JP2010199972A - Path control method, and node device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve control of a quality and efficient utilization of fixtures in a large area overlay network constituted of many nodes. <P>SOLUTION: Each node 1 includes: a cluster deciding section 12 which autonomously decides a cluster belonging thereto on the overlay network, and produces logic network configuration information 110; a link information measuring section 13 which periodically measures distance information between the node 1 and an arbitrary node inside the cluster adjacent to the cluster belonging thereto to set the measured information as link information 111; a link information sending/receiving section 14 which sends and receives the link information 111 between the node 1 and the other node; a routing table producing section 15 which produces topology information based on the link information, and produces a routing table 113 based on the topology information; and a path control section 16 which carries out packet transfer according to the routing table 113. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a path control method and a node device for ensuring connectivity, improving quality, and controlling traffic in a super-multi-node environment such as a wide area network such as the Internet or a sensor network.

  In the conventional route control, a static route control method is used for a relatively small and narrow network. In medium-sized networks, protocols such as broadcast and RIP (Routing Information Protocol) are often used.

  In slightly larger networks (several hundreds of nodes), OSPF (Open Shortest Path First) and IS-IS (Intermediate System to Intermediate System) are mainly used, but there are problems with scalability as the number of nodes increases. There is. Therefore, a method of sharing the routing table between nodes or statically connecting the partitioned OSPF / IS-IS areas to each other is also employed for improving the scalability.

In the case of a larger Internet area, it is stipulated that networks be interconnected by BGP (Border Gateway Protocol).
Thus, in a wide area and multi-node network, at present, OSPF / BGP which performs hierarchical control as described above is mainly used.

  FIG. 5 shows the hierarchical structure of Internet routing. FIG. 5 shows that static routing is used in a relatively small and narrow network area 100, and OSPF and IS-IS are used in a slightly larger network area 101. An area 102 indicates that BGP is used.

FIG. 6 shows a processing flow of a general link state type routing control system. Each router 200 has a link 201 connected to another router 200. Information representing a network composed of the router 200 and the link 201 is physical network information. Each router 200 holds this physical network information in advance.
Each router 200 measures / acquires connection information (presence / absence of connection and link quality) with the adjacent router 200 (step S1). This measured / acquired information is used as link information.

Subsequently, each router 200 transmits link information to another router 200 (step S2). This link information transmission is called flooding in OSPF.
Next, each router 200 creates topology information based on the collected link information, and creates a routing table 202 based on a Dijkstra algorithm or the like (step S3).
Each router 200 transfers the packet according to the routing table 202 (step S4).

In the existing methods as described above, the hierarchy is fixed, so when performing optimization based on some index such as shortest path search, path search to avoid congestion, optimization of resource usage, the optimization range is It remains a partial one, and because of its mechanism, only local optimization can be realized.
Overlay routing has been proposed as a technique that enables optimization in a wide area (see Non-Patent Document 1). Overlay routing is application level routing performed in the overlay network.

Go Hasegawa, “Overlay Routing and Network Free-riding Problem”, JANOG22 Meeting, Jul, 2008

  However, overlay routing is currently a mechanism that performs route calculation intensively in one place, and there is a problem that even if it can cope with a wide area, it can not cope with multi-node / large scale. .

  In view of the above-described problems, an object of the present invention is to realize quality control and efficient use of equipment in an overlay network composed of a large number of nodes over a wide area.

The path control method of the present invention includes a cluster determination procedure for autonomously determining a cluster to which each node device belongs on an overlay network, and creating logical network configuration information describing the belonging cluster and adjacent clusters, and each node device A link information measurement procedure for periodically measuring distance information with an arbitrary node device in a cluster adjacent to a cluster to which the cluster belongs based on the logical network configuration information, and using the measured information as link information; The topology information is created based on the link information transmission / reception procedure in which the device transmits / receives the link information to / from other node devices, and the link information measured by each node device and the link information received from other node devices. Creating a routing table based on this topology information , In which each node apparatus is characterized in that it comprises a routing procedure for packet forwarding in accordance with the routing table.
Further, in one configuration example of the route control method of the present invention, the cluster determination procedure includes a procedure for periodically reviewing a cluster to which the node device belongs based on network quality and updating the logical network configuration information. It is a feature.
Further, in one configuration example of the route control method of the present invention, the link information transmission / reception procedure receives the link information transmitted from another node device and sends a search request to the other node device. It includes a procedure for acquiring missing link information.
In one configuration example of the routing control method of the present invention, the routing table creation procedure creates the topology information by adopting information having the latest measurement time among a plurality of received link information. It includes a procedure for statistically processing a plurality of link information to create the topology information.
Further, in one configuration example of the route control method of the present invention, the route control procedure transfers a packet addressed to a transfer destination node device according to the routing table, and simultaneously transfers a node device belonging to an arbitrary cluster to the transfer destination node. It is equivalent to a device, and includes a procedure for transferring the packet to a node device belonging to this arbitrary cluster.

  Further, the node device of the present invention autonomously determines a cluster to which the node device belongs on an overlay network, creates a logical network configuration information describing the cluster to which the node belongs and an adjacent cluster, and the logical device Based on the network configuration information, the distance information is periodically measured with any node device in the cluster adjacent to the affiliated cluster, and the link information measurement means that uses the measured information as link information, and other node devices Creates topology information based on link information transmission / reception means for transmitting / receiving the link information between itself, link information measured by the own node device and link information received from another node device, and creates a routing table based on the topology information Routing table creation means for performing the processing according to the routing table. It is characterized in that and a path control means for Tsu-forwarding.

  According to the present invention, nodes on an overlay network are clustered, and autonomous decentralized route control is performed in units of clusters, thereby constructing a link state-based routing infrastructure in a wide area network composed of a large number of nodes. Therefore, various quality control and efficient use of facilities can be realized.

  In the present invention, each node periodically reviews the cluster to which the node belongs based on the network quality, so that the cluster configuration can be appropriately and autonomously reviewed when the network quality is extremely deteriorated, for example.

  In the present invention, the link information transmitted from the other node device is received and a search request is transmitted to the other node device to acquire the missing link information. It can correspond to a network consisting of nodes.

  In the present invention, topology information is created by adopting information having the latest measurement time among a plurality of received link information, or topology information is created by statistically processing a plurality of received link information. Even when the reliability and completeness of the link information is low, more appropriate topology information can be created.

  In the present invention, a packet is forwarded to a transfer destination node device according to the routing table, and at the same time, a node device belonging to an arbitrary cluster is made equivalent to the transfer destination node device, and also to a node device belonging to the arbitrary cluster. Forwarding reliability can be improved by transferring the packet.

It is a block diagram which shows the structure of the overlay network which concerns on embodiment of this invention. It is a block diagram which shows the structure of each node which participates in the overlay network which concerns on embodiment of this invention. It is a flowchart which shows the path control operation | movement of each node which participates in the overlay network which concerns on embodiment of this invention. It is a figure which shows the mode of acquisition of the link information in embodiment of this invention. It is a figure which shows the hierarchical structure of internet routing. It is a figure which shows the flow of a process of a link state type routing control system.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an overlay network according to an embodiment of the present invention.
An overlay network is logically formed on a wide area network such as an IP network. The overlay network has a large number of node devices (hereinafter abbreviated as nodes) 1-a, 1-b, 1-c, 1-d, and 1-e as components. Each node 1 has a node id which is unique identification information generated by a hash value or the like.

  The overlay network is divided into a plurality of clusters 2 (2-X, 2-Y, 2-Z). That is, each node 1 participating in the overlay network belongs to one of the clusters 2. Each cluster 2 has a cluster id which is unique identification information. In the present embodiment, this cluster 2 is used as a routing unit. Each cluster 2 has a virtual logical link 3 (3-XY, 3-YZ) connected to the adjacent cluster 2.

  FIG. 2 shows the configuration of each node 1 participating in the overlay network of FIG. Each node 1 autonomously determines a network connection unit 10 for connecting to the overlay network, a storage unit 11 for storing information, and a cluster 2 to which the node 1 belongs on the overlay network, and a logical network configuration A link information measurement unit 13 that periodically measures distance information between a cluster determination unit 12 that creates information and an arbitrary node 1 in the cluster 2 adjacent to the cluster 2 and uses the measured information as link information. A link information transmission / reception unit 14 that transmits / receives link information to / from another node 1, a routing table creation unit 15 that creates topology information based on the link information, and creates a routing table based on the topology information, routing And a route control unit 16 that performs packet transfer according to the table.

Next, the route control operation of the present embodiment will be described. FIG. 3 is a flowchart showing the path control operation of each node 1.
First, each node 1 acquires and holds the following logical network configuration information 110 (step S100). In the example of FIG. 1, there are three clusters 2-X, 2-Y, 2-Z, and logical links 3-XY between the clusters 2-X and 2-Y and between the clusters 2-Y and 2-Z, respectively. , 3-YZ exists.

In the case of the example in FIG. 1, when attention is paid to the node 1 belonging to the cluster 2-X, the logical network configuration information 110 held by the node 1 is as follows.
(Affiliation cluster id, (neighboring cluster list))
= (Affiliation cluster id, (adjacent cluster id-1, adjacent cluster id-2, ...))
= (Id-X, (id-Y))

  id-X and id-Y are the cluster ids of clusters 2-X and 2-Y, respectively. In an overlay network, the network structure is variable. For example, each node 1 may autonomously review its cluster when link information is extremely deteriorated (Ref. 1 “Kamei Kaoru, Kawahara Ryoichi,“ Proposal of Autonomous Clustering Method for Large-Scale Network ”, Proceedings of the IEICE General Conference, Vol. 2007 Communication, No. 2 (200707307), p. 110, 2007/03 ”). In this way, the cluster determining unit 12 of each node 1 performs the process of autonomously determining the belonging cluster.

  According to Document 1 above, each cluster 2 has a representative node 1-r. The cluster determination unit 12 of each representative node 1-r periodically performs full mesh measurement between the representative nodes. The measurement target includes network quality (for example, delay time). The cluster determination unit 12 of each representative node 1-r holds the measurement result vector.

  When the new node 1 participates in the overlay network, the cluster determination unit 12 of the new node 1 searches for the representative node 1-r for each cluster, measures the network quality with all the representative nodes 1-r, and measures Holds the result vector. The cluster determination unit 12 of the new node 1 compares the measurement result vector held by itself with the measurement result vector held by each representative node 1, and itself belongs to the cluster 2 to which the representative node 1-r having the closest inter-vector distance belongs. To belong. In this way, when the belonging cluster is determined, the cluster determination unit 12 of the new node 1 creates the logical network configuration information 110 as described above and stores it in the storage unit 11.

  Further, each node 1 participating in the overlay network periodically measures the network quality in the same manner as the new node, and determines its own cluster. The cluster determination unit 12 of each node 1 updates the logical network configuration information 110 by determining the belonging cluster.

  In order to determine the representative node 1-r, first, the initial representative node 1-r is randomly arranged, and the affiliation cluster is determined for each node as described above. Let the node be the new representative node 1-r. For the calculation of the center of gravity, the distance between vectors is used, and the node 1 at which the sum of the distances between vectors in each cluster 2 is the minimum is set as the center of gravity node. Then, the cluster determination unit 12 of each node 1 repeats the determination of the representative node 1-r and the determination of its own cluster.

  As described above, each node 1 can determine the belonging cluster 2 and create / update the logical network configuration information 110. This logical network configuration information 110 corresponds to conventional physical network information.

Next, the link information measuring unit 13 of each node 1 measures the link information 111 based on the logical network configuration information 110, and stores the measured link information 111 in the storage unit 11 (step S101). Specifically, the link information measuring unit 13 of each node 1 periodically measures distance information (delay time etc.) with an arbitrary node in a cluster adjacent to the cluster to which the node belongs. Furthermore, the link information measurement unit 13 also acquires the measurement time at this time. The format of the link information 111 is as follows.
(Measurement source node id, measurement source cluster id, measurement destination node id, measurement destination cluster id, distance information, measurement time)

When measuring the link information 111, a plurality of pieces of link information 111 may be acquired in the links between the same clusters, for example, by randomly selecting a partner.
As shown in the example of FIG. 4, when the measurement is performed from the node 1-a belonging to the cluster 2-Y to the node 1-b belonging to the cluster 2-X, the link information 111 is as follows.
(Id-a, id-Y, id-b, id-X, distance information, measurement time)

  id-a and id-b are node ids of the nodes 1-a and 1-b, respectively. The link information 111 corresponds to the link information acquired in the conventional step S1.

  Subsequently, the link information transmitting / receiving unit 14 of each node 1 transmits the link information 111 to the other node 1 and receives the link information 111 from the other node 1 (step S102). The link information transmission / reception unit 14 stores the information received from the other nodes 1 in the storage unit 11 as link information 112.

Regarding transmission of link information, it is not practical to propagate all link information over a wide area (broadcast) as in the past, so a method of propagating link information at random to some extent and link information that is randomly propagated are overlaid. It is realistic to combine with a method of actively acquiring missing link information when the entire network cannot be covered. Such processing can be found in Winny, which is a P2P (Peer to Peer) search system (see Reference 2, “winny.info key distribution”, http://winny.info/key.html). ). According to Document 2, each node 1 transmits information to the adjacent node 1, and acquires information from the entire network by sending a search request for the missing information.
The processing in step S102 described above corresponds to the conventional step S2.

  Next, the routing table creation unit 15 of each node 1 creates the routing table 113 based on the link information 111 and 112 (step S103). Specifically, the routing table creation unit 15 is based on the link information 111 and 112, and topology information (route information) representing a route between any two nodes and a route cost (delay in the present embodiment). And the routing table 113 is created based on the Dijkstra algorithm or the like.

  In the present embodiment, since the link information 111 and 112 are not completely flooded, an information acquisition process is required. Specifically, as described in step S102 above, topology information is created based on link information randomly propagated based on a certain probability and link information collected from the entire overlay network by a search request, and routing is performed. A table 113 is created.

  When creating the routing table 113, care must be taken when creating the topology information because the reliability and completeness of the link information 111 are lower than in the conventional case. That is, when creating the topology information, the routing table creation unit 15 has the same measurement source node id, the same measurement source cluster id, the same measurement destination node id, and the same measurement destination cluster id. Among the plurality of link information 111, the one with the latest measurement time is used.

  The routing table creation unit 15 may create topology information by performing some statistical processing on the link information 111. As statistical processing, for example, distance information is extracted from a plurality of link information 111 having the same measurement source node id, the same measurement source cluster id, the same measurement destination node id, and the same measurement destination cluster id. Thus, it is conceivable to obtain an average value of distance information. Further, a weighted average of the distance information included in the link information 111 may be obtained according to the proximity of the measurement time included in the plurality of link information 111 to the current time. In this case, the closer the current time is, the heavier the weight is.

In the created topology information, there may be a plurality of paths between any two nodes. Therefore, it is necessary to evaluate the cost of the link on the route by some method, calculate the optimum route, and create the routing table 113. In the present embodiment, for example, the Dijkstra algorithm is used as the optimum route calculation method. The Dijkstra algorithm is a route calculation method for selecting the route having the smallest or largest total value by summing the costs of the links on the route (delay in this embodiment). In this way, the routing table creation unit 15 creates the routing table 113 that represents the optimum route between any two nodes.
The processing in step S103 described above corresponds to the conventional step S3.

  Next, the path control unit 16 of each node 1 performs forwarding (packet transfer) according to the created routing table 113 (step S104). This forwarding requires caution because the reliability of the node 1 is low. Specifically, the path control unit 16 transfers the packet to the destination node 1 according to the routing table 113, and at the same time makes the node 1 belonging to the arbitrary cluster 2 equal to the destination node 1, and this arbitrary cluster The same packet may be transferred to the node 1 belonging to 2.

The path control unit 16 of the node 1 that has received the packet stores the packet in the storage unit 11 if the packet is addressed to itself, and transfers the packet according to the routing table 113 if the packet is not addressed to itself. Thus, the reliability of forwarding can be improved.
The processing in step S104 described above corresponds to the conventional step S4.

  As described above, in this embodiment, nodes on the overlay network are clustered, and autonomous distributed routing is performed in units of clusters, thereby enabling link state-based routing in a network composed of a large number of nodes. Infrastructure can be constructed, and various quality controls and efficient use of equipment can be realized.

  The node 1 described in the present embodiment can be realized by a computer having a CPU, a storage device, and an interface, and a program for controlling these hardware resources. The CPU of each node 1 executes the processing described in this embodiment in accordance with a program stored in the storage device.

  The present invention can be applied to a super multi-node environment such as a wide area network such as the Internet or a sensor network.

  DESCRIPTION OF SYMBOLS 1 ... Node, 2 ... Cluster, 3 ... Logical link, 10 ... Network connection part, 11 ... Memory | storage part, 12 ... Cluster determination part, 13 ... Link information measurement part, 14 ... Link information transmission / reception part, 15 ... Routing table preparation part , 16 ... path control unit, 110 ... logical network configuration information, 111, 112 ... link information, 113 ... routing table.

Claims (10)

A cluster determination procedure for autonomously determining a cluster to which each node device belongs on the overlay network, and creating logical network configuration information describing the belonging cluster and adjacent clusters,
A link information measurement procedure in which each node device periodically measures distance information with an arbitrary node device in a cluster adjacent to a cluster to which the node device belongs based on the logical network configuration information, and uses the measured information as link information; ,
A link information transmission / reception procedure in which each node device transmits / receives the link information to / from another node device;
A routing table creation procedure in which each node device creates topology information based on link information measured by its own node device and link information received from another node device, and creates a routing table based on this topology information;
A route control method comprising: a route control procedure in which each node device performs packet transfer according to the routing table. The route control method according to claim 1,
The route determination method characterized in that the cluster determination procedure includes a procedure of periodically reviewing a cluster to which the own node device belongs based on network quality and updating the logical network configuration information. The route control method according to claim 1,
The link information transmission / reception procedure includes a procedure of acquiring the missing link information by receiving the link information transmitted from another node device and sending a search request to the other node device. A characteristic route control method. The route control method according to claim 1,
In the routing table creation procedure, the topology information is created by adopting the information having the latest measurement time among the plurality of received link information, or the topology information is created by statistically processing the received plurality of link information. A path control method comprising a step of: The route control method according to claim 1,
The routing control procedure transfers a packet to a destination node device according to the routing table, and at the same time, makes a node device belonging to an arbitrary cluster equivalent to the transfer destination node device, and is addressed to the node device belonging to the arbitrary cluster. The route control method further includes a step of transferring the packet. Cluster determination means for autonomously determining a cluster to which the own node device belongs on the overlay network, and creating logical network configuration information describing the belonging cluster and an adjacent cluster;
Link information measuring means for periodically measuring distance information with any node device in the cluster adjacent to the cluster based on the logical network configuration information, and using the measured information as link information;
Link information transmitting / receiving means for transmitting / receiving the link information to / from another node device;
A routing table creating means for creating topology information based on link information measured by the own node device and link information received from another node device, and creating a routing table based on the topology information;
A node device comprising: path control means for performing packet transfer according to the routing table. The node device according to claim 6, wherein
The cluster determining means periodically reviews the cluster to which the node device belongs based on network quality, and updates the logical network configuration information. The node device according to claim 6, wherein
The link information transmission / reception means receives the link information transmitted from another node device and obtains missing link information by sending a search request to the other node device. Node device. The node device according to claim 6, wherein
The routing table creation means creates the topology information by adopting information having the latest measurement time among a plurality of received link information, or creates the topology information by statistically processing the received plurality of link information A node device characterized by: The node device according to claim 6, wherein
The route control means transfers the packet to the transfer destination node device according to the routing table, and at the same time, makes the node device belonging to an arbitrary cluster equivalent to the transfer destination node device, and addresses the node device belonging to the optional cluster. A node device for forwarding the packet to the network.

Images