WO2008016861A2 - Inférence de liaison dans des réseaux importants sur la base de données incomplètes - Google Patents

Inférence de liaison dans des réseaux importants sur la base de données incomplètes Download PDF

Info

Publication number
WO2008016861A2
WO2008016861A2 PCT/US2007/074679 US2007074679W WO2008016861A2 WO 2008016861 A2 WO2008016861 A2 WO 2008016861A2 US 2007074679 W US2007074679 W US 2007074679W WO 2008016861 A2 WO2008016861 A2 WO 2008016861A2
Authority
WO
WIPO (PCT)
Prior art keywords
network
node
port
nodes
root
Prior art date
Application number
PCT/US2007/074679
Other languages
English (en)
Other versions
WO2008016861A3 (fr
Inventor
Rajesh Balasubramaniam
James Mark Shaw
Original Assignee
Rajesh Balasubramaniam
James Mark Shaw
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Rajesh Balasubramaniam, James Mark Shaw filed Critical Rajesh Balasubramaniam
Publication of WO2008016861A2 publication Critical patent/WO2008016861A2/fr
Publication of WO2008016861A3 publication Critical patent/WO2008016861A3/fr

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/14—Routing performance; Theoretical aspects
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00—Data switching networks
    • H04L12/28—Data switching networks characterised by path configuration, e.g. LAN [Local Area Networks] or WAN [Wide Area Networks]
    • H04L12/46—Interconnection of networks
    • H04L12/4641—Virtual LANs, VLANs, e.g. virtual private networks [VPN]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/02—Topology update or discovery
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/48—Routing tree calculation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L45/00—Routing or path finding of packets in data switching networks
    • H04L45/48—Routing tree calculation
    • H04L45/488—Routing tree calculation using root node determination

Definitions

  • This invention relates to the field of network analysis systems, and in particular to a method and system for determining a topology of one or more networks based on information contained in Address Forwarding Tables (AFT) associated with switches of the network.
  • AFT Address Forwarding Tables
  • a preferred user interface for visualization and analysis tools generally provides a display of the topology of the network, wherein colors and other graphic features are used to illustrate features and performance characteristics related to the network nodes and their interconnecting links.
  • AFT address forwarding tables
  • the term 'simply connected' is used to indicate that two nodes are connected, either directly to each other, or via one or more intermediate nodes.
  • the ports that must be providing this connectivity can be determined. Lowekamp determined that, in an acyclic network, if the through sets of two ports contain an address in common, then those two ports cannot be simply connected; if a single pair of ports on a pair of communicating switches are found with through sets that do not contain an address in common, they must be simply connected.
  • a 'virtual link' can be added to the topology. If multiple nodes are identified as being simply connected to the same port, a virtual shared segment can be inferred, such as the use of a public (and not-modeled) network from this port to these multiple nodes.
  • Lowekamp's simply-connected determination provides a technique that is more robust for determining network topology than Breitbart's directly-connected determination when complete connectivity data is not available
  • both Lowekamp's and Breitbart's techniques are computationally demanding, particularly in a complex network comprising hundreds or thousands of switchs, each with multiple ports.
  • neither Lowekamp nor Breitbart addresses the merging of their techniques with other known link inference techniques and other connectivity tools; and neither addresses the use of additional information, such as the presence of particularly configured virtual local area networks (VLANs), such as protocol-filtered VLANs, or the presence of aggregate links that logically group individual physical links.
  • VLANs virtual local area networks
  • Lowekamp does not address the absence of connectivity information relative to entire segments of the network. Lowekamp assumes that all nodes are somehow connected, and, correspondingly, sufficient connectivity information is available for linking all nodes via a tree search starting from a root switch.
  • partitions a network into a set of independent partitions determines the topology of each partition, then merges the topologies to form a topology of the entire network.
  • the partitioning is hierarchical, wherein the network is partitioned to form individual VLAN partitions, and each of the VLAN partitions is further partitioned based on the nodes that are simply connected to each port of one or more selected root switches within the VLAN partition. Simple connections to each port are efficiently determined based on an aggregate address forwarding table associated with each node.
  • Ancillary information such as spanning tree or CDP data, may be used to facilitate efficient partitioning and/or to validate inferences that are made with incomplete information.
  • FIGs. IA- 1C illustrate an example partitioning of a network in accordance with this invention.
  • FIG. 2 illustrates an example flow diagram for inferring network topology in accordance with this invention.
  • FIG. 3 illustrates an example flow diagram for prioritized merging of links in accordance with this invention.
  • FIG. 4 illustrates an example block diagram of a link inference system in accordance with this invention.
  • the term 'node' is used to reference any network element, including routers, switches, bridges, terminals, and so on.
  • the term 'switch' is used to facilitate understanding of this invention by distinguishing nodes that can be configured to receive messages on one port and selectively forward that message to another of its ports.
  • a switch is generally referred to as a port when its function as a switch is immaterial to the topic being discussed.
  • FIG. IA illustrates a sample network comprising switches A-T, some of which are members of a virtual network Vl.
  • switches A-T some of which are members of a virtual network Vl.
  • AFT Address Forwarding Table
  • TS Through-Sets
  • the network is partitioned into smaller networks to reduce the combinatorial complexity.
  • each VLAN is processed as a separate entity.
  • Switches that have ports that are not assigned to a VLAN are partitioned together, as if they were members of a common VLAN, herein termed an 'implied virtual network'.
  • Multiple implied virtual networks can be defined, if there are characteristics available to distinguish among different subnetworks of nodes. If, for example, switches that have untagged ports that are assigned to a VLAN, such that a globally significant VLAN identifier is not available, are distinguishable from switches that are not assigned to a VLAN.
  • ports that support a particular protocol can be distinguished from those that support a different protocol, and different implied virtual network partitions can be defined based on the different protocols supported by the ports.
  • a single switch can appear in multiple VLAN partitions, depending on the configuration of its ports.
  • switch I includes ports that are members of VLAN Vl, and other ports that are members of the implied virtual network partition of non-VLAN ports (hereinafter VO).
  • VO virtual network partition of non-VLAN ports
  • VO ⁇ A, B, C, D, E, 1(3,4), R, S ⁇ ;
  • Vl ⁇ F, G, H, 1(1,2), J, K, L, M, N, O, P, Q, T ⁇ .
  • the 1(1,2) and 1(3,4) references are used to distinguish the ports of switch I in each partition; for each of the other switches, all of the ports of the switch are within the partition. [0017]
  • independent networks are effectively formed. Because each independent network will generally include fewer nodes than the original network, or at least will contain fewer ports than the original, determining the topology of each of these networks will involve fewer combinations that need to be tested, and thus the cumulative time for determining all of the topologies will likely be substantially less than the time to determine the overall network topology directly.
  • each of these independent networks is further partitioned by identifying the set of nodes that are simply connected to each port of one or more selected 'root-switches' within the partition.
  • switch A is selected as a first root-switch of network VO. If the address forwarding table of switch A is known to be complete, each simply connected node to each port of switch A will appear in this address forwarding table. However, in actual networks, rarely are forwarding tables known to be complete.
  • simply connected nodes are determined based on an aggregate address forwarding table associated with each node.
  • the aggregate address forwarding table of a node is the union of the AFTs of each port on the node. It can be shown if any of the following conditions are true, a simple connection must exist between two nodes:
  • ⁇ AFT(i) is the aggregation of all the AFT entries in all the ports(i) of the first node
  • ⁇ AFT(J) is the aggregation of all the AFT entries in all the ports(j) of the second node
  • ⁇ MAC (i) is the aggregation of all the port MAC addresses of all the ports(i) of the first node
  • ⁇ MAC(J) is the aggregation of all the port MAC addresses of all the ports(j) of the second node.
  • a 'synonym' table is maintained for tracking duplicate MAC addresses that are assigned to the same port.
  • a common MAC address replaces each occurrence of such synonymous addresses so that these tests, and others can be effectively applied.
  • FIG. 1C illustrates an example set of address forwarding tables for switches A-J and Q-T.
  • root-switch A is simply connected to each of the other nodes in the network VO, even though the address forwarding table of root-switch A does not expressly reference nodes I or R of this network. From the address forwarding tables of root switch A, it is known that node C is on one port, node D is on another, and nodes B, E, and S are on a third port; but it is unknown how nodes I and R are connected to root-switch A.
  • the Through-Set test of Lowekamp et al. can be applied.
  • the exhaustive Through-Set test can be bypassed or foreshortened by assessing the address table on each of the simply connected nodes. For example, if each node includes an identification of the other node on one of its ports, these ports must be simply connected. For example, in FIG. 1C, port 2 of switch B addresses switch R, and port 1 of switch R addresses switch B; therefore, port 2 of switch B must be simply connected to port 1 of switch R, and there is no need to apply the Through-Set test. Similarly, if a port of only one of the switches addresses the other, the other ports on that node need not be considered.
  • port 4 is used to connect to root-switch A, and thus we only need to compare the through set of this port to each of the ports of root- switch A.
  • a MAC address entry in the address table of switch I can be resolved to a particular port of root- switch A, then it would be known that port 4 of switch I connects to that port of root-switch A, and the Through-Set test would not be required.
  • the entry in the address table does not resolve itself to a particular port on switch A, and thus the Through-Test of Lowekamp et al. is applied.
  • the Through-Set of a port is an aggregation of all of the address forwarding tables of the switch except the address forwarding table of this port. Lowekamp et al. have shown that if the intersection of the Through-Set of two ports on different switches is not null, these two ports cannot form a simple connection. If sufficient information is available, the simply connected ports can be found by a process of elimination.
  • the Through-Set of port 4 of switch I is ⁇ E ⁇ ; that is, we know that data addressed to switch E that is received on port 4 will be routed through switch I from port 4 to port 3.
  • the Through-Set of ports 1, 2, and 3 of switch A are ⁇ D, B, E, S ⁇ , ⁇ C, B, E, S ⁇ , and ⁇ C, D ⁇ , respectively. Of these, only the through-set of port 3 of switch A provides a null intersection with the Through-Set of port 4 of switch I. Because this is the only viable option, we conclude that switch I must be in the set of nodes that are simply connected to port 3 of switch A.
  • port 3 of switch A is used to connect to switch B.
  • D ⁇ By comparing only the Through-Sets of port 3 of switch A ⁇ C, D ⁇ to the Through-Sets of all the ports of switch B ⁇ R, E, I ⁇ , ⁇ D, E, I ⁇ , and ⁇ D, R ⁇ , it can be determined that port 1 of switch B is simply connected to port 3 of switch A.
  • Each of the ports on the nodes that provide the simple connection to the root-node is defined as the root-port for that node; this designation will subsequently be used to trace the sets of nodes associated with each port of the root-switch, as detailed further below.
  • switches I and R are simply connected to port 3 of switch A, these switches can be added to the address table of port 3 to facilitate further connectivity determinations.
  • switch A can be added to the address table of the root-port of each simply connected node B, E, R, S that does not have this entry.
  • virtual network Vl is partitioned by selecting a first root switch, F, and determining all of the nodes that are simply connected to switch F using rule set (1), above.
  • rule set (1) a first root switch, F
  • only nodes G and P will be determined to be simply connected to switch F; and, because switch F has only one port, these nodes form the set of nodes associated with this port of root switch F.
  • Switch G will be added to the address table of this port on switch F, and switch F will be added to the address table of switch P, if not already present.
  • a second root switch, H is selected and the above process is repeated to determine that nodes I, J, N, and Q are coupled to port 1 of H; nodes K and T are coupled to port 2; and nodes L, M, and O are coupled to port 3.
  • Each of the address tables of these ports will be augmented to include the newly discovered simple connections, and switch H will be included in the address table of each of the root ports of the simply connected nodes, if not already present. In this example, all of the remaining nodes were associated with a port of root node H, and thus the partitioning of network Vl is complete.
  • IB illustrates the result of the partitioning of each virtual network VO and Vl . It is significant to note that by providing partitions based on ports of a root switch within each virtual network of the network of FIG. IA, the combinational complexity has been reduced from determining a topology of a twenty node network to determining topologies of five smaller networks, the largest of which is a five node network, and then stitching and/or overlaying these topologies to form the topology of the original network. Additionally, as noted above, by finding all of the simply connected nodes to a root switch, the address tables that are lacking information can be augmented to include this newly discovered connectivity information, thereby further simplifying the subsequent connectivity determinations.
  • the above partitioning of networks can be applied recursively to similarly divide each partition into smaller partitions.
  • the five node partition ⁇ B, E, I, R, S ⁇ at port 3 of switch A can be further partitioned by determining that switch B is directly connected to this port, defining switch B as a root switch of a lower level partition, and determining the simply connected nodes to each port of switch B.
  • the complexity is further reduced from finding a topology of a five node network to finding a topology of a pair of two node networks.
  • the four node partition ⁇ I, J, N, Q) off port 1 of switch H can be partitioned based on the sets of nodes associated with the ports of switch N; and, the three node partition off port 3 of switch H can be partitioned based on the nodes associated with the ports of switch M.
  • the addressing tables of the nodes can be augmented with the newly discovered connectivity information, thereby progressively simplifying the process of topology determination by filling in as much information as possible.
  • FIG. 2 illustrates an example flow-diagram for finding the topology of a network in accordance with this invention.
  • the loop 210-295 is applied for each expressly defined virtual network in the overall network, as well as any other identified independent networks, herein termed implied virtual networks.
  • the loop 215-290 is provided to assure that the connectivity of all switches within the virtual network is determined; if a switch's connectivity has been addressed, the processing of this loop is bypassed, at 220.
  • the as-yet-unprocessed switch is set as the current root switch, at 225.
  • the only time a switch will arrive at block 230 from block 225 is when the switch is not simply connected to the selected root switch. For example, in virtual network Vl, if switch H is selected as the first root switch, the switches F, G, and P will remain unchosen after the connectivity of all of the other nodes that are simply connected to switch H have been determined.
  • any of a variety of techniques can be used for selecting the first root switch, including a random selection.
  • the best root switch is the switch whose largest partition is smaller than any other switch's largest partition.
  • the switch whose longest address table is smaller than any other switch's longest address table would generally be preferred.
  • the completeness of the address forwarding table is unknown, a switch with many ports and a fairly uniform distribution of addresses among the ports would generally be preferred to a switch with few ports, or a switch with a skewed distribution of addresses among the ports.
  • Other selection criteria may be used, based on other available information. For example, if spanning tree data is available, the root switch of the spanning tree may be an appropriate choice.
  • the subsequent process beginning at 230 attempts to find the next-root switch that is directly connected to each of the ports of the current root switch.
  • each of these root switches is recursively processed from this point to determine the directly connected root switch from each of its ports, and this recursion is repeated for each subsequently determined root switch.
  • One of ordinary skill in the art will recognize that other techniques for determining the topology of each partition off the ports of a root switch may also be used.
  • the ports of the root switch are assessed to determine whether this port, at the OSI data link layer (Layer 2), is coupled to one or more devices at the OSI network layer (Layer 3), rather than to another switch.
  • the forwarding table of each port is assessed to determine if the port is coupled to one or more switches; if not, a Layer 3 connection is inferred. If there are multiple non-switch entries in the address table, a shared segment (e.g. a bridge) is inferred to provide the coupling of these multiple devices to the port.
  • this new root switch is a leaf node, i.e. a node with no further switches coupled to its non-root ports, the processing of the current partition is completed.
  • the 'next' block 280 is used to indicate that either the next partition 285, or the next switch 290, will be processed, depending upon how this branch was reached during the recursive processing.
  • each switch that has not yet been identified as a root switch or a switch that is linked to the root switch is assessed to determine its connectivity to the root switch. As noted above, this determination can be made quickly using rule set (1) above. If a switch is determined to be simply connected to the root port of the root switch, that switch is marked as a potential member of a shared segment with this root switch.
  • the loop 250-285 is structured to process each partition to determine the next root switch, or next shared segment of switches, that is directly connected to the port providing this partition.
  • the determination of a direct connection is substantially simplified by eliminating each of the switches that cannot be directly connected to the port of the root switch corresponding to this partition. It can be shown that two ports p and q cannot be directly coupled if:
  • port 1 of switch R is known to be the port that provides the simple connection to port 3 of root switch A.
  • the intersection of the forwarding table of port 1 of switch R ⁇ A,B,C,E ⁇ and port 3 of switch A ⁇ B,E,I,S,R ⁇ (as augmented when switches I and R were determined to be simply connected to port A) is not null, and therefore switch R is eliminated from further processing for determining a direct connection to port 3 of switch A.
  • switches E, I, and S can also be eliminated, thereby leaving only port 1 of switch B as the only port that can be directly connected to port 3 of switch A.
  • a direct connection to the current port of the root switch cannot be found directly, or if there is uncertainty in the determination, other sources of connectivity information may be used to further narrow the field of potential next root switches. For example, a number of tools exist for discovering 'neighbors' in a network, such as CDP (Cisco Discovery Protocol) and others. Similarly, spanning-tree data is often available from nodes that use spanning-tree routing algorithms. If such information is available, it is used to resolve ambiguities, and/or validate inferences made in the above processes.
  • the 'spanning-tree cost' of each node can be used to identify the most likely next root switch; or, any node that is placed in a blocking mode by the spanning-tree routing process can be ignored as a candidate next root switch.
  • such data may be used to bypass the above processes for any partition whose topology is well defined by this data.
  • next-root or next-shared-segment
  • an elimination and priority ordering scheme is preferably used to select the next-root from the remaining simply connected switches.
  • the aforementioned AFT-Intersection test (2) is applied to eliminate any switch whose root port cannot be directly connected to the current port of the root switch. From the remaining switches, eliminate any switch that has a simple connection on its root port with another remaining switch, because this would imply that this other switch is between the root switch and this switch. If more than one switch remains after this elimination, the remaining switches are assigned a 'score', typically based on a heuristic assessment of ports that are likely to be connected together, and the next root switch is selected based on this score. If multiple switches have the same score, one is selected arbitrarily.
  • the address table of the root port of the candidate switch includes the root switch, AND the address table of the current port of the root switch includes the candidate switch. 2.
  • the Through-Set intersection of the root port of the candidate switch and the current port of the root switch is null.
  • the address table of the root port of the candidate switch includes the root switch, or, the address table of the current port of the root switch includes the candidate switch. [0051] If, at 260, a single switch is identified as the next root switch, that switch is selected and marked as such, at 265. If a plurality of switches are identified as a shared segment that is connected to the port of the current root switch, then each of the switches in the shared segment is identified as a hub node and processed as the next root switch, at 270. [0052] At 275, the direct link between the port on the root switch and the root port of the new root switch is merged with the links that have been determined thus far. If the link corresponds to a trunk, each physical link corresponding to this logical aggregate is merged.
  • FIG. 3 illustrates an example flow diagram for a prioritized merging of links.
  • shared segments are inferred when there is insufficient information available to infer a simple direct connection; preferably, if a direct connection inference conflicts with a shared segment inference, the direct connection inference takes priority.
  • each link is inferred and added to the database of inferred links, it is compared with previously inferred links to determine if a priority inference should be applied.
  • the processing is dependent upon whether the new link is inferred to be a direct connection or a shared segment, at 310. If the new link is a direct connection, the loop 350-370 checks each of the endpoints to see if it is included in a prior inferred link, at 355. If, at 360, the prior inferred link is a shared segment, this endpoint is removed from the inferred shared segment, at 365.
  • the loop 320-345 checks each of the endpoints to see if it is included in a prior inferred link, at 325. If the prior link is a direct connection, this endpoint is removed from this new shared segment inference, at 340. If the prior link is also a shared segment, the new shared segment is merged with the prior shared segment, at 335.
  • networks are often partitioned into substantially independent networks using techniques other than virtual LANs.
  • the HVNES product from OPNET Technologies, Inc. allows multiple VNE (Virtual Network Environment) servers to be arranged in a hierarchical master/slave arrangement, wherein each slave network is disparately managed, with minimal connectivity between or among nodes of the slave networks.
  • each slave network can be independently analyzed to infer the topology within its domain, while a smaller 'upper-level' network comprising only the nodes that interconnect the slave domains can be analyzed to provide a topology for 'stitching together' these slave networks.
  • a priority merging of inferred links would be used, such as discussed above with regard to FIG. 3, wherein links that are determined at the upper level are given priority to those discovered within the slave networks.
  • FIG. 4 illustrates an example block diagram of a link inference system in accordance with this invention.
  • This system may be a hardware embodiment, a software embodiment, or a combination of both.
  • the preferred system includes a network partitioning component 440 that is configured to partition an input network into partitions based on a given partitioning criteria, such as partitions based on membership in a virtual network, or membership based on being simply connected to ports of a particular switch.
  • the network partitioning component 440 is configured to determine each set of simply connected nodes based on the aggregate AFT tests (1), detailed above.
  • the system also includes a root switch finder 450 that is configured to identify a subsequent/next root switch associated with each of the partitions identified by the network partitioner 440.
  • the root switch finder component 450 is configured to eliminate candidate nodes in this determination by applying the ATF intersection rule (2) to each root port of each candidate switch.
  • the root switch finder 450 includes heuristic tests that are configured to identify a likely next root port from among the candidate nodes, as detailed above.
  • a link merger 460 receives the determined connectivity to this next root switch from the current root switch and merges this information with prior determined links to create a network topology 470, using, for example, the example flow diagram of FIG. 3.
  • the root switch finder 450 is also preferably configured to use ancillary information, such as spanning tree information 452 and CDP information 454, to validate and/or determine or infer the next root switch associated with each partition.
  • the network partitioner 440 is preferably configured to partition the remaining nodes according to their connectivity to ports of the current root switch.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computer Security & Cryptography (AREA)
  • Small-Scale Networks (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

Un réseau est partitionné en un ensemble de partitions indépendantes, et la topologie de chaque partition est déterminée, ensuite fusionnée pour former une topologie du réseau entier. De préférence, le partitionnement est hiérarchique, le réseau étant partitionné pour former des partitions de réseau local virtuel (VLAN) individuelles, et chacune des partitions VLAN étant en outre partitionnée sur la base de nœuds qui sont simplement connectés à chaque port d'un ou de plusieurs commutateurs racines sélectionnés à l'intérieur de la partition VLAN. Des connexions simples à chaque port sont déterminées de manière efficace sur la base d'une table de transfert d'adresse agrégée associée à chaque nœud. Des informations auxiliaires, telles que des données d'arborescence ou de CDP, peuvent être utilisées pour faciliter une partition efficace et/ou valider des inférences qui sont déduites avec des informations incomplètes.
PCT/US2007/074679 2006-08-01 2007-07-29 Inférence de liaison dans des réseaux importants sur la base de données incomplètes WO2008016861A2 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US82102206P 2006-08-01 2006-08-01
US60/821,022 2006-08-01

Publications (2)

Publication Number Publication Date
WO2008016861A2 true WO2008016861A2 (fr) 2008-02-07
WO2008016861A3 WO2008016861A3 (fr) 2008-07-10

Family

ID=38997783

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2007/074679 WO2008016861A2 (fr) 2006-08-01 2007-07-29 Inférence de liaison dans des réseaux importants sur la base de données incomplètes

Country Status (1)

Country Link
WO (1) WO2008016861A2 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
GB2464125A (en) * 2008-10-04 2010-04-07 Ibm Topology discovery comprising partitioning network nodes into groups and using multiple discovery agents operating concurrently in each group.
US7756954B2 (en) 2004-04-21 2010-07-13 Dell Products L.P. Method for heterogeneous system configuration
US8625457B2 (en) 2007-12-03 2014-01-07 International Business Machines Corporation Method and apparatus for concurrent topology discovery

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5706440A (en) * 1995-08-23 1998-01-06 International Business Machines Corporation Method and system for determining hub topology of an ethernet LAN segment
WO1998018306A2 (fr) * 1996-10-28 1998-05-07 Switchsoft Systems, Inc. Procede et appareil de generation d'une topologie de reseau
US6628623B1 (en) * 1999-05-24 2003-09-30 3Com Corporation Methods and systems for determining switch connection topology on ethernet LANs
US20040255184A1 (en) * 2003-05-27 2004-12-16 Lucent Technologies Inc. System and method for determining the physical topology of a network having multiple subnets
US20060133297A1 (en) * 2004-12-17 2006-06-22 Quist Daniel A Network topology mapper

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5706440A (en) * 1995-08-23 1998-01-06 International Business Machines Corporation Method and system for determining hub topology of an ethernet LAN segment
WO1998018306A2 (fr) * 1996-10-28 1998-05-07 Switchsoft Systems, Inc. Procede et appareil de generation d'une topologie de reseau
US6628623B1 (en) * 1999-05-24 2003-09-30 3Com Corporation Methods and systems for determining switch connection topology on ethernet LANs
US20040255184A1 (en) * 2003-05-27 2004-12-16 Lucent Technologies Inc. System and method for determining the physical topology of a network having multiple subnets
US20060133297A1 (en) * 2004-12-17 2006-06-22 Quist Daniel A Network topology mapper

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
BREITBART Y ET AL: "TOPOLOGY DISCOVERY IN HETEROGENEOUS IP NETWORKS: THE NETINVENTORY SYSTEM" IEEE / ACM TRANSACTIONS ON NETWORKING, IEEE / ACM, NEW YORK, NY, US, vol. 12, no. 3, 1 June 2004 (2004-06-01), pages 401-414, XP001198226 ISSN: 1063-6692 cited in the application *
LOWEKAMP B ET AL: "TOPOLOGY DISCOVERY FOR LARGE ETHERNET NETWORKS" COMPUTER COMMUNICATION REVIEW, ACM, NEW YORK, NY, US, vol. 31, no. 4, 1 October 2001 (2001-10-01), pages 237-248, XP001115760 ISSN: 0146-4833 cited in the application *
YANTAO SUN ET AL: "The physical topology discovery for switched ethernet based on connections reasoning technique" COMMUNICATIONS AND INFORMATION TECHNOLOGY, 2005. ISCIT 2005. IEEE INTE RNATIONAL SYMPOSIUM ON BEIJING, CHINA OCT. 12-14, 2005, PISCATAWAY, NJ, USA,IEEE, vol. 1, 12 October 2005 (2005-10-12), pages 42-45, XP010875543 ISBN: 978-0-7803-9538-1 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7756954B2 (en) 2004-04-21 2010-07-13 Dell Products L.P. Method for heterogeneous system configuration
US8625457B2 (en) 2007-12-03 2014-01-07 International Business Machines Corporation Method and apparatus for concurrent topology discovery
GB2464125A (en) * 2008-10-04 2010-04-07 Ibm Topology discovery comprising partitioning network nodes into groups and using multiple discovery agents operating concurrently in each group.

Also Published As

Publication number Publication date
WO2008016861A3 (fr) 2008-07-10

Similar Documents

Publication Publication Date Title
US8089904B2 (en) Link inference in large networks based on incomplete data
US8670352B2 (en) Link inference in large networks based on incomplete data
Haddadi et al. Network topologies: inference, modeling, and generation
US11201809B1 (en) Network topology generation using traceroute data
US7720009B2 (en) Virtual private network (VPN) topology identifier
US20060256733A1 (en) Methods and devices for discovering the topology of large multi-subnet LANs
US8165038B2 (en) Network physical connection inference for IP tunnels
JP2019536331A (ja) 対話型ネットワーク分析プラットフォームのためのシステムおよび方法
US9253038B2 (en) End-to-end network access analysis
US9537749B2 (en) Method of network connectivity analyses and system thereof
US11546227B2 (en) Optimized detection of network defect exposure in network environment
CN109672562B (zh) 数据处理方法、装置、电子设备及存储介质
US20100094994A1 (en) Network structure information acquiring method and device
CN110932906A (zh) 基于snmp技术的数据中心网络拓朴结构发现方法及其拓朴结构发现系统
US8914503B2 (en) Detected IP link and connectivity inference
US7782797B2 (en) Methods and apparatus for self partitioning a data network to prevent address conflicts
US10615999B2 (en) Virtual local area network configuration system and method, and computer program product thereof
US9729426B2 (en) Stitching together partial network topologies
US20040215781A1 (en) Techniques for determining device connectivity in a network using protocol-specific connectivity information
WO2008016861A2 (fr) Inférence de liaison dans des réseaux importants sur la base de données incomplètes
US20140236878A1 (en) Inferring connectivity in the presence of conflicting network data
US20040158780A1 (en) Method and system for presenting neighbors of a device in a network via a graphical user interface
US20070025328A1 (en) Tracing routing differences
Andreev et al. An algorithm for building an enterprise network topology using widespread data sources
Ma et al. An algorithm of physical network topology discovery in multi-VLANs

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

NENP Non-entry into the national phase

Ref country code: RU

122 Ep: pct application non-entry in european phase

Ref document number: 07813511

Country of ref document: EP

Kind code of ref document: A2