WO2010058144A1

WO2010058144A1 - Method and device for discovering the level-3 topology of a corporate ip internet network

Info

Publication number: WO2010058144A1
Application number: PCT/FR2009/052286
Authority: WO
Inventors: Frédéric PAILLER; Romain Delordre
Original assignee: Centre National D'etudes Spatiales
Priority date: 2008-11-24
Filing date: 2009-11-24
Publication date: 2010-05-27
Also published as: FR2938992A1; FR2938992B1

Abstract

The invention relates to a method for discovering the level-3 IP topology of an internet network (2) including a step (110) of determining an initial subnetwork (6) to which a discovery terminal (3) is connected. The method comprises steps comprising, in a recursive loop of common path parameters (Nbsauts), searching the subnetworks (34, 36, 38, 40, 42, 44) with the same subnetwork rank (Nbsauts) from all the subnetworks (18, 20, 22) with a rank immediately lower by a number one than the common value (Nbsauts), the rank of a subnetwork being the minimum number of routers required to access said subnetwork from a discovery terminal, and assigning the common value of the common path parameter (Nbsaut) as subnetwork rank to the predetermined subnetworks (34, 36, 38, 40, 42, 44) with the same subnetwork rank (Nbsauts).

Description

Method and device for discovering the level 3 topology of a corporate IP internet network

The invention relates to a method for discovering the level 3 topology according to the OSI standard of an IP type enterprise network and more generally a corresponding mapping method.

The architecture of an IP-based enterprise computer network, that is to say using the Internet Protocol (IP), is evolving at such a speed that the associated documentation is not updated quickly enough to provide a real picture of the network topology. To this end, numerous computer tools have been developed for automatically discovering and graphically displaying the topology of the network.

The tools developed are for the most part network management tools that do not offer an adequate view of the network either because they do not provide enough information, for example by confining themselves to network equipment, or conversely because the intervention of the operator is necessary to inform additional information.

For this reason, tools have been proposed that implement Level 3 information discovery processes, level 3 being defined in the OSI (Open Systems Interconnection) classification. Level 3 information is primarily about routers and subnets, where a subnet is defined by interconnected IP network machines that can communicate with each other without having to go through a level 3 router.

Such search methods are not limited to the underlying subnets of a same router but can also go beyond the router.

The document EP 1 211 843 A1 describes a method for discovering the topology of a level 3 IP network according to the OSI standard which is not limited to the underlying subnetworks of the same router but which makes it possible to go beyond the router. The method described in this document firstly comprises a first step of discovering the different subnetworks of the entire network and for each of them of all their immediate connectivity links with each other. the underlying sub-networks and level 3 routers closest thereafter and then a second step of extracting the topology of the global network.

The discovery method described is certainly an exhaustive discovery method having a wide scope of the connectivity links of the different sub-networks with their neighbors but it has the disadvantage of being inefficient and slow to implement in the case where several subnets are looping back and forth across routers.

Indeed, the search is not oriented at first glance on the discovery of the topology in itself, and several times redundantly authorizes the discovery of the same sub-network at a different stage of the process.

The technical problem is to improve the efficiency of the discovery of the IP level 3 topology of the global network in the case where several subnetworks are looped back together through routers.

For this purpose, the subject of the invention is a method for discovering the IP level 3 topology of an internet type network comprising: a set of level 3 IP routers and a set of IP subnetworks forming interconnected nodes according to a predetermined graph topology, and a discovery terminal connected to one of the subnets of the set of networks called the initial subnetwork, the method comprising the steps of starting from the discovery terminal to: determining the sub-network initial network to which the discovery terminal is attached, assigning to the initial subnetwork the value zero as a subnet rank, characterized in that it comprises the steps of: in a recursive loop of current path parameter (Nbsauts), search subnets with the same subnet rank (Nbsauts) from the set of subnets with the rank of a number one below the current value (Nbsauts), the rank of any subnet relative to the discovery terminal being the minimum number of routers required to access said subnet from the discovery terminal, and assigning to the determined subnets with the same subnet rank (Nbsauts) the current value of the current path parameter (Nbsaut) as the subnet rank.

According to particular embodiments, the discovery method comprises one or more of the following characteristics:

the step of searching for subnets of the same rank Nbsaut comprises for each subnet of rank Nbsaut-1 scanned, a step of determining the routers connected as physical units to said subnetwork of rank Nbsaut -1 independently virtual routing implemented in any of the routers;

the method comprises, for each determined router, the steps of: determining the connected active physical interfaces, for each interface, determining the associated subnetwork and verifying that it is not a sub-network of rank lower than or equal to subnet current minus one, and in the case where the associated subnet is of a higher rank, register it in a memory of the subnets to be scanned at the next subnetwork rank index cycle Nbsauts + 1;

the method comprises the step of: registering the determined associated subnetwork in a memory of sub-networks already scanned;

during the step of registering a determined associated subnetwork in the memory of sub-networks already scanned, the IP address and the mask of the sub-network are recorded in the same record; the step of determining the routers comprises the steps of: recovering the address range of the scanned subnet, and for each IP address contained in the retrieved range, testing the activity of the sending address; a test message, and register the active address when an active machine is detected at this address, and for each registered active IP address, send a UDP request to the destination port associated with the SNMP service, and in the case where the response to this request indicates that the port is not closed, send a request of type sysName, and after reception of the answer, in case the variable sysName is filled and the machine filled is not already known in the list of records of routers, retrieve the list of IP interfaces of equipment with their mask associated with them, and in the case where two sub-networks are detected, the machine is determined as a level 3 router, register the determined router; the step of retrieving the list of physical interfaces comprises the steps of: querying the ipAddrTable table and more specifically the ipAdEntNetMask subset whose OID is .1.3.6.1.2.1.4.20.1.3 with an SNMP request of getBulkRequest type available from SNMP version 2c, retrieve from a single package all the data of the ipAdEntNetMask subset; and

- During the step of registering a specific router, the IP address of the determined router and the addresses of the subnets connected are recorded in the same record in the form of a list. The invention also relates to a mapping method comprising a topology discovery method as defined above and a network display step from the subnets and routers recordings, in the form of layers. concentric or parallel, each layer comprising the registered subnetworks of the same rank Nbsauts being distant from the initial sub-network of rank 0 proportionally to the rank of sub-network, and delimited by border zones formed by the routers of the same rank of router.

The invention also relates to a system for implementing the method of discovering the IP level 3 topology of an Internet type network comprising: a set of level 3 IP routers and a set of IP sub-networks forming nodes interconnected according to a predetermined graph topology, and a discovery terminal connected to one of the subnets of all the networks called the initial subnet, the discovery terminal being able to determine the initial subnet to which the discovery terminal is attached and to assign the subnet to the subnet; initial network zero as a subnet rank, characterized in that the discovery terminal comprises; means for searching according to a recursive loop of current path parameter (Nbsauts) the subnetworks having the same rank of sub-network

(Nbsauts) from the set of immediately lower rank subnets of a number one to the current value (Nbsauts), the rank of any subnet relative to the discovery terminal being the minimum number of routers necessary to access said subnet from the discovery terminal, and to assign to the determined subnets having the same subnet rank (Nbsauts) the current value of the current path parameter (Nbsaut) as a rank of subnet.

According to particular embodiments, the system comprises one or more of the following characteristics:

the means for searching the subnetworks of the same rank Nbsaut are able to determine routers connected as physical units to said subnetwork of rank Nbsauts -1 independently of a virtual routing implemented in any one of the routers; and

the means for searching the Nbsaut sub-networks of same rank are suitable for each determined router to determine the connected active physical interfaces, and for each interface to determine the associated subnet and to check that it is not a sub-network. a network of rank lower than or equal to the current rank of sub-network minus one, and in the case where the associated subnet is of a higher rank than registering it in a memory of the subnetworks to be scanned at the next cycle of sub-network rank index Nbsauts + 1.

The invention also relates to a recording medium of an instruction recording program that can be executed by a computer implementing a method as defined above. The invention will be better understood on reading the description of a single embodiment which will follow, given solely by way of example and with reference to the drawings in which:

FIG. 1 is a view of a general level 3 architecture of an Internet type network,

FIG. 2 is a flow diagram of a method for discovering the IP level topology of a network of the type described in FIG.

FIG. 3 is a detailed flow chart of a particular step of the flowchart described in FIG. 2. According to FIG. 1, an enterprise network of the Internet type, that is to say implementing the IP protocols. (Internet Protocol), also called Intranet is represented at level 3, level 3 being defined according to the OSI (Open System Interconnections) classification.

The network 2 comprises a set of logical subnetworks interconnected through routers according to a predetermined but unknown graph topology.

Thus, OSI model-level 2 components such as access switches are not considered in the level 3 discovery process. In addition, if virtual contexts are implemented on the same router considered as physical entity, they are represented here as independent routers.

The network 2 comprises a discovery terminal 3 with a computer 4 having means for determining subnetworks 5. The discovery terminal 3 is connected to any subnet 6 of the network called the initial subnet.

In addition to the discovery terminal 3 a first and second routers 8 and 10 are connected to the initial subnet 6.

The set 12 formed by the initial subnetwork 6 is defined as the set of subnets with a rank of zero subnetwork.

The first router 8 and the second router 10 are called routers having a router rank equal to 1. The set 12 of subnetworks is formed by the subnetworks which are connected to the first and second routers 8, 10, of router rank equal to 1, except the initial subnet 6 of subnet rank equal to zero.

Here, the set 12 called set of sub-array subnets equal to 1 is formed of the sub-networks 16, 18, 20 connected to the first router 8 and sub-networks 22, 24 connected to the second router 10 .

Tier 2 routers are routers connected to subnets of subnet rank equal to 1 that are not router rank routers equal to 1. Here, routers of router rank equal to 2 are routers 24, 26, 28,

Connected respectively to the subarrays 18, 20, 20 and 22, 22.

It should be noted that in FIG. 1, the subnetworks interconnected to at least two routers are each represented by a pattern having a gray interior while subnetworks connected to a single router are each represented by a pattern having a white interior. .

The set 32 of subnets is formed by the subnets connected to router rank routers equal to 2 except subnets of subnet rank equal to 1.

Here, the set 32 called set of sub-array subnets equal to 2 is formed by sub-networks 34, 36, 38, 40, 42, 44 respectively connected to routers of router rank equal to 2 , 24, 24, 26, 28, 28, 30.

Router rank routers equal to 3 are routers connected to subnet-rank subnets equal to 2 that are not router-rank routers equal to 2. Here, router-rank routers equal to 3 are routers 46, 48, 50 respectively connected to the sub-networks 34, 38, 42.

The set of subnets 52 is formed by the subnets that are connected to the router rank routers equal to 3 except subnets of subnet rank equal to 2. Here, the set 52 called together subnets of sub-network rank equal to 3 is formed by sub-networks 54, 56, 58, 60 connected to routers of router rank equal to 3 respectively routers 46, 48, 48, 50. Thus, an ordered description of the topology at level 3 of the network 2 is performed by the computer 4 of the discovery terminal 3 in which the routers and the subnets are ordered according to their rank of router and their rank respectively. subnet that were assigned to them by the computer 4.

In general, the subnet rank assigned by the computer 4 and represented by the variable called Nbsauts, for a given subnet relative to the discovery terminal 3 is the minimum number of routers required to access said subnetwork from the discovery terminal 3. In general, the router rank of a router assigned by the computer 4 to a router is equal to the minimum rank of the subnet or subnets connected thereto plus the number 1.

It should be noted that in the topology of the network 2 given by way of example in FIG. 1, the sub-network 22 is connected to both routers 28 and 30 of rank 2 and to router 10 of rank 1.

It should also be noted that a subnetwork loop exists. It is defined by the chain of connections connecting the initial subnet 6 to the router

10, the router 10 to the subnet 22, the subnet 22 to the router 28, the router

28 to subnet 20, subnet 20 to router 8, and router 8 to the initial subnet 6.

According to FIG. 2, a flowchart 100 of the discovery method comprises a set of steps traversed according to a recursive type algorithm during which the network is probed by increasing rank of subnet, that is by incrementing the parameter Nbs. . The discovery of the topology of the IP network is implemented from a given point in the network, more precisely here the discovery terminal 3 by the exchange of request and response messages belonging to the Internet protocols.

First, when Nbsaut is equal to zero, the subnet rank subnet equal to zero on which the search terminal 3 is located is determined as well as the router rank routers equal to 1.

The discovery terminal 3 implementing the algorithm in search of routers of router rank equal to 1 performs this search. Once the routers are detected, the computer 4 tries to find to which other subnets these are connected, the computer 4 stores them and finally we scan the current subnet to gather the information on the machines. connected to it. Then, the search for routers on the stored subnets can resume and so on successively.

The discovery process thus advances iteratively jump jump incrementing variable Nbsauts until reaching the jump limit set at the beginning of the process or until reaching the limits of the network or routers not delivering information on their external interconnections. More specifically, in a first initialization step 102, a first work stack 104 of the current subarrays to be scanned corresponding to the rank of the current value of the Nbsaut variable, as well as a stack 106 for preparing the subnets to be used. scan at the next hop are emptied that is to say put in a reference state corresponding to the vacuum. In addition, a memory 108 for storing already scanned subnets and routers is also initialized by setting to a reference state corresponding to the vacuum. The current variable of rank of subnetworks, called Nbsauts is set to zero and a maximum number of rank of subnets to be discovered is set also in step 102 by an entry of the operator. In a next step 110, the initial subnet 6 is determined.

Then in a step 112, the initial subnet 6 determined is stacked in the work stack 104.

In a step 114 of branching a first loop, a recursive discovery loop, parameterized by the value of the number of jumps, Nbsauts, is initiated by the initial value of Nbsaut equal to zero.

In a step 116 subsequent to the branching step 112, the current value of the Nbsauts parameter is compared with the value of the predetermined maximum hopping number set during the initialization step 102.

When the maximum hop count is exceeded, then the process is terminated in the stop step 120.

Conversely, when the number of jumps current Nbsauts is less than or equal to the maximum number of jumps, then a set of steps of scanning of current subnets corresponding to a subnet rank equal to Nbsauts is implemented.

In a step 122, the work stack 104 is consulted.

In the case where the battery 104 is not empty, in a step 124, the battery 104 is unstacked from a sub-network unit to start the path of a second loop.

In a step 126, the routers connected to the subnetwork depilated at 124 and being analyzed are determined.

In a step 128 for connecting a third routing loop of the routers determined in step 126, each current router associated with a path pointer of the third loop is examined.

In a next step 130, the interfaces of the scanned current router being examined and corresponding to the current pointer of the third loop are determined. In a subsequent step 132 of connecting a fourth loop of the interfaces determined in step 128, each scanned interface associated with a path pointer of the fourth loop is examined.

In a next step 134, for the scanned interface being examined, the associated subnet is determined if one exists. If it does not exist, the pointer of the fourth loop is incremented and the method jumps to the connection step 132.

If there is an associated subnet, in a next step 136, it is examined whether the network has already been scanned and examined by going to read the storage memory 108 of sub-networks already scanned. If the subnet has already been scanned and examined by being stored in the memory 108, then the pointer of the fourth loop is incremented and the next interface is examined by skipping to step 132.

If the subnet has not already been scanned, in a step 138 subsequent to 136, it is checked whether the subnet is in the work stack 104. If the subnet is registered in the work stack 104, then the interface pointer of the fourth loop is incremented and the next interface is examined by skipping to step 132. If the subnet is not registered in the work stack 104, then in a step 140 the associated subnet is stacked in the subnet stack 106 to be scanned at the next hop.

In a step 142, it is examined whether all the interfaces of the current determined router have been scanned and if this is not the case, the method jumps to step 132.

When it is determined in step 142 that all interfaces have been examined, it is determined in step 144 whether all the routers determined in step 126 have been examined. If there is still at least one router to examine, then the process jumps to step 128.

If all the routers have been examined, the current subnet depicted in step 124 is analyzed in step 146 and the information associated with this current subnet is added to the storage file 108 of the already scanned subnets. . Among the information associated with this subnet is the subnet rank assigned to it the current value of Nbsauts by the computer 4.

Then, the process jumps to step 122, and in the case where the working stack 104 is not empty, all the steps 124, 126, 128, 130, 132, 136, 138, 140, 142 are repeated. , 144, 146.

In the case where the stack 104 is empty, in a step 148, the contents of the subnetting preparation stack 106 are transferred to the work stack 104, the preparation stack 106 then being in an empty reference state.

In a next step 150, it is checked whether the contents of the work stack 104 are empty.

If the working stack 104 is empty, then the discovery process is completed in the stopping step 120.

Otherwise, the path parameter of the first loop, Nbsauts, is incremented by one in step 152 and the first discovery loop is again implemented by progressing to step 116 and then to step 122 if the maximum reach of the discovery set by the maximum allowable number of jumps allows. Thus, each sub-network is analyzed once and the storage memory 108 of sub-networks already scanned contains in an orderly manner the topology of the network in an order governed by the parameter of the number of hops.

A graphical display of the topology of the level 3 network is performed directly from the structured data contained in the storage memory.

As a variant, the graphic display is made as and when the entities of the network are discovered, the distance of the entities discovered with respect to the search terminal being a function of the parameter Nbsauts. In Fig. 3, step 126 is described in detail in a set of steps.

In a first step 200, the range of IP addresses of the subnet being examined is recovered. From the IP address of the subnet and its mask provided in step 124, the IP addresses of the minimum and maximum terminal of the machines that this subnet contains are determined.

For example, if the IP address of the subnet is 192.168.30.0 and the mask is 255.255.255.0, then the range is defined by the minimum bound of 192.168.30.0 and the maximum bound of 192.168.30.255. As a result, the available addresses in this range are 192.168.30.1, 192.168.30.2, ..., 192.168. 30,254.

In a branching step 202, a fifth loop is traveled for each IP address in this range except the one that has already been registered from the already discovered router. Thus, a router already listed is not scanned again and examined. In a next step 204, for each IP address 204, the equipment associated with this address is tested to see if it is active or not.

For this, the computer 4 sends an ICMP Echo Request command (ICMP designating in English Internet Control Message Protocol) to the polled machine identified by its IP address which is supposed to respond by an ICMP Echo Reply. In case the polled machine does not respond, another method is used which consists in using the TCP protocol in a devious way in relation to its originally intended purpose, the presence of a machine being detected even if it does not respond to NCMP Echo Request. For this, the principle is to send a TCP ACK type packet on the port 80 for example, the packet usually used to transfer data and TCP designating in English Transmission Control Protocol (protocol part of the heart of Internet protocols). Since the remote polled machine has not previously established a connection and does not understand why it receives this message, it responds with an RST message (abbreviation of Reset) that indicates its presence.

This test is also valid for a TCP SYN packet that requests the remote machine to establish connections. If the port of the remote machine on which the packet is sent, is closed then this one will answer by a packet of type RST, in the opposite case it will begin the second stage of the establishment of the connection by the sending of a SYN / ACK package, two scenarios that will reveal the presence of the targeted machine.

In a next step 206, the computer 4 tests whether the machine has responded thus revealing its presence on the sub-network. If it has not done so, the method loops back to step 202. If it has issued a response, the method then proceeds to a next step 208.

In step 208, the active IP address is stored.

Then in a step 210, one tests if all the IP addresses of the range of the loop were traversed.

If in step 210 all the IP addresses of the range traveled have been tested, it leaves the fifth loop to continue at a step 212.

In the branching step 212, a sixth loop is traversed by a pointer describing all the IP addresses found to be active. In a step 214 for each current active IP address, the computer 4 sends a User Datagram Protocol (UDP) packet to the port 161 associated with the Simple Network Management Protocol (SNMP) service. The package does not contain data but just the appropriate parameters to check if the recipient has on this port a service is open, closed or filtered.

In a next step 216, the state of the port is tested. If it is closed, there is no access to the SNMP protocol and then we go to step 150. Otherwise we continue to a step 218 consecutive. In step 218 a request of type get-request is sent on the variable

SysName of MIB-II (RFC 1213 of the machine), to raise the name of the machine.

In a step 220, if the variable SysName is filled in, a step 222 is implemented. Otherwise, we go back to step 212 of connection of the sixth loop, the pointer being on the next active IP address.

In step 222, if this name is already known among the routers already discovered, it returns to step 212 of branching of the sixth loop, the pointer being on the next active IP address. Otherwise we continue to a next step 224.

In step 224, the ipAddrTable table and more precisely the ipAdEntNetMask subset (whose OID (Object Identifier abbreviation) is .1.3.6.1.2.1.4.20.1.3) with a SNMP request of type GetBbulkRequest available at from SNMP version 2C, which allows to retrieve from a single packet all the data of the subset. IpAdEntNetMask informs for each IP interface the mask associated with it. The IP address of the interface is found thanks to the OID of the field:

Table I

In a step 226, it is tested whether the machine is connected to two sub-networks or more precisely if it has two active IP addresses that is to say not belonging to the same subnet in which case we can consider that it runs at least at level 3 of the OSI model. An SNMP manageable machine that has more than one different subnet IP interface is then considered a router. Finally, with the IP address and mask found in the ipAdEntNetMask table, a subnet is determined.

The advantage of identifying a subnet by the ipAdEntNetMask table is that this method finds cases where a single router has two or more IP addresses in the same subnet. This configuration appears in the event that one of the HSRP protocols (Hot

Standby Router Protocol) or VRRP (Virtual Router Redundancy Protocol, described in RFC 3768) are implemented and show in the table two gateway IP addresses including a virtual one. The ipAdEntNetMask table is used to list subnets to which the router is connected.

Queries on MIB objects of type Table are requests of the type GetRequest available from version 2C of SNMP.

They make it possible to repatriate all the values of the table in a packet and not by incrementation as in version 1 where it is necessary to send a request GetNext to traverse each of the values of the table.

In the case where a single value is requested, then we use a request of the type get-request, the requests getBulkRequest and get-request generating a response of the type get-response. In a step 228 if the test performed in step 226 results in a router detection, the router is stored in the storage memory 106 of the subnets to be scanned at the next hop. The router is registered with its router rank to which the current value Nbsauts has previously been assigned by the computer 4. The sixth loop is continued until at step 230 where it is tested if IP addresses still active n ' have not been examined yet. If this is the case, the process is continued at step 212 of branching. Otherwise, the step

126 is complete and the discovery process then continues to step 128.

The discovery process described in Figure 2 traverses the routers and subnets by structuring them according to the minimum number of jumps to go from a subnet to the discovery terminal, which avoids to browse several times the same sub-network. in the case where there exists a sub-network loop such as that described by the sub-networks 6, 20, 22.

In addition, identifying a subnet by the ipAdEntNetMask table makes it possible to find all the interfaces of a router, regardless of any Virtual Virtualization and Forwarding (VRF) virtual partitioning implemented on this router.

Claims

A method for discovering the IP level 3 topology of an internet type network (2) comprising: a set of IP routers (8, 10, 24, 26, 28, 30, 46, 48, 50) of level 3 and a set of IP subnetworks (6, 18, 20, 22, 24, 34, 36, 38, 40, 42, 44) forming interconnected nodes according to a predetermined graph topology, and a discovery terminal ( 3) connected to one of the subnets of the set of networks called the initial subnetwork (6), the method comprising the steps of starting from the discovery terminal (3) to: determining (110) the sub-network initial network (6) to which the discovery terminal (3) is attached, assigning to the initial subnet (6) the value zero as a rank of sub-network, characterized in that it comprises the steps of: in a recursive loop of current path parameter (Nbsauts), search the subnets (34, 36, 38, 40, 42, 44) with the same rank of subnet (Nbsauts) from of all subnets (18, 20, 22) of immediately lower rank of a number one than the current value (Nbsauts), the rank of any subnet relative to the discovery terminal (3) being the minimum number of routers required to access said subnet from the discovery terminal (3), and assigning to the determined subnets (34, 36, 38, 40, 42, 44) having the same rank of sub-networks. network (Nbsauts) the current value of the current path parameter (Nbsaut) as a subnet rank.

2. The discovery method as claimed in claim 1, wherein the step of searching for sub-networks of the same rank Nbsaut comprises for each sub-network of rank Nbsaut-1 scanned, a step of determining (126) connected routers. as physical units to said rank Nnauts -1 subnet regardless of virtual routing implemented in any of the routers.

3. The discovery method as claimed in claim 2, characterized in that the method comprises for each determined router the steps of: determining (130) the connected active physical interfaces, for each interface, determining (134) the associated subnetwork and verify that it is not a subnet of rank lower than or equal to the current rank of subnet minus one, and in the case where the associated subnet is of a higher rank, save it in a memory ( 106) subnets to be scanned at the next sub-network rank index cycle Nbsauts + 1.

4. Discovery method according to claim 3, characterized in that it comprises the step of:

Registering (146) the determined associated subnet in a memory (108) of already scanned subnets.

5. Discovery method according to claim 4, characterized in that during the step of recording (146) a determined subnetwork determined in the memory (108) sub-networks already scanned, the address IP and the subnet mask are saved in a single record.

The discovery method according to claim 2, characterized in that the step of determining (126) the routers comprises the steps of: recovering (200) the address range of the scanned subnet, and for each IP address contained in the retrieved range, testing the activity of the address (206) sending (204) a test message, and storing (208) the active address when an active machine is detected at that address, and for each registered active IP address, send (214) a UDP request to the destination port (161) associated with the SNMP service, and in case the response to this request indicates that the port is not closed, send ( 218) a request of the type sysName, and after reception of the answer, in the case where the variable sysName is filled and the machine entered is not already known in the list of the records of the routers, retrieve (224) the list of the interfaces IP equipment with their mask that their are associated, and in the case where two subnetworks are detected, the machine is determined (226) as a level 3 router, register (228) the determined router.

7. Method of topology discovery according to claim 6, characterized in that the step (224) for recovering the list of physical interfaces comprises the steps of: interrogating the ipAddrTable table and more specifically the subset ipAdEntNetMask whose 'OID is .1.3.6.1.2.1.4.20.1.3 with a SNMP request of type getBulkRequest available from SNMP version 2c, retrieve from a single package and all data from the subset ipAdEntNetMask.

Method for discovering a topology according to claim 6, characterized in that during step (228) of the recording of a determined router, the IP address of the determined router and the addresses of the subnetworks connected are recorded in the same record as a list.

A mapping method comprising a topology discovery method defined according to one of claims 1 to 8 and a network display step from the subnets and routers recordings, in the form of concentric or parallel layers, each layer grouping the registered sub-networks of the same rank Nbsauts being distant from the initial sub-network of rank 0 proportionally to the rank of sub-network, and delimited by border zones formed by the routers of the same rank of router.

A system for discovering the IP level 3 topology of an Internet type network comprising: a set of level 3 IP routers and a set of IP subnets forming interconnected nodes according to a predetermined graph topology, and a discovery terminal (3) connected to one of the subnets of all the networks called initial subnet (6), the discovery terminal (3) being able to determine the initial subnet (6) to which is attached the discovery terminal (3) and assigning to the initial subnet (6) the value zero as a subnet rank, characterized in that the discovery terminal (3) comprises; means (5) for searching according to a current path parameter recursive loop (Nbsauts) the subnetworks (34, 36, 38, 40, 42, 44) having the same subnet rank (Nbsauts) from the set of subnets (18, 20, 22) of rank immediately below a number one to the current value (Nbsauts), the rank of any subnetwork relative to the discovery terminal (3) being the minimum number of routers required to access said subnet from the discovery terminal (3), and to assign to the determined subnets (34, 36, 38, 40, 42, 44) having the same rank of sub-networks. network (Nbsauts) the current value of the current path parameter (Nbsaut) as a subnet rank.

11. Discovery system according to claim 10, characterized in that the means (5) for searching the Nbsaut sub-networks of the same rank are capable of determining routers connected as physical units to said rank N-subnet. 1 independently of a virtual routing implemented in any of the routers.

12. The discovery system according to claim 11, characterized in that the means for searching the Nbsaut sub-networks of the same rank are suitable for each determined router to determine the connected active physical interfaces, and for each interface to determine the subnetwork. associated and verify that it is not a subnet of rank less than or equal to the current rank of subnet minus one, and in the case where the associated subnet is of a rank higher than the record in a memory (106) of the subnets to be scanned at the next sub-network rank index cycle Nbsauts + 1.

13. Instruction recording medium capable of being executed by a computer (4) implementing a defined method according to one of claims 1 to 9.