JP2008288872A - Peer-to-peer communication establishment apparatus, method, and program - Google Patents

Abstract

An object of the present invention is to optimize traffic by performing direct communication between clients without going through a virtual hub.
The present invention relates to a virtual network in which a plurality of clients having a communication link with at least one of a plurality of relay nodes communicate with each other via the relay node. An apparatus for establishing a peer-to-peer link p between specific clients 30 (# 2, # 6), and two consecutive clients among any two clients 30 (# 2, # 6) among a plurality of clients A global flow that is a communication flow that is made through the above relay nodes 20 (# 1 to # 4) and satisfies a predetermined standard is detected, and a peer-to-peer link p is established between clients in which the global flow is detected. A network management module 40 is provided.
[Selection] Figure 9

Description

  The present invention relates to a peer-to-peer communication establishment apparatus, method and program, and more particularly to an apparatus, method and program for establishing peer-to-peer communication based on global flow detection in a virtual network.

  For example, a virtual network constructed by software on an existing physical network (hereinafter referred to as “underlay network”) such as a TCP / IP network is referred to as an overlay network.

  As main overlay networks (hereinafter also referred to as “virtual networks”), what is called a peer-to-peer (P2P) network (for example, Non-Patent Documents 1 to 7) and virtual local area networks (hereinafter “ Referred to as “virtual LAN”).

  Peer-to-peer networks are currently widely used for file exchange (Winny, BitTorrent, etc.), voice communication (Skype, etc.) and the like. Depending on the method, the following two major steps are taken: (1) Host and content search depend on special servers, but final data transmission / reception is performed between peer-to-peer hosts, and (2) data In addition to transmission / reception, the host and content search itself can be divided into those that do not require a special server and are distributed and coordinated among hosts in completely equal positions. In some cases, the above two intermediate methods using a host group that plays a special role are used for searching hosts and contents, although they are not specific servers.

  On the other hand, a typical virtual LAN includes Softether (Packetix) (for example, Non-Patent Documents 8 to 9) manufactured by SoftEther Corporation. Softeter (Packetix) is a software group for constructing virtual Ethernet (registered trademark), which is an example of a virtual network. For example, a virtual Ethernet construction method using Softeter (Packetix) or similar software is as follows.

  That is, first, software that emulates an Ethernet switching hub (hereinafter referred to as “virtual hub”) is operated on a certain computer (hereinafter, the computer that operates the virtual hub is referred to as “server”).

  Next, a computer wishing to participate in virtual Ethernet connects to a virtual hub (hereinafter referred to as a virtual network interface) via a virtual network interface (hereinafter referred to as “virtual network interface”) constructed by software. The computer that operates is called “client”).

  A general configuration of such a virtual Ethernet is shown in FIG.

  For example, the underlay network 150 that is a TCP / IP network is a physical network that serves as a basis for constructing the virtual Ethernet 154 that is an overlay network. The underlay network 150 includes a firewall 156, a router 157, and a link layer communication device (a hub or a switching hub in the case of Ethernet). The router 157 is a network layer device that constitutes a TCP / IP network, and is a general device, and therefore, detailed description thereof is omitted here. The firewall 156 is a security device installed at the boundary between the internal network 160 and the external network 161, and since it is a general device, its detailed description is omitted here.

The client 153 is a computer that is connected to the virtual hub 151 that constitutes the virtual Ethernet 154, and is connected to the virtual hub 151 via the virtual network interface 152. In this way, the virtual hub 151 functions as a relay node that holds connections with a plurality of clients 153. The server 155 is a network virtualization apparatus in which such a virtual hub 151 is arranged. In this specification, the server 155 that is such a network virtualization apparatus and the client 153 that is a terminal controlled by the network virtualization apparatus are collectively referred to as a network virtualization system. The virtual Ethernet 154 is constructed by connecting a plurality of clients 153 each having a virtual network interface 152 to one or a plurality of virtual hubs 151.
The Institute of Electronics, Information and Communication Engineers, September 2004, “P2P General I: Challenge of Brokerless Model” The Institute of Electronics, Information and Communication Engineers, October 2004, "P2P General Remarks II P2P Technology" IEICE Journal, December 2004, "P2P General Review III P2P Service and Business" IEICE Journal, January 2005, "P2P General Remarks IV Latest Trends and Future Prospects" ASCII, Inc., UNIX (registered trademark) MAGAZINE, September 2005, "Basic knowledge of P2P technology [1]" ASCII Co., Ltd., UNIX MAGAZINE, October 2005, “Basic knowledge of P2P technology [2]” ASCII Co., Ltd., UNIX MAGAZINE, November 2005, "Basic knowledge of P2P technology [3]" ASCII Corporation, Official SoftEther Usage Guide http: // www. softther. com, SoftEther Corporation website

In such a conventional virtual Ethernet 154, all traffic between the clients 153 belonging to the virtual Ethernet 154 passes through the virtual hub 151, that is, the server 155, and between the clients 153 belonging to the virtual Ethernet 154, or between the client 153 and the server. The positional relationship with 155 does not take into account the topology of the underlay network 150. Therefore,
1. It is difficult to ensure the scalability of the virtual hub 151.
2. The load on the underlay network 150 increases, and the capital investment cost increases.
3. Management costs such as installation of a plurality of virtual hubs 151 and switching of clients 153 between the plurality of virtual hubs 151 increase.
The problem arises.

  The present invention has been made in view of such circumstances, and a peer-to-peer communication establishment device capable of optimizing traffic by performing direct communication between clients without going through a virtual hub, It is an object to provide a method and a program. In particular, a peer-to-peer communication establishment device, method for optimizing traffic by communicating directly between clients communicating via a plurality of virtual hubs without going through those virtual hub groups, And to provide a program.

  In order to achieve the above object, the present invention takes the following measures.

  In other words, the invention of claim 1 is a specific client in a virtual network in which a plurality of clients having a communication link with at least one of a plurality of relay nodes communicate with each other via a relay node. An apparatus for establishing a peer-to-peer link between two or more clients among a plurality of clients via a continuous two or more relay nodes and satisfying a predetermined standard Global flow detection means for detecting a certain global flow, and peer-to-peer link establishment means for establishing a peer-to-peer link between clients whose global flows have been detected by the global flow detection means.

  Then, communication between clients in which the peer-to-peer link has been established by the peer-to-peer link establishment means is performed via the peer-to-peer link.

  According to a second aspect of the present invention, in the peer-to-peer communication establishment apparatus according to the first aspect, the global flow detecting means is the number of hops reached for each of the leaf nodes that are relay nodes that do not relay the communication flow between the relay nodes for all of the relay nodes. And the root node determining means for determining the root node that is the relay node having the smallest maximum hop count or all the relay nodes having the maximum hop count smaller than a predetermined threshold, and the root node determining means. The two roots are detected by detecting a global flow candidate that is a communication flow satisfying a predetermined standard while tracing the relay node from the root node to the leaf node side along the virtual network using the root node as a starting point. If continuous global flow candidates are detected between nodes, The Barufuro candidate and a global flow determining means for determining a global flow.

  According to a third aspect of the present invention, in the peer-to-peer communication establishment apparatus according to the first aspect, the global flow detecting means is a relay node having a maximum load for all of the relay nodes or all relays having a load greater than a predetermined threshold. A root node determining unit that determines a root node that is a node, and a relay node that does not relay a communication flow between relay nodes from the root node along the virtual network, starting from the root node determined by the root node determining unit. If a global flow candidate that is a communication flow that satisfies a predetermined standard is detected while tracing a relay node to a certain leaf node side, and if a continuous global flow candidate between two leaf nodes is detected, then Global flow size that determines global flow candidates as global flows And a means.

  According to a fourth aspect of the present invention, the root node determining means sets the target relay nodes for the determination of the root node not to all of the relay nodes but to all of the leaf nodes. Is a peer-to-peer communication establishment apparatus.

  According to the invention of claim 5, the global flow detection means starts the relay nodes from the relay nodes to the leaf node side that is a relay node that does not relay the communication flow between the relay nodes along the virtual network. The global flow candidate that is a communication flow that satisfies a predetermined standard is detected while tracing, and if a continuous global flow candidate is detected between two leaf nodes, the continuous global flow candidate is detected as a global flow. The peer-to-peer communication establishment apparatus according to claim 1.

  In the invention of claim 6, the global flow detection means is the same regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and in all relay nodes between the two leaf nodes. The peer-to-peer communication establishment apparatus according to claim 1, further comprising a global flow determination unit that determines a communication flow between the two leaf nodes as a global flow when a communication flow that satisfies the condition is detected.

  In the invention according to claim 7, the global flow detection means is the same regarding address information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and in all relay nodes between the two leaf nodes. The peer-to-peer communication establishment apparatus according to claim 1, further comprising a global flow determination unit that determines a communication flow between the two leaf nodes as a global flow when a communication flow that satisfies the condition is detected.

  In the invention of claim 8, the global flow detection means is the same regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and in all relay nodes between the two leaf nodes. A global flow determination means is provided for determining a communication flow between the two leaf nodes as a global flow when a communication flow that satisfies the conditions and satisfies the same condition regarding address information is detected. 1 is a peer-to-peer communication establishment apparatus.

  In the virtual network in which a plurality of clients having a communication link with at least one of the plurality of relay nodes respectively communicate with other clients via the relay node, the invention of claim 9 is provided between specific relay nodes. A peer-to-peer link establishing a peer-to-peer link between two relay nodes among a plurality of relay nodes, and a communication flow that satisfies a predetermined standard between at least one other relay node Global flow detection means for detecting a global flow, and peer-to-peer link establishment means for establishing a peer-to-peer link between two relay nodes in which a global flow has been detected by the global flow detection means.

  Then, communication between the relay nodes in which the peer-to-peer link is established by the peer-to-peer link establishment means is performed via the peer-to-peer link.

  The invention of claim 10 is a method corresponding to the apparatus of claim 1. The inventions of claims 11 to 19 are programs corresponding to the devices of claims 1 to 9.

  According to the present invention, it is possible to realize a peer-to-peer communication establishment apparatus, method, and program capable of optimizing traffic by directly communicating between clients without going through a virtual hub. .

  In particular, a peer-to-peer communication establishment device, method for optimizing traffic by communicating directly between clients communicating via a plurality of virtual hubs without going through those virtual hub groups, And a program can be realized.

  The best mode for carrying out the present invention will be described below with reference to the drawings.

(First embodiment)
A first embodiment of the present invention will be described.

  FIG. 1 is a conceptual diagram showing a configuration example of a typical network 10 to which the peer-to-peer communication establishment method according to the first embodiment of the present invention is applied. The network 10 may be a LAN such as Ethernet (registered trademark) or a WAN to which a plurality of LANs are connected via a public line or a dedicated line. In the case of a LAN, it is composed of a number of subnets via routers as necessary. In the case of WAN, a firewall or the like for connecting to a public line is provided as appropriate. However, in order not to obscure the essence of the present invention, illustration and detailed description thereof are omitted, and only the virtual hub 20, the client 30, the network management module 40, and the LAN 50 are shown.

  In the network configuration shown in FIG. 1, the client 30 (# 1) is connected to the virtual hub 20 (# 2), the client 30 (# 2), the client 30 (# 3) is connected to the virtual hub 20 (# 1), and the client 30 (# 2). The clients 30 (# 6) to # 4) are connected to the virtual hub 20 (# 4), respectively. The virtual hubs 20 (# 1) to (# 4) are connected in series as shown in the figure. Note that the LANs 50 (# 1) to (# 6) represent segments of the underlay network (usually segments of the IP network), and the segments are usually connected by a router (not shown). The virtual hub 20 (# 1) and the clients 30 (# 1) to (# 3) are in the LAN 50 (# 1), the virtual hub 20 (# 2) is in the LAN 50 (# 2), and the virtual hub 20 (# 3) Belongs to the LAN 50 (# 3), the virtual hub 20 (# 4) belongs to the LAN 50 (# 4), and the clients 30 (# 4) to (# 6) belong to the LAN 50 (# 5).

  Further, the virtual hubs 20 (# 1) to (# 4) are connected to the network management module 40. The network management module 40 can belong to any of the LANs 50 (# 1) to (# 6). Alternatively, it can be operated on a computer such as the virtual hub 20 (# 1) to (# 4) or the client 30, or can be operated on another computer. In FIG. 1, as an example, it is shown as belonging to any one of the LANs 50 (# 1) to (# 6).

  2, 3, and 4 are functional block diagrams illustrating configuration examples of the virtual hub 20, the client 30, and the network management module 40.

  As shown in FIG. 2, the virtual hub 20 as shown in the virtual hub 20 (# 1) to (# 4) in FIG. 1 includes an Ethernet switching hub 21, a forwarding table 22, a connection link control unit 23, a virtual port 24, A network management interface unit 25, a peer-to-peer link control unit 26, an underlay network protocol group 27, and a physical network interface 28 are provided.

  The Ethernet switching hub 21 performs the same processing as a so-called general Ethernet switching hub.

  The forwarding table 22 is a table that holds mapping between the destination MAC address of the Ethernet frame and the output destination port. Since it is generally used for Ethernet frame transfer processing by an Ethernet switching hub, details are omitted here.

  The connection link control unit 23 is a part that holds and controls a connection (link) in the underlay network with the client 30 or another virtual hub 20, and the Ethernet frame received from the Ethernet switching hub 21 is an appropriate underlay network. To the link. Further, the Ethernet frame received from the underlay network is transferred to the Ethernet switching hub 21.

  The virtual port 24 is a concept corresponding to a communication port of a normal Ethernet switching hub in the virtual hub 20, and is a logical cable, that is, another virtual hub 20 using some means of an underlay network, or a virtual network interface 32. (It will be described later with reference to FIG. 3). Accordingly, the virtual port 24 represents an end point of connection (link) in the underlay network with the other virtual hub 20 or the client 30 (particularly, the virtual network interface 32) accommodated by the own station in the connection link control unit 23. They are distinguished by a unique identifier within the virtual hub 20.

  The network management interface unit 25 is connected to the other virtual hub 20 or the client 30 (particularly, the virtual network interface 32) accommodated by the own station, the traffic amount between the connection links in the Ethernet switching hub 21, and the traffic Protocol, the positional relationship information of the client 30 in the underlay network, etc. are collected and managed, and further the information is notified to the network management module 40.

  The peer-to-peer link control unit 26 instructs the client 30 to generate and delete a peer-to-peer link in response to an instruction from the network management module 40 via the network management interface unit 25.

  The underlay network protocol group 27 is generally TCP / IP and higher-level protocols (for example, HTTP) that use them.

  The physical network interface 28 is a physical network interface used for actual communication.

  As shown in FIG. 3, the clients 30 (# 1) to (# 6) shown in FIG. 1 include an application / upper protocol group 31, a virtual network interface 32, an underlay network protocol group 37, and a physical A network interface 38 is provided. Further, the virtual network interface 32 includes an Ethernet frame transmission / reception unit 33, a transmission destination management table 34, and a connection link control unit 35. Further, the connection link control unit 35 includes a plurality of virtual ports 36.

  The application / upper protocol group 31 is a user application that uses the virtual network interface 32 and upper protocols (for example, TCP / IP, HTTP, etc.).

  In the virtual network interface 32, the Ethernet frame transmission / reception unit 33 stores the data from the application / upper protocol group 31 in the Ethernet frame and sends it to the connection link control unit 35. In addition, the Ethernet header is removed from the Ethernet frame received from the connection link control unit 35 and passed to the application / upper protocol group 31.

  In the virtual network interface 32, the connection link control unit 35 is a part that holds and controls the connection (link) in the underlay network with the virtual hub 20 or another client 30, and is received from the Ethernet frame transmission / reception unit 33. Send out the Ethernet frame to the appropriate underlay network link. In addition, the Ethernet frame received from the underlay network is transferred to the Ethernet frame transmitting / receiving unit 33.

  The virtual port 36 provided in the connection link control unit 35 represents an end point of connection (link) in the underlay network with the other virtual hub 20 in the connection link control unit 35, and is represented by a unique identifier in the virtual network interface 32. Differentiated.

  The transmission destination management table 34 is a table that holds a mapping between a peer-to-peer link and the MAC address of the peer of the link. As will be described later, the Ethernet frame (unicast frame) addressed to the peer registered in the transmission destination management table 34 is transferred to the corresponding peer-to-peer link, and the other destination Ethernet frame (unicast frame), broadcast frame, and multicast frame are It is output to the link with the virtual hub 20.

  The underlay network protocol group 37 is generally TCP / IP and a higher level protocol (for example, HTTP) using them.

  The physical network interface 38 is a physical network interface used for actual communication.

  The network management module 40 as shown in FIG. 1 executes a program recorded on a recording medium such as a magnetic disk for executing the peer-to-peer communication establishment method of the present invention, or a program downloaded via a communication network such as the Internet. It is a device realized by a computer that is read and whose operation is controlled by this program.

  This program can be executed by a computer (computer) such as the network management module 40. For example, a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), optical disk (CD-ROM, DVD or the like), stored in a recording medium such as a semiconductor memory, or transmitted and distributed via a communication medium such as the Internet. The program stored in the recording medium is configured in the computer such as the network management module 40 with software means (including not only the execution program but also the table and data structure) that is executed by the computer such as the network management module 40. Includes setting program. Then, when this program is read from a recording medium or a communication medium into a computer such as the network management module 40, the information collection unit 41, the root node determination unit, and the network management module 40 are operated. 42, the global flow detection unit 43, and the peer-to-peer link control unit 44 execute processing.

  The information collection unit 41 communicates with the network management interface unit 25 of each virtual hub 20, and from each virtual hub 20 to the other virtual hub 20 and the client 30 (particularly, the virtual network interface 32) accommodated by the virtual hub 20. The connection link state of the network, the traffic volume transferred between the connection links in the Ethernet switching hub 21, the traffic protocol, the positional relationship information of the client 30 in the underlay network, etc. are periodically collected and managed. Further, a connection topology state (hereinafter, connection tree information) of the virtual hub 20 and the virtual network interface 32 is created from the above information.

  Here, for example, as shown in FIG. 5, it is assumed that traffic T that may match a certain condition occurs between the client 30 (# 2) and the client 30 (# 6). Then, the information collection unit 41 collects information as described above from the network management interface unit 25 of each virtual hub 20 (# 1) to (# 4). Further, among the collected information, from the connection link information held by each virtual hub 20 (# 1) to (# 4), as shown in FIG. 6, each virtual hub 20 (# 1) to (# 4), And the connection state of the client 30 (virtual network interface 32) accommodated by them, that is, connection tree information (tree data structure) is created.

  In FIG. 6, for convenience of illustration, the virtual hub 20 (# 3) is shown as the root of the tree, but it means that the root node in the subsequent processing is the virtual hub 20 (# 3). I don't mean.

  The root node determination unit 42 determines the root node from the connection tree information as shown in FIG. The root node refers to the virtual hub 20 where the global flow detection process is started. A global flow is a local flow that satisfies a certain condition such as exceeding the traffic volume reference value or storing the same upper layer protocol information across two or more consecutive virtual hubs 20. In this case, these continuous local flows are regarded as one global flow and are called global flows.

  On the other hand, the local flow means one of the following two definitions in this specification. In the first definition, it is assumed that a traffic is generated between two virtual ports 24 in a certain virtual hub 20 (a state where forwarding is performed), and there is a flow between these virtual ports 24. This is called local flow.

In the second definition, an arbitrary two MAC addresses learned in the forwarding table 22 and a pair of output ports mapped to them in a certain virtual hub 20 are referred to as a local flow. In this case, the mapping is expressed as follows.
{Virtual port m: MAC1} => {virtual port n: MAC2}
Here, m and n are integers for identifying virtual ports, MAC1 and MAC2 are MAC addresses mapped to the virtual ports, and “⇒” represents a direction (details will be described later).

  A specific process in which such a root node determination unit 42 determines a root node will be described later.

  The global flow detection unit 43 detects a global flow using the root node determined by the root node determination unit 42 as a start node. Details of this detection method will be described later. Note that the global flow detection unit 43 can also perform the global flow detection process using all the virtual hubs 20 as root nodes without using the specific virtual hub 20 as a root node.

  When a global flow is detected by the global flow detection unit 43, the peer to peer link control unit 44 instructs the client 30 to generate a peer to peer link via the network management interface unit 25 of the virtual hub 20. For example, as shown in FIG. 5, the global flow detection unit 43 is between the virtual hub 20 (# 1) and the virtual hub 20 (# 4), that is, the client 30 (# 2) and the client 30 (# 6). When a global flow is detected, the peer-to-peer link control unit 44 instructs the client 30 (# 2) and the client 30 (# 6) to create a peer-to-peer link between them and shift to peer-to-peer communication. put out.

  On the other hand, when the global flow detection unit 43 does not detect the global flow, the peer-to-peer link control unit 44 does not perform any further processing, and the client 30 communicates via the virtual hub 20 as before. .

  As shown in the flowchart of FIG. 7, an instruction to generate a peer-to-peer link is issued from the peer-to-peer link control unit 44 to the related clients 30 (for example, the client 30 (# 2) and the client 30 (# 6)). (S1) and the connection link control unit 35 of the client 30 that has received the instruction generates a communication path using a peer-to-peer link by using some means of the underlay network (for example, a TCP connection, an SSL session, etc.) ( S2), the client 30 (# 2) registers the virtual Ethernet MAC address of the client 30 (# 6), and the client 30 (# 6) registers the same MAC address of the client 30 (# 2) in the transmission destination management table 34. (S3).

  Thereafter, as shown in the flowchart of FIG. 8, the connection link control unit 35 of the client 30 (# 2) receives the transmission data (S11), and the Ethernet frame (destination MAC address) for the client 30 (# 6) The client 30 (with the MAC address of # 6) is sent to the peer-to-peer link (S12: No, S13: Yes, S14), and the Ethernet frame having the other destination MAC address (unicast) is a virtual hub ( # 1) (S12: No, S13: No, S15). Broadcast frames and multicast frames are also sent to the link with the virtual hub 20 (# 1) (S12: Yes, S15).

  Similarly, the client 30 (# 6) sends an Ethernet frame for the client 30 (# 2) (having the MAC address of the client 30 (# 2) as the destination MAC address) to the peer-to-peer link, and otherwise The Ethernet frame having the destination MAC address, the broadcast frame, and the multicast frame are transmitted to the link with the virtual hub 20 (# 4).

  As a result, traffic between the client 30 (# 2) and the client 30 (# 6) does not pass from the virtual hub 20 (# 1) to the virtual hub 20 (# 4), and these virtual hubs 20 (# The load on the groups 1) to (# 4) and the load on the LAN 50 (# 2), LAN 50 (# 3), and LAN 50 (# 4), which are underlay networks, are reduced.

  Finally, FIG. 9 shows a state where peer-to-peer communication between the client 30 (# 2) and the client 30 (# 6) is established by the above processing. A solid line p indicates the peer-to-peer communication link p, and a broken line d indicates the communication link d via the virtual hub 20.

  When the detected global flow does not satisfy a certain condition, that is, when it is no longer a global flow, the global flow detection unit 43 detects it and the peer-to-peer link control unit 44 detects the associated client 30. , Instructing the deletion of the peer-to-peer link and the transition to communication via the virtual hub 20.

  In this case, as shown in the flowchart of FIG. 10, the processing on the client 30 side is triggered by the reception of an instruction from the network management module (S21), the peer-to-peer link deletion (S22), and the corresponding from the transmission destination management table 34. The client MAC address is deleted (S23).

  As described above, the network management module 40 can be operated on the virtual hub 20 or the client 30 on which the virtual network interface 32 operates, or can be operated on another computer. Communication between the information collection unit 41, the peer-to-peer link control unit 44 of the network management module 40, the network management interface unit 25 of each virtual hub 20, and the peer-to-peer link control unit 26 is a general one using an underlay network. Since it is realizable, the detailed description is abbreviate | omitted here.

  Next, three examples of the root node determination method performed by the root node determination unit 42 will be described.

(Root node determination method 1)
Here, the virtual hub at the center of the tree on the topology is the root node.

  The root node determination unit 42 of the network management module 40 investigates the number of hops reaching all the leaf nodes for all the virtual hubs 20, and sets the virtual hub having the smallest maximum number of hops as the root node. A leaf node is a virtual hub 20 that has zero or one connection to another virtual hub 20, in other words, a virtual hub 20 that does not relay traffic between other virtual hubs 20.

  As an example, with reference to FIG. 11, for the virtual hub 20 (# 1) to the virtual hub 20 (# 7), each leaf node (virtual hub 20 ( The number of hops to # 1), (# 4), (# 6), (# 7)) is as shown in FIG. 12, and as a result, the virtual hub 20 (# 3) becomes the root node. Depending on the topology, it is conceivable that a plurality of virtual hubs 20 satisfy the conditions of the root node. In this case, separately, the root node is set by at least one of the following a) to c). decide.

a) A plurality of root nodes satisfying the above conditions are all set as root nodes.
b) Choose one at random.
c) Consider the load (it can be said to be a combination with the root node determination method 2 described later).

  Furthermore, all virtual hubs 20 whose maximum hop count is below a predetermined threshold can be set as the root node. Further, since a method for counting the number of hops between nodes in a data structure forming a tree is a general method, detailed description thereof is omitted here.

(Root node determination method 2)
Here, the virtual hub 20 having a high load is set as the root node.

  The root node determination unit 42 of the network management module 40 has, for all virtual hubs 20, the number of accommodated nodes, CPU usage rate, memory usage rate, local flow traffic volume, local flow count, local flow average load (total load / local load). The load information such as the number of flows) and the traffic amount of a specific upper layer protocol having a local flow is examined, and the virtual hub 20 with the highest load is set as the root node.

  Furthermore, all the virtual hubs 20 in which the load exceeds a predetermined threshold can be set as the root node. Here, for example, in FIG. 13, it is assumed that the traffic volume handled by each of the virtual hubs 20 (# 1) to (# 7) is a load, and the most loaded virtual hub 20 is determined as the root node. Here, a very high traffic t occurs between the client 30 (# 4) and the client 30 (# 6), and there is not much traffic between the other clients 30. If the virtual hub 20 (# 5) can be selected as the root node by this method, then in the global flow detection process performed in the global flow detection unit 43, for example, the global flow is selected from the one having a small number of hops to the leaf node. For example, when performing a detection process, a global flow between the virtual hub 20 (# 6) and the virtual hub 20 (# 7), that is, between the client 30 (# 4) and the client 30 (# 6) is quickly performed. It becomes possible to detect.

(Root node determination method 3)
In this method, the leaf node is set as the root node.

  The root node determination unit 42 of the network management module 40 investigates the load of the virtual port (the amount of traffic transmitted and received by the virtual port) for all leaf nodes, and determines the leaf node having the virtual port with the highest load as the root node. And Furthermore, it is possible to set all leaf nodes whose load exceeds a predetermined threshold as the root node.

  For example, in FIG. 14, the client 30 (# 1) is a file server, and the client 30 (# 10) is a computer that stores the backup, and between the client 30 (# 1) and the client 30 (# 10). Assume that a file system backup is in progress. In this state, for the virtual hubs 20 (# 1), (# 4), (# 6), and (# 7) that are leaf nodes, the traffic volume of all the virtual ports held by them is investigated. For example, if the traffic volume of the virtual port 3 of the virtual hub 20 (# 4) is maximized and the virtual hub 20 (# 4) can be determined as the root node, the global flow performed in the global flow detection unit 43 thereafter. In the detection process, between the virtual hub 20 (# 2) and the virtual hub 20 (# 4) that handle the file system backup traffic, that is, between the client 30 (# 1) and the client 30 (# 10). It becomes possible to detect the global flow. In this method, when the same load as the route node determination method 2 is adopted as the load, it can be considered that the target virtual hub 20 in the route node determination method 2 is limited to leaf nodes.

  Next, three examples of the global flow detection method performed by the global flow detection unit 43 will be described.

(Global flow detection method 1)
In this method, a global flow is detected using traffic information. Note that the local flow in this method is the local flow defined in Definition 1 above.

  The global flow detection unit 43 of the network management module 40 traces the virtual hub 20 in the direction of all the leaf nodes, with the root node determined by any one of the route node determination methods 1 to 3 as a process start node. However, the local flow including the virtual port in the root node direction of each virtual hub 20 is compared with the predetermined traffic condition, and if the condition is satisfied, the connection destination of the virtual port in the leaf node direction of the local flow The same processing is proceeded to the virtual hub 20, and if there is a continuous local flow between the two leaf nodes, it is determined as a global flow.

  For example, it is assumed that a local flow as shown in the figure exists in the virtual hub 20 (# 1) to the virtual hub 20 (# 7) in FIG. A local flow b indicated by a thick arrow b indicates traffic exceeding 1 Mbps (transferred between virtual ports of local flows), and a local flow h indicated by a thin arrow h indicates traffic less than 1 Mbps.

  Here, the global flow detection process when the traffic threshold is 1 Mbps in terms of the number of bits transferred between virtual ports and the root node is the virtual hub 20 (# 3) will be described.

  1) The virtual hub 20 (# 3) detects that the local flow by the virtual port 0-virtual port 1 pair exceeds the threshold.

  2) Next, the virtual port 0, which is the end point in the leaf node direction of the local flow, the virtual hub 20 (# 2) and the virtual hub 20 (# 5) to which the virtual port 1 is connected are respectively virtual nodes that are root nodes. The local flow including the virtual port 0 in the direction of the hub 20 (# 3) is checked, the local flow by the virtual port 0-virtual port 1 pair of the virtual hub 20 (# 2), and the virtual hub 20 (# 5) It detects that the local flow by the virtual port 0-virtual port 1 pair exceeds the threshold.

  3) Similarly, it is detected that the local flow by the virtual port 0-virtual port 2 pair of the virtual hub 20 (# 6) ahead of the virtual hub 20 (# 5) exceeds the threshold value.

  4) Through the processing of 1) to 3) above, between the two leaf nodes, the virtual hub 20 (# 2) and the virtual hub 20 (# 6), that is, the client 30 (# 1) and the client accommodated by them 30 (# 5) and a global flow is determined to exist.

  5) The processing is returned to the virtual hub 20 (# 3), and the local flow by the virtual port 0-virtual port 2 pair exceeds the threshold in the virtual hub 20 (# 3) in the same manner as described above. Is detected.

  6) Next, the local flow is checked for the virtual hub 20 (# 2) and the virtual hub 20 (# 4) that are the connection destinations of the virtual port 0 and the virtual port 2 of the virtual hub 20 (# 3). It is detected that the local flow by the virtual port 0-virtual port 1 pair of the hub 20 (# 2) and the local flow by the virtual port 0-virtual port 3 pair of the virtual hub 20 (# 4) exceed the threshold. To do.

  7) By the processing so far, between the two leaf nodes, the virtual hub 20 (# 2) and the virtual hub 20 (# 4), that is, the client 30 (# 1) and the client 30 (# 10) that they accommodate ) And a global flow exists.

  Normally, the above process is executed by a recursive method starting from the root node. FIG. 16 shows a global flow detection process flow by the network management module executed for each virtual hub 20 in the recursive process. The root node can execute a recursive process including steps S41 to S45 in both directions of two leaf nodes, and can detect a global flow when both reach the leaf node. Note that the following i) to iii) need to be taken into consideration regarding the determination of whether or not the local flow is to be compared with the threshold value and the determination of whether or not processing can continue in the leaf node direction. .

  i) Only the connection destination of the virtual port included in the local flow is the processing continuation target (S41).

  Even if there is a virtual port in the leaf node direction, the global flow detection process is not continued for a virtual port that is not included in the local flow paired with the virtual port in the root node direction. In the above example, it is the virtual port 2 in the virtual hub 20 (# 5), and the processing is not continued to the connection destination virtual hub 20 (# 7).

  ii) Only a local flow including a virtual port in the root node direction is a target for comparison processing with a threshold value (S42).

  Even if a local flow exists, if it does not include a virtual port in the direction of the root node, it will not be compared with a threshold value. In the above example, the local flow is a virtual port 1-virtual port 2 pair in the virtual hub 20 (# 2), and comparison processing with a threshold value is not performed. This is because it is not included in the flow including the virtual node 20 (# 3) that is the root node. For example, if the virtual hub 20 (# 2) is selected as the root node, this local flow is also a target of processing. Would be.

  iii) Only the connection destination of the virtual port in the leaf node direction included in the local flow that satisfies the condition is the target of the detection process continuation (S43: No, S44).

  Even if there is a local flow including a virtual port in the root node direction, the global flow search process is not continued to the connection destination of the virtual port in the leaf node direction of the local flow if the condition is not satisfied by comparison with the threshold value. . In the above example, since the local flow by the virtual port 0-virtual port 2 pair of the virtual hub 20 (# 2) does not exceed the threshold, the virtual hub 20 (# 1) that is the connection destination of the end point of this local flow. The global flow detection process is not continued.

  As described above, the example of the detection processing of two global flows based on the traffic information, in particular, the number of transfer bits per unit time between virtual ports has been described with reference to FIG. 15. However, the traffic information is not limited to this. , The number of transfer bits per unit time between virtual ports, the number of transfer bytes per unit time between virtual ports, the number of transfer frames per unit time between virtual ports, and a specific size or more per unit time between virtual ports Detection of data including number of transfer frames, specific upper layer protocol information between virtual ports (upper layer protocol information may include protocol type, IP address, UDP / TCP port number, etc.), unit between virtual ports Number of transfer bits / bytes / frames (including top-level protocol information per hour) The layer protocol information, protocol type, it is also possible to use information such as IP address, UDP / TCP port number, etc. can be considered).

(Global flow detection method 2)
In this method, a global flow is detected using address information. Note that the local flow in this method is the local flow defined in Definition 2 described above.

  The global flow detection unit 43 of the network management module 40 traces the virtual hub 20 in the direction of all the leaf nodes, with the root node determined by any one of the route node determination methods 1 to 3 as a process start node. However, the local flow that includes the virtual port in the direction of the root node of each virtual hub 20 and the local flow that triggered the virtual hub 20 on the virtual hub 20 to which the virtual port in the direction of the root node of the local flow is connected. If the value of the MAC address pair with the flow and the direction match, proceed to the virtual hub 20 that is the connection destination of the virtual port in the leaf node direction of the local flow, and finally If there is a continuous local flow between two leaf nodes, it is determined as a global flow .

  For example, in the virtual hub 20 (# 1) to the virtual hub 20 (# 7) in FIG. 17, the MAC address is learned as shown in the figure, that is, each output port (virtual port) and the MAC address are mapped. Suppose that Here, the global flow detection process when the root node is the virtual hub 20 (# 1) will be described. In the virtual hub 20 (# 1), there are a plurality of local flows. Here, a case where attention is paid to a local flow consisting of a pair of {virtual port n: MAC3} → {virtual port m: MAC6} will be described.

  In the local flow notation, “⇒” indicates the “direction” of the MAC address pair. That is, even if the MAC address pair values are equal, if this direction is reversed, the local flows are not regarded as belonging to the same global flow, and the global flow detection process is not continued. A situation in which the MAC address pair values are equal and the directions are reversed does not normally occur except when a loop occurs.

Now, paying attention to the local flow consisting of the pair {virtual port n: MAC3} ⇒ {virtual port m: MAC6},
1) In the virtual hub 20 (# 1), it is detected that a local flow with a pair of {virtual port 2: MAC3} → {virtual port 0: MAC6} exists.

  2) Next, for the virtual hub 20 (# 2) to which the virtual port 0 that is the end point in the leaf node direction of the local flow is connected, the virtual port 2 in the direction of the virtual hub 20 (# 1) that is the root node is included. The local flow is checked, and on the virtual hub 20 (# 2), a local flow having the same MAC address pair value and direction as on the virtual hub 20 (# 1), {virtual port 2: MAC3⇒virtual port 0 : MAC6} is detected.

  3) Next, for the virtual hub 20 (# 3) to which the virtual port 0 that is the end point in the leaf node direction of the local flow is connected, the virtual port 0 in the direction of the virtual hub 20 (# 1) that is the root node is included. The local flow is checked, and on the virtual hub 20 (# 3), the local flow having the same MAC address pair value and direction as the virtual hub 20 (# 1) and virtual hub 20 (# 2), {virtual It detects that port 0: MAC3⇒virtual port 1: MAC6} exists.

  4) Similarly, on the virtual hub 20 (# 5) and the virtual hub 20 (# 7), a local flow having the same value and direction of the MAC address pair, {virtual port 0: MAC3⇒virtual port 2: MAC6} , {Virtual port 1: MAC3⇒virtual port 1: MAC6} are detected.

  5) By the processing of 1) to 4), between the two leaf nodes, the virtual hub 20 (# 1) and the virtual hub 20 (# 7), that is, the client 30 (# 3) and the client 30 accommodated by them. It is determined that a global flow exists between (# 6) and.

Normally, the above process is executed by a recursive method starting from the root node. The global flow detection processing flow by the network management module 40 executed for each virtual hub in the recursive processing is basically the same as that in FIG. 16, and only the second condition determination processing portion is different.

  As described above, an example of the global flow detection process based on the address information, particularly the MAC address has been described with reference to FIG. 17, but an upper layer address such as an IP address can be used as the address information. is there. However, processing becomes complicated and performance degradation may occur.

(Global flow detection method 3)
In this method, a global flow is detected using both traffic information and address information.

  Each global flow detection method is as described in the global flow detection methods 1 and 2, but the following events may occur.

  That is, when the traffic information is based on the global flow detection method 1, the traffic volume between the virtual ports is mainly used except when referring to the upper-end protocol information, particularly the end-end information such as the IP address. It pays attention only to it, and does not care about the transmission source and the destination. In other words, since traffic with a mixture of various sources and destinations is checked, even if a global flow is detected based on the amount of traffic, it is detected based only on traffic between clients 30 at a specific end-end. Global flows are likely to be inaccurate. On the other hand, since traffic with a mixture of transmission sources and destinations is seen, in an environment where communication is performed randomly between many clients 30, a flow that shows a remarkable tendency at the end end, that is, a global flow is displayed. It may be difficult to find (end-to-end flows are buried).

  On the other hand, in the case of being based on address information as in the global flow detection method 2, if the example described in the global flow detection method 2 is taken, the Ethernet between the client 30 (# 3) and the client 30 (# 6) If frames are exchanged, each virtual hub 20 (# 1), (# 2), (# 3), (# 5), (# 7) on the route learns the MAC addresses of these clients. For this reason, although it is an extreme example, even if only one Ethernet frame is transmitted and received and there is no traffic after that, it may be regarded as a global flow.

  In view of the above circumstances, the present method detects a global flow using both traffic information and address information.

  For example, first, a global flow is detected using address information (MAC address information). Subsequently, the traffic amount of each local flow included in the global flow, that is, the traffic amount between the two virtual ports forming the local flow of each virtual hub 20 is checked, and the traffic amount of all the local flows is a certain condition. Only when the condition is satisfied, it is finally determined that the flow is a global flow.

  In this way, it is possible to reliably detect a global flow between end-to-end clients.

  In the global flow detection methods 1 to 3, the case where only the leaf node can be the end point of the global flow is illustrated, but according to the application area, one end point or both end points are It may be an intermediate node that is a virtual hub 20 that is not a leaf node, that is, a virtual hub 20 that relays traffic between other virtual hubs 20.

  Therefore, according to the peer-to-peer communication establishment method according to the present embodiment according to the present embodiment, it is possible to avoid traffic concentration and high load on a specific relay node group, the corresponding relay node group, and the entire virtual network. It is possible to ensure scalability. Further, since the traffic does not pass through a specific relay node group, it is possible to reduce the load on the underlay network and to suppress the capital investment. Furthermore, communication throughput between clients 30 and response time can be improved by performing direct communication between clients 30.

(Second Embodiment)
The second embodiment is an application example different from the first embodiment of the peer-to-peer communication establishment method of the present invention. Therefore, a different point from 1st Embodiment is demonstrated and duplication description is avoided.

  That is, in the first embodiment, the example in which the peer-to-peer communication is established between the two clients 30 by the peer-to-peer communication establishment method of the present invention has been described. The present invention is not limited to establishing peer-to-peer communication between the two, and can also be applied to peer-to-peer communication between two virtual hubs 20. That is, the present invention can also be applied to peer-to-peer direct communication between two virtual hubs 20 connected via another virtual hub 20.

  For example, in FIG. 17, it is detected that the traffic between the virtual hub 20 (# 4) and the virtual hub 20 (# 5) is very large on the virtual hub 20 (# 3). That is, when a global flow is not detected for some reason between the virtual hub 20 (# 4) and the virtual hub 20 (# 6), or between the virtual hub 20 (# 4) and the virtual hub 20 (# 7). In addition, direct communication can be performed between the virtual hub 20 (# 4) and the virtual hub 20 (# 5).

  As described above, also by the peer-to-peer communication establishment method according to the present embodiment according to the present embodiment, it is possible to avoid traffic concentration and high load on a specific relay node group, the corresponding relay node group, and the virtual network. Overall scalability can be ensured. Further, since the traffic does not pass through a specific relay node group, it is possible to reduce the load on the underlay network and to suppress the capital investment. Furthermore, by directly communicating between the virtual hubs 20, it is possible to improve the communication throughput, response time, etc. between the virtual hubs 20.

(Applied technology field)
The peer-to-peer communication establishment method of the present invention can be widely applied not only to the virtual Ethernet as described above but also to a virtual network construction technique in which traffic is concentrated on a certain node.

  The best mode for carrying out the present invention has been described above with reference to the accompanying drawings, but the present invention is not limited to such a configuration. Within the scope of the invented technical idea of the scope of claims, a person skilled in the art can conceive of various changes and modifications. The technical scope of the present invention is also applicable to these changes and modifications. It is understood that it belongs to.

The conceptual diagram which shows the structural example of the typical network with which the peer to peer communication establishment method which concerns on 1st Embodiment is applied. The functional block diagram which shows the structural example of a virtual hub. The functional block diagram which shows the structural example of a client. The functional block diagram which shows the structural example of a network management module. The conceptual diagram which shows the state which the traffic generate | occur | produced in the typical network shown in FIG. The figure which shows an example of the connection tree information produced by the information collection part. The flowchart which shows the flow of the production | generation process of a peer to peer link. The flowchart which shows the flow of the communication process by a peer to peer link. The conceptual diagram which shows the state which established the peer-to-peer communication in the typical network shown in FIG. The flowchart which shows the flow of the deletion process of a peer to peer link. FIG. 3 is a conceptual diagram of a network configuration for explaining a route node determination method 1; The figure which shows the hop number to the leaf node in the network structure of FIG. The network composition conceptual diagram for demonstrating the root node determination method 2. FIG. FIG. 5 is a conceptual diagram of a network configuration for explaining a route node determination method 3; 1 is a conceptual diagram of a network configuration for explaining a global flow detection method 1. FIG. The flowchart which shows the flow of the global flow detection process by a recursive process. The network block conceptual diagram for demonstrating the global flow detection method 2. FIG. The network conceptual diagram which shows the general structural example of virtual Ethernet.

Explanation of symbols

  b ... local flow, d ... communication link, h ... local flow, m ... virtual port, n ... virtual port, p ... peer-to-peer communication link, T ... traffic, 0-3 ... virtual port, 10 ... network, 20 ... virtual hub 21 ... Ethernet switching hub, 22 ... Transfer table, 23 ... Connection link control unit, 24 ... Virtual port, 25 ... Network management interface unit, 26 ... Peer-to-peer link control unit, 27 ... Underlay network protocol group, 28 ... Physical network Interface: 30 Client 31: Upper protocol group 32 Virtual network interface 33 Ethernet frame transmission / reception unit 34 Transmission destination management table 35 Connection link control unit 36 Virtual port 37 Underlay network protocol Group, 38 ... physical network interface, 40 ... network management module, 41 ... information collection unit, 42 ... route node determination unit, 43 ... global flow detection unit, 44 ... peer-to-peer link control unit, 50 ... LAN, 150 ... underlay network , 151 ... Virtual hub, 152 ... Virtual network interface, 153 ... Client, 154 ... Virtual Ethernet, 155 ... Server, 156 ... Firewall, 157 ... Router, 160 ... Internal network, 161 ... External network

Claims (19)

Apparatus for establishing a peer-to-peer link between specific clients in a virtual network in which a plurality of clients having a communication link with at least one of a plurality of relay nodes communicate with each other via the relay node Because
A global flow detecting means for detecting a global flow that is a communication flow that satisfies a predetermined standard and is made between two arbitrary clients among the plurality of clients via two or more continuous relay nodes; ,
Peer-to-peer link establishment means for establishing a peer-to-peer link between clients for which a global flow has been detected by the global flow detection means,
A peer-to-peer communication establishment apparatus, wherein communication between clients whose peer-to-peer link has been established by the peer-to-peer link establishment means is performed via the peer-to-peer link. In the peer-to-peer communication establishment apparatus according to claim 1,
The global flow detection means includes
For all of the relay nodes, the number of hops reached for each leaf node that is a relay node that does not relay the communication flow between the relay nodes is investigated, and the relay node with the smallest number of maximum hops or the maximum number of hops is determined in advance. Root node determination means for determining root nodes that are all relay nodes smaller than the threshold value;
A communication flow that satisfies the predetermined criteria while tracing a relay node from the root node to the leaf node side along the virtual network, starting from the root node determined by the root node determining means. A global flow determining unit that detects a global flow candidate and determines that the continuous global flow candidate is the global flow if a continuous global flow candidate is detected between two leaf nodes. A peer-to-peer communication establishment device. In the peer-to-peer communication establishment apparatus according to claim 1,
The global flow detection means includes
Root node determination means for determining a relay node having a maximum load for all of the relay nodes or a root node that is all relay nodes having the load larger than a predetermined threshold;
Starting from the root node determined by the root node determination means, following the relay node from the root node to the leaf node side that is a relay node that does not relay the communication flow between the relay nodes along the virtual network , Detecting a global flow candidate that is a communication flow that satisfies the predetermined criteria, and detecting a continuous global flow candidate between two leaf nodes, the continuous global flow candidate is defined as the global flow. A peer-to-peer communication establishment apparatus, comprising: a global flow determination unit for determining.   4. The peer-to-peer communication according to claim 2 or 3, wherein the root node determining means sets all the leaf nodes instead of all the relay nodes as target relay nodes when determining the root node. Establishing device.   The global flow detection means, starting from each of the relay nodes as a starting point, tracing the relay node along the virtual network toward the leaf node that is a relay node that does not relay the communication flow between the relay nodes. A global flow candidate that is a communication flow that satisfies a predetermined criterion is detected, and if a continuous global flow candidate is detected between two leaf nodes, the continuous global flow candidate is detected as the global flow. The peer-to-peer communication establishment apparatus according to claim 1.   The global flow detection means is a communication flow that satisfies the same conditions regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and in all relay nodes between the two leaf nodes. 2. The peer-to-peer communication establishment apparatus according to claim 1, further comprising: a global flow determination unit that determines that the communication flow between the two leaf nodes is a global flow when the two are detected.   The global flow detection means is a communication flow that satisfies the same condition regarding address information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and all relay nodes between the two leaf nodes. 2. The peer-to-peer communication establishment apparatus according to claim 1, further comprising: a global flow determination unit that determines that the communication flow between the two leaf nodes is a global flow when the two are detected.   The global flow detection means satisfies the same conditions regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes, and all relay nodes between the two leaf nodes, and The peer-to-peer communication establishment apparatus according to claim 1, further comprising a global flow determination unit that determines a communication flow between the two leaf nodes as a global flow when a communication flow that satisfies the same condition regarding address information is detected. . A plurality of clients having a communication link with at least one of a plurality of relay nodes each establish a peer-to-peer link between specific relay nodes in a virtual network in which the clients communicate with each other via the relay node. A device,
A global flow detecting means for detecting a global flow that is a communication flow that satisfies a predetermined criterion and is made between at least one other relay node among two relay nodes among the plurality of relay nodes; ,
Peer-to-peer link establishment means for establishing a peer-to-peer link between two relay nodes for which a global flow has been detected by the global flow detection means,
A peer-to-peer communication establishment apparatus characterized in that communication between relay nodes in which a peer-to-peer link has been established by the peer-to-peer link establishment means is performed via the peer-to-peer link. A method of establishing a peer-to-peer link between specific clients in a virtual network in which a plurality of clients each having a communication link with at least one of a plurality of relay nodes communicate with each other client via the relay node Because
Detecting a global flow that is a communication flow between two arbitrary clients among the plurality of clients via two or more continuous relay nodes and satisfying a predetermined standard;
Establishing a peer-to-peer link between clients where the global flow is detected;
A peer-to-peer communication establishment method, characterized in that communication between clients with which the peer-to-peer link has been established is made via the peer-to-peer link. A program for establishing a peer-to-peer link between specific clients in a virtual network in which a plurality of clients each having a communication link with at least one of a plurality of relay nodes communicate with each other via the relay node Because
A function of detecting a global flow that is a communication flow that satisfies a predetermined standard and that is made between two arbitrary clients among the plurality of clients via two or more continuous relay nodes;
A program for causing a computer to realize a function of establishing a peer-to-peer link between clients in which the global flow is detected. 12. The program of claim 11, wherein
The detecting function is further
For all of the relay nodes, the number of hops reached for each leaf node that is a relay node that does not relay the communication flow between the relay nodes is investigated, and the relay node with the smallest number of maximum hops or the maximum number of hops is determined in advance. A function to determine root nodes that are all relay nodes smaller than the threshold value,
Starting from the determined root node as a starting point, a global flow candidate that is a communication flow that satisfies the predetermined criteria is detected while following a relay node from the root node to the leaf node along the virtual network. And a function that, when a continuous global flow candidate is detected between two leaf nodes, determines that the continuous global flow candidate is the global flow. 12. The program of claim 11, wherein
The detecting function is further
A function for determining a relay node having a maximum load for all of the relay nodes or a root node that is all relay nodes having the load larger than a predetermined threshold;
Starting from the determined root node as a starting point, tracing the relay node from the root node to the leaf node that is a relay node that does not relay the communication flow between the relay nodes along the virtual network A function that detects a global flow candidate that is a communication flow that satisfies a criterion and, if a continuous global flow candidate is detected between two leaf nodes, determines that the continuous global flow candidate is the global flow; Including programs.   The program according to claim 12 or 13, wherein the determining function sets all the leaf nodes, not all of the relay nodes, as target relay nodes when determining the root node.   The detecting function starts with all of the relay nodes as starting points, tracing the relay nodes along the virtual network to the leaf nodes that are relay nodes that do not relay the communication flow between the relay nodes. A function for detecting a global flow candidate that is a communication flow that satisfies a defined standard and detecting the continuous global flow candidate as the global flow if a continuous global flow candidate is detected between two leaf nodes. 12. The program of claim 11 comprising:   The detection function is a communication flow that satisfies the same conditions regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and all relay nodes between the two leaf nodes. 12. The program according to claim 11, including a function of determining, when detected, a communication flow between the two leaf nodes as a global flow.   The detecting function is a communication flow that satisfies the same condition regarding address information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and all relay nodes between the two leaf nodes. 12. The program according to claim 11, including a function of determining, when detected, a communication flow between the two leaf nodes as a global flow.   The detection function satisfies the same condition regarding traffic information in two leaf nodes that are relay nodes that do not relay a communication flow between relay nodes and all relay nodes between the two leaf nodes, and addresses 12. The program according to claim 11, further comprising a function of determining a communication flow between the two leaf nodes as a global flow when a communication flow satisfying the same condition regarding information is detected. A plurality of clients having a communication link with at least one of a plurality of relay nodes each establish a peer-to-peer link between specific relay nodes in a virtual network in which the clients communicate with each other via the relay node. A program,
A function of detecting a global flow that is a communication flow that is made between at least one other relay node among the plurality of relay nodes via at least one other relay node and satisfies a predetermined criterion;
A program for causing a computer to realize a function of establishing a peer-to-peer link between two relay nodes in which a global flow is detected by the global flow detection means.

Images