• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A method of extracting, from a physical topology, a virtual topology, the physical network topology including a plurality of switches connected to each other using their ports, the method comprising the following steps: sub-topology distribution of leaf switches according to the characteristics of the target topology, these subtopologies being included in the virtual topology; - inclusion of the neighboring switch (s) connected only to or to the switches (s) of the same subtopology to this subtopology; - Split neighboring switch (s) connected to a first subtopology and a second subtopology in two virtual switches, the first virtual switch comprising the ports of this neighbor switch with which it is connected to said first subtopology, the second virtual switch comprising the ports of this neighbor switch with which it is connected to said second subtopology.
    公开号:FR3037463A1
    申请号:FR1555428
    申请日:2015-06-15
    公开日:2016-12-16
    发明作者:Jean-Noel Quintin；Alain Cady
    申请人:Bull SA；
    IPC主号:
    专利说明:

    [0001] BACKGROUND OF THE INVENTION The present invention generally relates to the adaptation of a "high-speed" network topology to a specific routing algorithm, and more particularly to the adaptation of an unstructured topology to a routing algorithm for a structured topology. The logical structures of a broadband network, called topologies, are generally divided into two classes: structured and unstructured. Structured topologies denote topologies defined by a mathematical formalism (theoretical description). For example, the PGFT, for "Parallel Ports Generalized Fat Trees" (Zahavi E., "D-Mod-K-Routing Providing Non-blocking Traffic for Shift Permutations on Real Life Fat Trees", http: //webee.technion.acil / publication-link / index / id / 574, 2010) are described by a formula for reconstructing the topology from a set of factors. Hypercubes are also mentioned (Bhuyan, L. N., & Agrawal, D.P. (1984), "Generalized hypercube and hyperbus structures for a computer network", Computers, IEEE Transactions on, 100 (4), 323-333). These network structures have been theoretically studied to efficiently transmit messages between machines. To perform transmissions efficiently within the network, routing algorithms dedicated to these topologies have been developed. These algorithms are also present in the documents mentioned above. These algorithms are efficient but can only be used with the topologies according to the formalism for which they have been developed. Conversely, unstructured topologies do not follow specific formalism and / or have not been studied for routing. Generally, these are topologies that do not respect a particular construction. They are potentially close to a formalism but can not be taken into account by the routing algorithm. It follows that a structured topology favors the performance of routing algorithms, both in terms of routing efficiency and in terms of their execution time. The use of routing algorithms for structured topologies such as "Flattened-Butterfly", "HyperX", "Tore", or "PGFT" confirm the interest of this type of network topology. However, a structured network topology is expensive both in number of network switches (or in English terminology "switch") than cables. Thus, reducing the cost of the network infrastructure by decreasing the number of network switches while maintaining performance in terms of routing efficiency has become a concern of network architects.
    [0002] In this case, by reducing the number of network switches, the theoretical characteristics of the topology are no longer preserved and the routing algorithms that are specific to it are therefore no longer usable in the state. Indeed, knowing that a step of analysis and validation of the structure of the physical topology is generally required to perform a topology routing, if the structure of a topology is modified so that it is no longer suitable to the routing algorithm that is previously intended, the routing can be achieved according to this routing algorithm. In this respect, the known solutions, such as that proposed by the "OpenSM" software for "InfiniBand", deal only with the physical topology. In particular, according to these solutions, if the characteristics extracted from a modified physical topology do not correspond to the expectations of the routing algorithm, a fallback solution based on a more generic algorithm, less efficient and more expensive in execution time is generally adopted. In addition, most current network switches use a single routing table. It is, therefore, not possible to consider this type of network switches as several separate entities to offset a reduction in their number. Otherwise, the routing table of a network communicator would be written multiple times with conflicting data that can not be merged. In addition, for efficient routing, a destination should be reached by two different links regardless of the link through which the packet arrives, which can not be achieved with a single network switch routing table. An object of the present invention is, therefore, to overcome the aforementioned drawbacks.
    [0003] Another object of the present invention is to adapt a network topology, which following a modification in its structure (in particular by reducing the number of network switches) is no longer adapted to the routing algorithm which is specific, to the same routing algorithm so as to maintain the routing performance in this high-speed network. Another object of the present invention is to be able to reduce the cost of a structured topology while remaining adapted to a routing algorithm intended for a structured topology. To this end, according to a first aspect, the invention relates to a method of extracting, from a physical network topology, a virtual topology adapted to a predefined routing algorithm, the physical topology of network including a plurality of switches connected to each other by means of their ports for receiving and / or transmitting packets to or from leaf switches of said physical network topology, said network physical topology being different from one network physical topology of predefined type of network topology adapted to the predefined routing algorithm, this method comprising the following steps: a) identification, in the physical network topology, of a plurality of 25 sub-topologies, these sub-topologies; -topologies that do not have a switch in common, each subtopology being of the predefined type of network topology and including at least one leaf switch, no a subtopology not connected to a switch of another subtopology; B) expanding each subtopology by including one or more switches directly connected to it and not directly connected to any other subtopology so that the extended subtopology is of the predefined type of network topology; 3037463 4 C) when a switch is directly connected to more than one subtopology, splitting this switch into as many virtual switches, each virtual switch comprising the split switch port (s) by which this split switch is 5 connected to the switches (s) of a subtopology, the ports with which the split switch is connected to a switch not yet belonging to any subtopology being distributed between the virtual switches, these virtual switches being considered as switches; D) repeating steps b) and c) until each switch is included in a subtopology, the virtual topology being composed of said subtopologies. Advantageously, this method further comprises a step of use by the routing algorithm of the extracted virtual topology.
    [0004] Advantageously, the extracted virtual topology is a structured topology adapted to a routing algorithm intended for a structured topology, a structured topology being a topology defined by a predetermined mathematical formalism. The predefined type of network topology is selected from a list comprising a PGFT network topology, a 2D torus network topology, a 3D torus network topology, a HyperX network topology. According to a second aspect, the invention relates to a routing method in an unstructured physical topology of a high-speed network according to a routing algorithm for a structured topology, a structured topology being a topology defined by a predefined mathematical formalism, this method comprises the following steps: - discovery of the unstructured physical topology; - analysis and validation of the discovered physical topology; Extracting, according to the extraction method presented above, a virtual topology adapted to the routing algorithm; - analysis and validation of the structure of the virtual topology ex- tract; 3037463 5 - calculation of the routing tables adapted to the extracted virtual topology; merge the routing tables associated with the physical topology so as to obtain routing tables adapted to the structured virtual topology; - Loading, in the extracted virtual topology, the merged routing tables. According to a third aspect, the invention relates to a computer program product implemented on a memory medium, which can be implemented within a computer processing unit and includes instructions for implementing it. of the method summarized above. Other characteristics and advantages of the invention will appear more clearly and concretely on reading the following description of preferred embodiments, given below with reference to the appended drawings in which: FiG.1 schematically illustrates a structured network topology of any type; FiG.2 schematically illustrates a PGFT-type structured network topology ("Parallel Ports Generalized Fat-Trees") used in the following to illustrate an embodiment; FiG.3 schematically illustrates an unstructured network topology for extracting a structured virtual topology adapted to a routing algorithm for a structured network topology; and FiG.4 and FiG.5 schematically illustrate steps of a method of extracting a structured virtual topology from an unstructured network topology according to one embodiment. FIG. 1 shows a structured topology 10. By way of nonlimiting examples, the structured physical topology 10 is of the "PGFT", Hypercube, Butterfly, HyperX, 2D torus, 3D torus or De Bruijn graph type. .
    [0005] A specific routing protocol is implemented in this structured physical topology 10. This routing protocol allows the establishment of route (s), or path (s), between a source node and a destination node of the network. A route is formed by the succession of links 15 and network switches 11-14 to borrow to connect these two nodes. In this regard, network switches 11-14 are configured to perform distributed routing based on routing elements (e.g., routing tables) defining the rules for routing data packets to their destination.
    [0006] In addition, a network switch 11-14 includes a port routing table 0-3. The routing elements are independent of the input port 0-3 of a network switch 11-14 through which data to router arrives. On the other hand, the link 15 on which data is routed depends on the input ports 0-3 and the destination of this data.
    [0007] The implementation of a routing protocol in the structured physical topology is generally preceded by the following steps: discovery of the structured physical topology of the network; - analysis and validation of the structure of this structured physical topology 10; The calculation of the routing tables; and - the loading of the routing tables. Figure 2 illustrates a structured physical topology 200 comprising switches 20-29 and in which a PGFT specific routing algorithm may be implemented. The particular type of PGFT topology is given here for illustrative purposes only and is in no way limiting. The embodiments described above extend directly and unambiguously to different types of topology such as, but not limited to, Hypercube, Butterfly, HyperX, 2D torus, 3D torus, or De Bruijn graph. Structured topologies are expensive, mainly because of the number of top-of-the-line switches (so-called "top switches"). In this case, the structured physical topology 200 of FIG. 2 is a PGFT comprising two switches 20-21 from the top of the network. Moreover, in the theoretical framework of the PGFT routing, it is not possible to reduce the number of switches 20-21 from the top of the network without losing characteristic properties of the topology. By way of example, by replacing in FIG. 3 the switches 20-21 of the top of the network of FIG. 25 by a single high-end switch 30, the structure of the resulting unstructured topology 300 is not more suited to the PGFT specific routing algorithm. The physical network topology 300 is different from the physical topology 200 of PGFT type network. It follows that the modified structure of the unstructured topology 300 can not be validated, and the PGFT routing can not be realized (in particular because from the switch 30 of the top of the network there is at most one path to a leaf switch 35-38 through the switches 31-34). More generally, in order to maintain the routing algorithm for the structured topology 200 (in this example, PGFT) and to be able to implement it in the unstructured topology 300, a structured virtual topology is extracted from the topology Unstructured 300. More generally, for the routing in an unstructured topology 300, resulting from the modification of a structured topology 200, this routing being according to a routing protocol intended for the structured topology 200, the following 20 steps are performed: - discovery of the modified physical topology 300; - analysis and validation of the discovered physical topology; extracting a structured virtual topology adapted to the routing algorithm that is intended for a structured topology; Analysis and validation of the structure of the structured virtual topology; - calculation of the routing tables adapted to the structured virtual topology; merge the routing tables associated with the physical topology so as to obtain the routing tables adapted to the structured virtual topology; - loading the merged tables.
    [0008] The step of extracting a structured virtual topology adapted to the routing algorithm intended for a structured topology has the effect of: decorrelating the memory topology of the physical topology; and 5 - separating the network switches into logically independent entities. To illustrate the step of extracting a structured virtual topology adapted to the routing algorithm, the unstructured topology 400 (FIG. 4) is used by way of example only. The unstructured topology 400 is said to be unstructured because, among other things, the top-of-network switch 41 has two different paths to reach a 45-48 leaf switch. This topology does not represent, therefore, a PGFT. In this example, a PGFT-structured virtual topology 500 is extracted from an unstructured physical topology 400 that is not a PGFT. The routing algorithms subsequently use only the structured virtual topology, and not the unstructured (real) physical topology 400 that is not supported. To identify such a structured virtual topology, we begin by considering the leaf switches 45-48. The 45-48 leaf switches are those that are directly connected to the compute nodes. Each of the sheet switches 45-48 is placed in a subtopology G1: 45, G2: 46, G3: 47 and G4: 48. The subtopologies G1-G4 have no switch in common and each includes a leaf switch. .
    [0009] Each of the G1-G4 subtopologies is considered to be of the PGFT type, as is the type of the physical network topology 200. The subtopologies G1-G4 having exactly the same neighbors are then merged. In this case, G1 and G2 are merged into an SG subtotology. Likewise, G3 and G4 are merged into an SG2 subtopology. In other words, the subtopologies G1-G4, respectively constituted by the leaf switches 45-48, are merged into the subtopologies SG1 and SG2 so that the subtopologies G1-G4 having the same neighboring switches are located. in the same subtopology SG1 or 3037463 9 SG2 (the switch (s) neighboring a certain switch 41-48 are the set of switches directly connected to this switch). The new SG1 and SG2 subtopologies are of the PGFT type. In other words, by identifying the switches 42-44 directly connected to the switches 45-48 of the unstructured physical topology 400 (or the neighboring switches 42-44 of the switches 45-48), the switches 45 -48 having the same neighboring switches are divided into SG1-SG2 subtopologies. These SG1-SG2 subtopologies are included in the virtual topology to be extracted.
    [0010] Switches 42-44 which are above the leaf switches 45-48 are now considered in the unstructured physical topology 400. Switches 42-44 are the direct neighbors (or first neighbors) of the leaf switches 45-48. Two adjacent neighbor switches 42-44 of the leaf switches 45-48 are separated if they are connected to two different SG1-SG2 subtopologies. Thus, a switch 42-44 which is connected only to the switch (s) of the same subtopology SG1-SG2 is included (or added) in this subtopology SG1-SG2. In this case, the switch 42 is connected only to the subtopology SG1. The switch 44 is connected only to the subtopology SG2.
    [0011] The switches 42 and 44 are thus included respectively in the subtopology SG1 and the subtopology SG2. On the other hand, the switch 43 is connected to the two subtopologies SG1 and SG2. This switch 43 is split into two virtual switches 43-SG1 and 43-SG2 respectively included in the subtopology SG1 and the subtopology SG2.
    [0012] The ports of the physical switch 43 which has just been split into two virtual switches 43-SG1 and 43-SG2 are to be distributed between these two virtual switches 43-SG1 and 43-SG2. For this, we start by considering one of the SG1-SG2 subtopologies (in this case, SG1). An outbound link of SG1 is selected with originated a lower level sub-topology G1-G2 (G1 for example) and connected to virtual switch 43-SG1. The link connecting port 5 of switch 45 to port 0 of switch 43 (i.e., the link [45: 43: 0]) is considered. Next, the links connecting the other lower level subtopologies included in SG1 to the virtual switch 43-SG1 are searched. The link connecting the port 5 of the switch 46 to the port 1 of the switch 43 (that is to say the link [46: 5 -> 43: 1]) is added to the links considered. Since all lower level subtopologies in SG1 (in this case G1 and G2) are considered, ports 0 and 1 of switch 43 are assigned to virtual switch 43-SG1. The virtual switch 43-SG1 with its ports 0 and 1 is added to the subtopology SG1. Similarly, for the subtopology SG2, the ports 2 and 3 of the switch 43 are assigned to the virtual switch 43-SG2 which is added to the subtopology SG2. As a result, the physical switch 43 which is connected to the sub-topology switches SG1 and / or switches of the sub-topology SG2 is split into two virtual switches 43-SG1 and 43-SG2. The virtual switch "43-SG1", comprising the ports 0-1 with which it is connected to or switches of the subtopology SG1 is assigned to this subtopology SG1. The virtual switch 43-SG1, comprising the ports 2-3 with which it is connected to the switches of the second subtopology SG2, is assigned to this sub-topology SG2. The "uplink" ports 4-5 of the physical switch 43 are also to be distributed between the subtopologies SG1 and SG2 included in the structured virtual topology. In this respect, the ports by means of which the split switch 41 is connected to a switch which does not yet belong to any subtopology, in this case the connector 41, are distributed between the virtual switches 43-SG1 and 43 -SG2. This distribution can be done arbitrarily. In the example of FIG. 4, the ports 4 and 5 of the switch 41 are respectively assigned to the virtual switch 43-SG1 and 43-SG2.
    [0013] As for switches 42-44, the switches above are now considered in an unstructured physical topology 400. In this example, only switch 41 is connected to more than one subtopology. This switch is therefore split into as many virtual switches as sub-topologies to which it is connected, that is to say two virtual switches 41-SG1 and 41-SG2. Indeed, for links belonging to switches shared between two subtopologies (SG1 and SG2), one of these shared links is considered by placing it arbitrarily in one of the subtopologies. Links connecting to other subtopologies via a switch, virtual or not, not yet considered are searched. Virtual switch 41-SG1 includes ports 0 and 3. The second virtual switch 41-SG2 includes ports 1 and 2. The new subtopologies SG1 and SG2 are of type PGFT. More generally, the extraction of the virtual topology is performed by successive approaches from the leaf switches 45-48 to the switches of the top of the network (top switch) 41, adding to each approach to the switches at the top of the network the switches to the 15 sub-topologies to which they are only connected and splitting / splitting the switches connected to more than one subtopology into as many virtual switches. FIG. 5 shows the structured virtual topology, composed of the subtopology SG1 and SG2, which has just been extracted from the unstructured physical topology 400. In this illustrative example, each of the two physical switches 41 and 43 is cut off. in two virtual switches. The extracted structured virtual topology includes two more switches than the actual topology. Considering the two virtual switches 43-SG1 and 43-SG2 produced from the physical switch 43, one consisting of the ports 0 and 1 and the other of the ports 2 and 3, the routing algorithm passes the packets. arriving on ports 0 and 1 to the subtopology SG1, without the possibility of reaching the other subtopology SG2, and vice versa. In addition, the routing elements "in particular the routing tables" are independent. Thus no conflict is possible: the routing tables of a port can be written to a value without disturbing or influencing the other ports. It is, therefore, not possible to corrupt memory. In addition, the virtual switching of the switches makes it possible to rearrange the memory topology into a suitable topology for the specific routing algorithms, while reducing costs. The structured virtual topology retrieved from an unstructured physical topology (itself derived from a modified structured physical topology) retains the original routing algorithm. Indeed, in this example, the extracted structured virtual topology is a real PGFT adapted to the routing protocol. Routing algorithms dedicated to physical topologies of this type will work on virtual topologies belonging to this same type of topology. More generally, the topology used by the routing algorithm is based on the virtual topology extracted over the physical and logical networks. Maintaining a virtual structured topology in accordance with the routing algorithm thus makes it possible to maintain the performance of the latter. It should be noted that, although the type, considered above, of the network topology is the PGFT (a network topology based on a tree structure), the same approach applies to other types network topologies so that the method of extracting the virtual topology can be extended to other topologies. Indeed, as non-limiting examples, in the case of a 2D torus network topology (a network topology based on a rectangle), the first subtopologies G1-G4 are each leaf switch ( switches directly connected to compute nodes). Starting from a Tore 2-D subtopology, this network subtopology is extended to include other switches so that the new subtopology remains of the Tore 2-D type. For this, one side of the size n rectangle of a Tore 2-D type topology is taken into consideration to search for switches connected to each of its lines. When a switch is connected to different subtopologies, it can be divided into several virtual switches to extend the dimension of these subtopologies; in the case of a 3D torus network topology (a block-based network topology), the first subtopologies are each leaf switch (switches having connected computing nodes). Starting from a subtopology, we search for a set of other subtopologies allowing extension according to the basic scheme of a Tore 3-D topology which is a block. Each side of the keypad is taken into consideration to search for switches that are connected to it. If a switch is connected to different subtopologies, then this switch is divided into as many virtual switches; in the case of a network topology of the HyperX type (a network topology based on a set of connected switches all interconnected), the first subtopologies may be the leaf switches (switches directly connected to calculation nodes) . The set of subtopologies is extended in parallel by finding all the adjacent switches interconnected.
    [0014] Switches belonging to the same subtopology must be connected to the switches of this same subtopology. If a switch is connected to several sub-topologies, it is then split into as many virtual calculators which will be considered as switches later on to extend the sub-topologies. Due to the recursion of the topology, the extension for HyperX is close to that of PGFT; composition of sub-topologies from the previously identified sets of switches: the leaf switches (45-48) having the same neighbor switches initialize the subtopologies (SG1 and 5G2). These sub-topologies will be merged to generate the global virtual topology. Obtaining a structured virtual topology from an unstructured physical topology advantageously makes it possible to generalize the use (and thus widen the field of application) of the routing algorithms for physical topologies. unsupported without having to modify these routing algorithms dedicated to these physical topologies; - reduce the cost of network infrastructure while maintaining network performance in terms of routing; To uncouple the physical topology from the routed topology; - to split a physical switch into virtual switches thanks to the tables by ports. Advantageously, the method described above for extracting, from a physical network topology, a virtual topology adapted to a predefined routing algorithm, allows: - the splitting of a physical switch into virtual switches : Each physical switch port may belong to a different virtual switch. Thus, a virtual switch is a port set belonging to the same physical switch; the identification of a set of switches that can be used as a basis for the construction of a target virtual topology. For example, for a PGFT, the leaf switches are supposed to represent the switches of the first level of the PGFT. This first can be formed of physical or virtual switches; the extension of the basic set by splitting the physical switches so that the structure of the extracted virtual topology coincides with the structure dedicated to the routing algorithm (for example PGFT, Torus, HyperX, dragonfly).
    [0015] It should be noted that the terms "network communicator", "switch" or "switch" are used interchangeably to refer to any routing equipment in a high-speed network.
    权利要求:
    Claims (6)
    [0001]
    REVENDICATIONS1. A method of extracting, from a physical network topology (400), a virtual topology (500) adapted to a predefined routing algorithm, the physical network topology (400) including a plurality of switches (41). -48) connected to each other by means of their ports (0-5) for receiving and / or transmitting packets to or from leaf switches (45-48) of this physical network topology (400), this the physical topology (400) of the network being different from a physical topology (200) of network of predefined type of network topology adapted to the predefined routing algorithm, this method comprising the following steps: a) identification, in the topology network (400) of a plurality of subtopologies (G1-G4), these subtopologies (G1-G4) having no switch (41-48) in common, each subtopology being predefined type of network topology and including at least one leaf switch (45 -48), no switch of a subtopology being connected to a switch of another subtopology; b) extending each subtopologies by including one or more switches directly connected thereto and not directly connected to any other subtopology so that the extended subtopology is of the predefined type of network topology; c) when a switch (43) is directly connected to more than one subtopology, splitting this switch into as many virtual switches (43-SG1, 43-SG2), each virtual switch including the one or more ports of the split switch with which this split switch is connected to the switches (s) of a subtopology, the ports with which the split switch is connected to a switch that does not yet belong to any subtopology being distributed between the virtual switches, these virtual switches 30 being considered as switches; d) repeating steps b) and c) until each switch is included in a subtopology, the virtual topology being composed of said subtopologies. 3037463 16
    [0002]
    2. The method of claim 1, further comprising a step of use by the routing algorithm of the extracted virtual topology.
    [0003]
    3. The method of claim 1 or 2, wherein the extracted virtual topology is a structured topology adapted to a routing algorithm for a structured topology, a structured topology being a topology defined by a predetermined mathematical formalism.
    [0004]
    The method according to any of the preceding claims, wherein the predefined type of network topology is selected from a list comprising a PGFT network topology, a 2D torus network topology, a 3D torus network topology, a HyperX network topology.
    [0005]
    5. A routing method in an unstructured physical topology of a high-speed network according to a routing algorithm for a structured topology, a structured topology being a topology defined by a predetermined mathematical formalism, this method comprising the following steps : - discovery of the unstructured physical topology; - analysis and validation of the discovered physical topology; extracting, according to a method presented in any one of claims 1 to 4, a virtual topology adapted to the routing algorithm; - analysis and validation of the structure of the extracted virtual topology; - calculation of the routing tables adapted to the extracted virtual topology; merge the routing tables associated with the physical topology so as to obtain routing tables adapted to the structured virtual topology; - Loading, in the extracted virtual topology, the merged routing tables.
    [0006]
    6. Computer program product implemented on a memory medium, capable of being implemented in a computer processing unit and comprising instructions for the implementation of a method according to one of claims 1 at 4.
    类似技术:
    公开号 | 公开日 | 专利标题
    EP3107253B1|2020-07-08|Transformation of unstructured network infrastructures into structured virtual topologies adapted to specific routing algorithms
    US7561534B2|2009-07-14|Methods of network routing having improved resistance to faults affecting groups of links subject to common risks
    EP3155771B1|2019-11-27|Data routing method and switch for network
    CN103312615B|2016-07-13|Use the longest prefix match search of variable number prefix
    US20100027442A1|2010-02-04|Constructing scalable overlays for pub-sub with many topics: the greedy join-leave algorithm
    CN108718278B|2021-04-27|Message transmission method and device
    FR2948247A1|2011-01-21|METHOD AND SYSTEM FOR HIGH PERFORMANCE AND AUTOMATED MANAGEMENT OF VIRTUAL NETWORKS.
    FR2784525A1|2000-04-14|PROCESS FOR ALLOCATING ORTHOGONAL CODES IN A MOBILE RADIOCOMMUNICATIONS SYSTEM OF THE MULTIPLE ACCESS TYPE BY CODES DISTRIBUTION USING CODES OF VARIABLE LENGTH
    CN106936680B|2020-08-07|System and method for intercommunication among heterogeneous networks of cloud computing platform
    EP2572478B1|2017-07-19|Method of optimizing routing in a cluster comprising static communication links and computer program implementing this method
    US8842543B2|2014-09-23|Trap-free shortest link-disjoint paths
    US7729353B1|2010-06-01|System, method and computer program product for discovering network connectivity among network devices based on topological information
    CN106878170B|2020-04-21|Method and device for determining forwarding path
    Enyedi2011|Novel algorithms for IP fast reroute
    JP5673667B2|2015-02-18|Packet classifier, packet classification method, packet classification program
    EP3531641B1|2020-10-28|Method for establishing communication routes between nodes of a cluster of computers, corresponding computer program and cluster of computers
    US10965532B2|2021-03-30|Software-defined networking device based on service profile learning
    US8842544B2|2014-09-23|Trap-free shortest link-and-shared risk link group-disjoint paths
    Risso et al.2013|Using metaheuristics for planning resilient and cost-effective multilayer networks
    FR3025905A1|2016-03-18|ALLOCATION OF RESOURCES
    WO2008016861A2|2008-02-07|Link inference in large networks based on incomplete data
    Braem et al.2014|Mapping a community network
    EP2727403B1|2020-05-13|Method for extinguishing routers in a communications network and router implementing this method
    FR2960732A1|2011-12-02|METHOD FOR PSEUDO-DYNAMIC ROUTING IN A CLUSTER COMPRISING STATIC COMMUNICATION LINKS AND COMPUTER PROGRAM USING THE SAME
    EP3671456A1|2020-06-24|Method for deploying a task in a supercomputer, method for implementing a task in a supercomputer, corresponding computer program and supercomputer
    同族专利:
    公开号 | 公开日
    US9985868B2|2018-05-29|
    EP3107253B1|2020-07-08|
    US20160366045A1|2016-12-15|
    FR3037463B1|2017-06-23|
    EP3107253A1|2016-12-21|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    WO2002062076A2|2001-01-31|2002-08-08|Tulip Networks Ltd.|Distributed multicasting for an atm network|
    US20140204805A1|2013-01-23|2014-07-24|Cisco Technology, Inc.|Multi-Node Virtual Switching System|
    US20150139035A1|2013-11-15|2015-05-21|Cisco Technology, Inc.|Dynamic Virtual Port Instantiation for Scalable Multitenant Network|FR3078220A1|2018-02-22|2019-08-23|Bull Sas|METHOD OF ESTABLISHING COMMUNICATION ROUTES BETWEEN NODES OF A CLUSTER OF COMPUTERS, COMPUTER PROGRAM AND CLUSTER OF CORRESPONDING COMPUTERS|US8145760B2|2006-07-24|2012-03-27|Northwestern University|Methods and systems for automatic inference and adaptation of virtualized computing environments|
    US7991859B1|2009-12-28|2011-08-02|Amazon Technologies, Inc.|Using virtual networking devices to connect managed computer networks|
    WO2013020126A1|2011-08-04|2013-02-07|Midokura Pte. Ltd.|System and method for implementing and managing virtual networks|
    CN104704759B|2012-10-08|2017-11-24|华为技术有限公司|The transmitting function virtualization of optical-fiber network based on wavelength-division multiplex |
    ES2617196T3|2013-04-19|2017-06-15|Entuity Limited|Route identification in a network of mixed routing / switching devices|
    US20140369230A1|2013-06-18|2014-12-18|Alcatel-Lucent Usa Inc.|Virtual Chassis Topology Management|
    US9602387B2|2014-12-29|2017-03-21|Juniper Networks, Inc.|Network topology optimization|
    US9780909B2|2014-12-29|2017-10-03|Juniper Networks, Inc.|Network topology optimization with feasible optical paths|FR3076142A1|2017-12-21|2019-06-28|Bull Sas|METHOD AND SERVER OF TOPOLOGICAL ADDRESS ALLOCATION TO NETWORK SWITCHES, COMPUTER PROGRAM AND CLUSTER OF CORRESPONDING SERVERS|
    FR3100069B1|2019-08-23|2021-10-29|Bull Sas|A rapid method of establishing communication routes between computers in a supercomputer|
    法律状态:
    2016-05-24| PLFP| Fee payment|Year of fee payment: 2 |
    2016-12-16| PLSC| Publication of the preliminary search report|Effective date: 20161216 |
    2017-05-23| PLFP| Fee payment|Year of fee payment: 3 |
    2018-05-25| PLFP| Fee payment|Year of fee payment: 4 |
    2020-06-26| PLFP| Fee payment|Year of fee payment: 6 |
    2021-06-25| PLFP| Fee payment|Year of fee payment: 7 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1555428A|FR3037463B1|2015-06-15|2015-06-15|TRANSFORMATION OF UNSTRUCTURED NETWORK INFRASTRUCTURES TO STRUCTURED VIRTUAL TOPOLOGIES ADAPTED TO SPECIFIC ROUTING ALGORITHMS|FR1555428A| FR3037463B1|2015-06-15|2015-06-15|TRANSFORMATION OF UNSTRUCTURED NETWORK INFRASTRUCTURES TO STRUCTURED VIRTUAL TOPOLOGIES ADAPTED TO SPECIFIC ROUTING ALGORITHMS|
    EP16170087.7A| EP3107253B1|2015-06-15|2016-05-18|Transformation of unstructured network infrastructures into structured virtual topologies adapted to specific routing algorithms|
    US15/173,789| US9985868B2|2015-06-15|2016-06-06|Transformation of unstructured network infrastructures into structured virtual topologies suitable for specific routing algorithms|
    [返回顶部]