linked list L _RUT corresponding to RUT. Similar to the method of calculating L _N , first configure the RUT according to the actual use of the RUT, that is, configure the number of areas in the RUT and the number of ports in each area, so as to calculate the high-dimensional ordered linked list L _RUT corresponding to the RUT. Then compare with L _N corresponding to each node in the topology graph until a node _LRUT matches L _N , that is, L _RUT and L _N contain the same number of areas, and the number of ports in each area is the same. The computational complexity of the function CAL(N) is O(If ² ), and the total complexity of the RUT selection algorithm is O(K×If ² ), where K is the number of nodes in the topology graph, and If is the The maximum number of ports. Since the topological graph usually has a large number of similar nodes, it is often possible to find a node suitable for the RUT configuration without traversing all the nodes, so the above-mentioned complexity is the time complexity of the worst case (RUT selection loss).

After selecting an appropriate node as the RUT, it is also necessary to configure IP addresses for the ports of each node in the entire network. Because OSPF does not use the IP address aggregation strategy, for simplicity, the user only needs to configure the IP addresses of the RUT and IRTS ports, and the internal addresses of the network model are automatically generated. The algorithm first sets the IP address of the RUT and its adjacent ports, then checks each node one by one, and sets the port of the node and the IP address of the directly connected opposite port. In order to save address space and meet the needs of large-scale network models, each link uses a subnet with a mask of 255.255.255.252, and the two ports of the link use two available host addresses in this subnet respectively.

Based on the above technologies, Tsinghua University successfully developed a distributed Internet routing test system IRTS, and realized the strength test. The system uses a distributed multiprocessor based on the Compact PCI bus, and chooses the distributed real-time operating system VxWorks as a development platform. Its main control board uses Motorola MPC750 processor to run OAM subsystem, topology generation subsystem, RUT selection subsystem and LSA generation subsystem. The line card uses Motorola 860 processor, which mainly runs the protocol test subsystem. The line card uses FPGA to implement various Ethernet and WAN ports, including Gigabit Ethernet ports, 100M Ethernet ports and 2M synchronous ports.

Using the designed and realized IRTS as the experimental bed, we have carried out a strength test on the CISCO2600 router by testing the routing interaction process when the CISCO2600 router starts in the Internet. The test process assumes that when the RUT is started, other routers in the network interact in good condition, and each router has a corresponding routing table and link state information base. The test process realized large-scale high-speed OSPF routing interaction between the entire network and RUT. By measuring and analyzing its interactive characteristics, we obtained the upper limit of computational complexity and capacity support of RUT's OSPF implementation. Experiments show that IRTS has achieved the system design goal very well.

Figure 9 shows the relationship between the link state advertisement (LSA) response delay and the network size during the test. The Y axis represents the mathematical expectation E(D) of the delay D between sending an LSA and receiving the corresponding response from the IRTS system, and the X axis N represents the network scale used, that is, the number of subnets.

First, analyze the graph qualitatively. When the scale N<4000, Y=E(D)=F(N) is a decreasing function of the independent variable N. This shows that according to the protocol standard, the OSPF implementation of the RUT performs a delayed acknowledgment (DelayedAcknowledge) after receiving the link state information. In this response process, when the RUT receives an LSA, it does not immediately send an Ack packet response, but waits for a certain time interval or receives a certain number (response threshold) of LSAs, and then generates a response message confirmation These LSAs save bandwidth and increase network processing capabilities. However, due to the limited number of LSAs, at the end of the interaction process, there will always be some unanswered LSAs whose number does not reach the RUT response threshold, so the RUT will be in a waiting state, expecting to receive more LSAs, until timeout. The delay of this part of LSAs is relatively large, and we call such LSAs with a large delay due to delayed responses as LSAs in the waiting state. However, because the receiving speed of the previous LSAs is very fast, the waiting time is negligible compared with the processing time of the RUT, so it can be considered that these LSAs have no waiting delay. As the number of LSAs increases, the ratio of LSAs in the waiting state to all LSAs decreases continuously, so E(D) decreases as N increases.

When N continues to increase, because the RUT continues to operate the link state database, calculate the shortest path tree, and the calculation scale is also expanding, the load on the RUT processor increases, and the processing time increases, which means that the received LSA cannot be confirmed in time , so E(D) increases as N increases.

Next, a statistical model is used to quantitatively analyze the data in Figure 9. Due to the large variation range of N, first take the logarithm of N, make the abscissa X=ln(N), and the ordinate Y=E(D), and fit the curve by the least square method on the logarithmic coordinate graph, such as Figure 10 shows. Quadratic, cubic and quartic curve regressions were performed in the figure respectively. It can be seen that the quartic curve regression is closer to the test data, indicating that RUT has high scalability and can better adapt to large-scale network environments, and its performance is E(D)∝O((lnN) ⁴ ). The quartic equation is:

E(D)＝0.0160(lnx) ⁴ -0.4082(lnx) ³ +3.8641(lnx) ² -16.4158(lnx)+28.8157

Extrapolating from the above empirical formula, the mathematical expectation of the LSA response delay time of the RUT when N is other values can be estimated. For example, when N=43,000, E(D)=5.10s. As the protocol standard stipulates, if no confirmation information is received within 5 seconds, the update packet will be resent, so at this time, a large number of LSAs will be retransmitted because the processing speed of the RUT cannot keep up. If this happens, it will not only waste a lot of bandwidth, but also cause a serious burden on the RUT, causing the efficiency of the RUT to drop sharply or even fail. Therefore, it can be considered that this is the upper limit of the number of routing table entries that the RUT can bear.

It can be seen that the present invention achieves the expected purpose of realizing the routing protocol test based on the network model.

Claims (3)

1. The routing protocol testing method based on the punch-through-end network model uses the randomly generated feed-through-end topology model TS, which is characterized in that it is a distributed real-time operating system as a development platform based on the punch-through-end network topology The router routing protocol testing method of the model, it aims at an actual tested router, each node of the pass-through-end network topology model represents a router, and each edge represents the actual link between routers, and then according to the above-mentioned tested router The number of ports and the actual use requirements, find an appropriate node in the above topology to represent the router under test, so that all other nodes can exchange routing information with the router under test, so that the router under test can operate on such a large scale Routing protocol testing in a network environment; it contains the following steps in sequence: (1) According to the actual router under test, use the pass-through-end network topology model to configure the network scale and the interconnection relationship between each layer; (2) Generate a corresponding large-scale network topology model; (3) according to the port number and the node type of the above-mentioned tested router, locate the layer where the above-mentioned tested router is located; (4) judge whether there is above-mentioned tested router satisfying condition in above-mentioned network topology model, if not, then return to step (2); (5) Otherwise, the above-mentioned network topology model is mapped to an Internet structure composed of multiple autonomous systems AS and each autonomous system AS internally uses the Open Shortest Path First OSPF protocol; (6) According to the requirements of the OSPF protocol, the Type 5 link state advertisement LSA generation subsystem converts the link state information outside the autonomous system into a Type 5 link state advertisement, and sends it to the node directly connected to the router under test The corresponding protocol test subsystem; (7) Select a port of a node connected to the pass-through domain node at the location of the above-mentioned router under test as the current port; (8) In the AS where the router under test is located, the Type 4 LSA generation subsystem generates Type 4 link state advertisement LSAs for each AS border router, and sends these LSAs to the protocol test sub-system corresponding to the current interface system; (9) Determine the corresponding other routers in the AS for the current interface, and the LSA generation subsystem of the first to third types of link state advertisements forms the link state advertisements of the first to third types according to the protocol requirements, and sends them to the corresponding protocol test subsystem; (10) each link state advertisement subsystem sends an LSA description packet to the above-mentioned router under test through its respective protocol testing subsystem; (11) After the above-mentioned router under test receives the corresponding description packet, it sends the LSA request packet, and after the protocol testing subsystem receives the request packet, it sends the corresponding LSA update packet, and starts timing each link state update process simultaneously; (12) The router under test will send the LSA confirmation packet after receiving the LSA update packet according to the OSPF protocol, and the protocol test subsystem will end the timing after receiving the corresponding LSA confirmation packet; (13) process all ports of the above-mentioned tested router by steps (7)～(12); (14) Statistical average delay time and output; (15) END. 2. The routing protocol testing method based on the pass-through-end network model according to claim 1, wherein the mapping relationship between the pass-through-end domain topology model TS and the Internet structure described in step (5) is as follows: TS Topology Model - BGP, OSPF-oriented Internet; The entire topology model - the Internet; A pass-through domain and all stub domains it connects to - an autonomous system; Links connecting different pass-through domains——BGP interactive links; Passthrough domain—OSPF backbone area; stub domain - OSPF non-backbone area; Nodes connecting different pass-through domains - AS border routers; Nodes in the pass-through domain - OSPF domain border routers; Nodes in the stub domain——OSPF intra-area routers; Links connecting different stub domains——OSPF virtual links. 3. The system based on the routing protocol testing method of the punch-through-end network model is characterized in that: it is a kind of Internet routing testing system IRTS, and it contains: topology generation subsystem, tested router selection subsystem, link state advertisement generation Subsystems LSA1-LSA5, protocol testing subsystem, operation and management OAM subsystem, wherein the selected subsystem of the router under test controls the above-mentioned topology generation subsystem and is connected to its data output terminal, and the link state advertisement generation subsystem is connected with the above-mentioned router under test Select the data output terminal of the subsystem to connect and generate 5 types of link state advertisements. The protocol test subsystem is connected to the data output terminal of the above-mentioned LSA generation subsystem and interacts with the router under test for routing protocols, and operates and manages the OAM subsystem. Control the above-mentioned topology generating subsystem, tested router selecting subsystem, link state advertisement generating subsystem and protocol testing subsystem.

Images