JP5252571B2 - Network management system and management method - Google Patents

Description

  The present invention relates to a technique for reducing power consumption of the entire network in a network, and in particular, dynamically configures the network configuration according to the usage status of a plurality of network devices such as routers and switches existing on the network. The present invention relates to a network management system and a management method to be changed.

  The traffic volume in the Internet is expected to increase rapidly in the future due to the increase in the network population and the spread of various communication services such as moving image distribution. Fig. 7 is an estimate of traffic volume and power consumption as shown in the "Green IT Initiative Conference (1st)" held by the Minister of Economy, Trade and Industry. In 2025, the traffic volume was compared with 2006. It is estimated to be 190 times. As the traffic volume increases, IT devices such as servers and routers that process the traffic tend to increase, and the power consumption is estimated to increase five times in 2025 compared to 2006.

  From this background and from the viewpoint of cost reduction and improvement of corporate value due to corporate social responsibility, while maximizing the operational efficiency of the network such as servers and routers, the power consumption in the network is reduced to save energy. Technology will become increasingly important in the future.

Conventional techniques related to power saving of a network can be classified into the following four types according to the part that aims to save power.
(1) Power saving technology for facility equipment.
(2) Power saving technology related to hardware.
(3) Power-saving technology for platforms and middleware.
(4) Power saving technology related to protocol, software, and operation.

As research on the facility equipment of (1) described above, thermal planning technology (Non-patent Document 1), high-voltage DC power feeding technology to IT equipment (Non-Patent Document 2), and the like have been proposed.
In addition, research on the hardware of (2) above includes multi-core technology and CPU power control technology (Non-Patent Documents 3 and 4) by Intel and AMD, and EEE (Energy Efficient Ethernet (registered trademark) of the standardization organization IEEE )) (Non-patent document 9) and Green Ethernet (registered trademark) (Non-patent document 12) by D-Link have been proposed.
Furthermore, as a research on the platform and middleware of (3) described above, virtualization technologies such as VMware, Xen, and Hyper-V (Non-Patent Documents 5 and 6) have been proposed.

In addition, regarding the power saving technology by the protocol (4), software, and operation described above (4), a virtual represented by VMware's DPM (Distributed Power Management) (Non-patent Document 7, Patent Document 1, Non-patent Document 8). There is an operation management technology targeted at a computer platform.
This virtualization technique is a technique for making one computer resource (physical machine) physically appear as a plurality of computer resources (virtual machines). Computer resources (CPU, memory, network bandwidth, hard disk, etc.) of physical machines are divided and allocated to virtual machines.
In other words, the allocation amount of the computer resources of the physical machine to the virtual machine is dynamically changed according to the usage situation by applying the virtualization technology, and unnecessary physical machines are stopped or suspended.
However, in the above-described technology, it is possible to change the configuration according to the use situation and to save power by stopping or halting unnecessary devices. However, a server group in which the application range of power saving is virtualized In addition, there is a problem that it is limited to the same network segment.

In IEEE EEE (Energy Efficient Ethernet (registered trademark)) (Non-Patent Document 9) of the standardization organization, for example, changing the interface speed according to the utilization rate of the network interface (node communication interface) is being considered. Yes. Some routers and switches (nodes) realize power saving by stopping the power supply of ports unnecessary for communication and changing the output signal strength according to the cable length (Non-Patent Document 10, 11, 12).
However, in any technique, the target range of power saving is limited to a single node. Therefore, even when the utilization rate of the entire network is low, all nodes on the network need to be activated, and there is an inconvenience that power consumption cannot be reduced in the entire network.

The technique described in Non-Patent Document 13 focuses on a node (switch) on a network having links in which a plurality of physical links are logically single links by a link aggregation technique. At present, regardless of the amount of traffic flowing through the link, aiming to save power in a switch that is always operating at full capacity, the number of physical links is increased according to the amount of traffic flowing through the link, and other physical links are set to sleep. State.
However, because the scope of power saving is limited to the links between nodes, even if the entire network utilization rate is low (some links can be in the sleep state) Node must be activated, and power consumption cannot be reduced in the entire network.

Further, in the technique described in Non-Patent Document 14, a power saving technique in a management server that centrally calculates the route of the optical network for an optical network as a typical target is proposed. As shown in FIG. 6, this management server 60 uses an existing protocol such as SNMP (Simple Network Management Protocol) and uses the node usage status (the amount of traffic flowing through the node interface, etc.) from all the nodes 61 in the network. collect.
Then, the management server 60 calculates the network configuration so that the power consumption is minimized according to the collected usage status of each node, and the management server 60 changes the topology setting of each node 61 so that the calculated configuration is obtained ( Which route of the layer 2 topology 62 is used to flow traffic is changed). In other words, the topology configuration of the node 61 is changed according to the usage status, and further, power saving is achieved by stopping or halting the node 61 that is no longer necessary for communication (no longer relaying traffic due to a route change). Has been done.

  However, Non-Patent Document 14 is a centralized management approach in which each node is managed by a management server, and the amount of calculation required for the management server to obtain a network configuration that minimizes power consumption as the number of devices increases. There was a problem that the number increased explosively.

In addition, RIP (Routing Information Protocol) or OSPF can be used as a technology for keeping nodes (routers) on the network exchange information with other nodes and keeping the layer 3 topology (IP routing table) always appropriate and up-to-date. Dynamic routing techniques represented by (Open Shortest Path First) have been proposed.
However, this dynamic routing technology does not take into account the usage status when configuring the layer 3 topology, so the network configuration (addition or deletion of devices, breakdowns, broken links, etc.) (Unless it is) fixed use, it is not possible to save power.

JP2007-310791A

HP Thermal Planning Service: http://h50146.www5.hp.com/doc/catalog/services/pdfs/AS_070202_1.pdf Department of Energy National Research Institute LBNL (Lawrence Berkeley National Laboratory): "Energy-Efficient Direct-Current-Powering Technology Reduces Energy Use in Data Centers By Up to 20 Percent" http://www.lbl.gov /Science-Articles/Archive/EETD-DC-power.html V. Raghunathan, M. Srivastava and R. Gupta, "A Survey of Techniques for Energy Efficient On-Chip Communication" in Proceedings of Design Automation Conference 2003, Anaheim, CA, Jun. 2003. T. Pering, T. Burd and R. Bordersen, "The Simulation and Evaluation of Dynamic Voltage Scaling Algorithms," in Proceedings of the International Symposium on Low Power Electronics and Design, Monterey, CA, Aug. 1998. VMware: http://www.vmware.com/ Xen: http://www.xen.org/ VMware Distributed Power Management (DPM) Keisuke Hatazaki, Yoshifumi Takamoto, "Power-saving Policy Operation Technology for Systems Using Server Virtualization," IEICE Technical Report Computer System Study Group CPSY2006-44 Vol.106, No.436, pp. 37- 42, Dec. 2006. IEEE 802.3 Energy Efficient Ethernet (registered trademark) Study Group: http://grouper.ieee.org/groups/802/3/eee_study/index.html M. Gupta and S. Singh, "Greening of the Internet," in Proceedings of ACM SIGCOMM 2003, Karlsruhe, Germany, Aug. 2003. M. Gupta and S. Singh, "Energy conservation with low power modes in Ethernet LAN environments," in Proceedings of IEEE INFOCOM, Anchorage, Alaska, May 2007. D-Link Green Ethernet (registered trademark): http://www.dlink.com/corporate/environment/dlink-green-ethernet (registered trademark) / Yoshiyuki Kawano, Yutaka Fukuda, Hitomi Tamura, Kenji Kawahara, Yuji Oie, “Proposal of Power Saving Method Using Dynamic Physical Link Control,” IEICE Technical Report Information Network Study Group IN2007-216 Vol. 107, No.525, pp.343-348, Mar. 2008. Yutaka Arakawa, Daisuke Ishii, Aya Tsuruzaki, Naoaki Yamanaka, Hiroyuki Ishikawa, Yasuhiro Sonami, "Network Reconfiguration Technique for Low Power Consumption of Network," IEICE Technical Report Photonic Network Study Group PN2008-16 Vol .108, No.183, pp.13-18, Aug. 2008.

  In the technologies described in Patent Document 1 and Non-Patent Documents 1 to 14 described above, there was no idea of reducing power consumption in a distributed manner for all nodes existing on the network.

  The present invention has been proposed in view of the above circumstances, and in order to reduce the power consumption of the entire network, it can be implemented in a part or all of a plurality of network devices such as routers and switches existing on the network. An object of the present invention is to provide a network management system and management method that dynamically changes the configuration of a network in accordance with the use status of the network.

In order to achieve the above object, the present invention is an autonomous distributed dynamic network that aims to reduce power consumption in the entire network while considering communication quality from the user's viewpoint such as transfer rate and transfer delay. Are a network management system and management method.
That is, in the network management system according to claim 1, in at least a plurality of nodes of a network in which a large number of nodes are connected by links,
A connection form grasping means for grasping the current connection form of the network and the history of the connection form by obtaining interface information of each node including those in a stopped state to be described later ;
Information sharing means for exchanging information about the interface usage status information of each node including those in a stopped state, which will be described later, and information on the cost of each link between the nodes;
Topology acquisition means for newly obtaining an IP routing table (layer 3 topology) from information of the connection form grasping means and information sharing means;
Topology change means for dynamically changing the layer 3 topology based on the result of the topology acquisition means;
Due to the configuration change of the layer 3 topology by the topology change means, the power to the unused links and nodes other than those corresponding to WakeOnLAN are shut off and all functions (all functions including the OSPF extension part) are stopped. A node stop means for setting the stop state;
A node return means for returning a link or a node from a stopped state according to an instruction from another node;
The node stop unit and the node return unit are a power management unit provided to perform power management of the node or link, and the power management unit restarts the stopped link or node so that an adjacent node A function to send a packet instructing restart to a stopped node and a function to start a packet when a packet instructing restart of a link or node whose power is stopped are provided. It is said.

  A network management system according to a second aspect of the present invention is the network management system according to the first aspect, wherein the topology acquisition unit changes the cost of each link from the information obtained by the information sharing unit, whereby the IP routing table (layer 3 topology) It is characterized by restructuring.

A third aspect of the present invention is a network management method performed in at least a plurality of nodes of a network in which a large number of nodes are connected by links, and includes the following procedures.
A procedure for grasping the current connection form of the network and the history of the connection form by obtaining interface information of each node including those in a stopped state to be described later .
A procedure for obtaining information by exchanging information on interface usage status of each node including those in a stopped state, which will be described later, and information on the cost of each link.
A procedure for newly obtaining an IP routing table (layer 3 topology) from the information of the connection form grasping means and the information sharing means.
When the newly obtained layer 3 topology increases the number of links, a procedure for returning the node from the stopped state to the operating state by instructing the node where the adjacent node has stopped to restart.
A procedure for dynamically changing to the newly obtained layer 3 topology.
A procedure for shutting down all the functions by shutting off the power except for the portion corresponding to WakeOnLAN for links and nodes that have become unused due to the topology change.

According to the network management system and management method of the present invention, a new IP routing table (layer 3 topology) is obtained from the information of the connection form grasping means and the information sharing means for grasping the current connection form and the history of the connection form. Changing the topology of the network dynamically by changing the topology, stopping the node itself or links between nodes that are no longer necessary for communication (stopping all functions by shutting off the power except for the part corresponding to WakeOnLAN) Status). As a result, a distributed power saving technique for the entire network is realized, a power saving effect is obtained by stopping unnecessary links or nodes, and the power consumption of the entire network can be reduced.

Currently, a network is generally designed for a time zone in which the amount of data handled by the network is the largest, and its configuration is fixedly used. For this reason, even when the network usage rate is low, the number of routers and switches or links between them does not change. Therefore, the amount of power consumption is high and the usage rate is low. Then it does not change greatly.
On the other hand, application of the present invention stops nodes (routers) and links that are no longer necessary for communication, so that the power consumption of the entire network can be reduced in proportion to the amount of data.

It is the model figure which imaged the operation | movement of the time zone with a high utilization rate in the network management system of this invention, and a low time zone. It is a block diagram which shows the function structure at the time of implement | achieving an autonomous distributed power saving protocol by extending | stretching the network management system of this invention by OSPF. It is a flowchart figure which shows the processing sequence of the functional unit of an autonomous distributed power saving protocol. It is a flowchart figure which shows the processing sequence of the flooding information update part of FIG. It is a model figure which shows the operation example of the autonomous distributed power saving protocol by the extension of OSPF. It is a model figure which shows the network structure proposed by the nonpatent literature 14. It is the graph which showed the estimation of traffic amount and power consumption.

A network management system and management method according to an embodiment of the present invention will be described with reference to the drawings. The network management system and management method of the present invention are realized by a network composed of a plurality of nodes (routers) in which an autonomous distributed power saving protocol is implemented.
The autonomous distributed protocol for realizing the network management system and management method according to the present invention includes a function 1 (network usage status management function) for grasping network link information, an interface that changes power consumption of each node, and changes with time. Usage status information, function 2 for sharing link cost information (network usage status database function), function 3 for managing link cost used for calculating layer 3 topology (flooding information management function), and network usage status In order to change the layer 3 topology according to the network, the function 4 (flooding information update function) for changing the link cost according to the network usage rate and the flooding information managed by the function 3 are exchanged every time there is an update. The interface of all routers in the network The function 5 (flooding information database function) for grasping the flooding information set for each, the function 6 (flooding information history database function) for managing the history of flooding information, and the flooding information managed by the functions 5 and 6 Function 7 for obtaining the layer 3 topology and changing the layer 3 topology (route determination function) and function 8 for stopping the link and node that are no longer needed and returning the link and node from the stopped state (power circuit management function) By doing each, energy saving in the network is planned. Note that information about the power consumption of each node may not be acquired.

  That is, as shown in FIG. 1, in a time zone with a high utilization rate, a layer 3 topology is used by using all of the layer 2 topologies 2 to which the nodes 1 to which the management system of the present invention is applied are physically connected. 3 is configured, but in times when the usage rate is low, unnecessary nodes and links are stopped by the function 8 described above, and the new link cost updated by the function 4 is used to change to the layer 3 topology 3 by the function 7. By doing so, the number of nodes 1 and links 2 in the operating state is reduced (configuration change from left to right in FIG. 1), and energy saving of the entire network is achieved.

Hereinafter, a specific configuration of the management system of the present invention will be described with reference to FIG. 2, taking as an example a case where an autonomous distributed power saving protocol is realized by extending a link state routing protocol (OSPF) in a node. .
The link state type routing protocol (OSPF) that has existed in the past includes a link state monitoring module 11, a flooding information management unit 13 that shows information about the interface including the cost of each link of the node itself, and a link state monitoring module 11. The flooding information database (link state database) 12 that stores flooding information by the flooding information management unit 13 and the exchanged flooding information and link (layer 2 topology) information, and the flooding information database (link state) A route determination unit 14 for determining a layer 3 topology based on the database 12. The function 1 for grasping the link information of the network is based on the diversion of the procedure in which the link state monitoring module 11 of the existing link state type routing protocol exchanges flooding information with other nodes and stores it in the flooding information database 12. Realized (execution of function 1).

In addition to these configurations, the management system of the present invention extends the network usage status management unit 15, the network usage status database 16, the flooding information update unit 17, the power management unit 21, and the flooding information history database 22. Are provided.
The control unit 19 controls the entire system by transmitting and receiving information between the units.

The network usage status management unit 15 indicates information related to the bandwidth of each link of the node itself, the amount of traffic transmitted and received by the link, and the power consumption of the node. The network usage status of the network usage status management unit 15 is the same as when the flooding information from the flooding information management unit 13 is exchanged with other nodes by the link state monitoring module 11 and stored in the flooding information database 12. All network usage statuses exchanged are stored in the network usage status database 16 (execution of function 1 and function 2). In addition, the network usage status database 16 is used for grasping usage status that changes from moment to moment. Therefore, by referring to the network usage status database 16, the usage statuses of all routers and interfaces (corresponding to the protocol of the embodiment) in the current network can be known.
In addition, the flooding information management unit 13 obtains information (for example, corresponding to link cost in the case of OSPF) for calculating what route the node uses to propagate traffic (layer 3 topology). Is managed for each interface (execution of function 3).

  The flooding information update unit 17 changes the cost (flooding information 13) assigned to each link based on the information in the network usage status database 16 (execution of function 4) (sequence shown in FIG. 4 described later).

The flooding information database (link state database) 12 is updated by exchanging the flooding information 13 with another node by the link state monitoring module 11 (execution of function 5). That is, the node exchanges with other nodes in the network every time the flooding information (set for each interface) managed by the function 3 is updated. The node stores the exchanged information in the flooding information database 12 by the function 5. By referring to the flooding information database 12, the flooding information set for each interface of all routers (corresponding to the protocol of the embodiment) in the current network can be known.
Similarly, the network usage status database 16 is updated by exchanging the network usage status of the network usage status management unit 15 with other nodes. Each node can grasp the latest information on the power consumption of each node, the usage status information of the interface that changes from time to time, and the cost of the link by the information in the flooding information database 12 and the network usage status database 16. (Execution of function 2).
The link cost (flooding information managed by the flooding information management unit 13) is initially allocated by the operator arbitrarily based on, for example, the transmission delay of the route (link), but the flooding information update unit 17 is assigned to each node. The link cost is changed (increased or decreased) according to the usage status (network usage status database 16).

  The route determination unit 14 includes topology acquisition means for newly obtaining an IP routing table (layer 3 topology) from information (flooding information history database 22) of the connection form grasping means and information sharing means, and a layer corresponding to the network usage situation 3 Topology is calculated (execution of function 3). Furthermore, the route determination unit 14 includes topology change means for dynamically changing the layer 3 topology based on the calculation result, and changes the configuration to the calculated layer 3 topology (execution of function 7).

The power management unit 21 manages the power supply to each node and link. In order to restart the stopped link or node, the adjacent node transmits a packet instructing the restart to the stopped node. A function to start when a packet instructing reactivation is received while the power is stopped, and a function to notify other nodes that reactivation has failed when reactivation fails.
The packet for instructing re-operation is implemented with a function equivalent to the product name Wake OnLAN (WOL) that has already been commercialized, or a network-compatible power control switch is installed with the node, and the power control switch is It is done by starting up. If it is an unnecessary node, it stops itself.
In other words, the power management unit 21 includes a node stop unit that shuts down and stops power to links and nodes that are not used due to the configuration change of the layer 3 topology by the topology change unit, and instructions (packets) from other nodes. ) To return the link or itself (node) from the stop state of the power stop, and the link or node is stopped and the link or node is returned from the stop state, respectively. (Execution of function 8) is designed to save energy in the entire network.

The flooding information history database 22 has a database for managing history of flooding information of nodes in a power-off state in addition to the function of the flooding information database 12 for managing the current flooding information. The above-described flooding information database (link state database) 12 indicates the current state of the network at the present time. For example, when there is a node whose power is stopped by the protocol of this embodiment, the flooding information of the node is the flooding information database. 12 is deleted.
On the other hand, the flooding information history database 22 is the same as the function of the flooding information database 12 except that the flooding information of the node is not deleted from the database except in the following cases (1) or (2). is there).
(1) When a notification packet indicating that the node is not managed by the protocol of this embodiment arrives from the node.
(2) When the power management unit 21 fails to execute the power management function (re-operation of the node), and a packet notifying that has arrived.

  Next, a processing sequence of the autonomous distributed power saving protocol implemented in each node will be described with reference to FIG. The processing sequence of FIG. 3 is repeatedly executed by each node.

  When the autonomous decentralized power saving protocol is activated in each node, first, the functions 3, 5 and 9 described above are executed to exchange interface information and the like with other nodes, and the network connection form (layer 2 topology) , Link information) is grasped (step S101). That is, each node exchanges flooding information with other nodes, and further monitors the aliveness of the link to obtain how the nodes are connected and the network is configured (current topology information).

  When the grasp of the link information is completed, the functions 1 and 2 are executed to exchange and share the power consumption of the node, the interface usage information that changes from moment to moment, and the link cost information (step S102). ). That is, each node exchanges the network usage status and obtains how much interface is used and how much traffic is flowing in the entire network. At this time, the cost of the link is changed based on the usage status of the network. The change procedure will be described later.

Next, by using the information obtained in step S101 and step S102 and executing functions 4, 5, and 6, a layer 3 topology that takes into account power consumption is calculated (changed based on network usage) By calculating the layer 3 topology from the cost of the link, the layer 3 topology considering the power consumption is obtained) (step S103). The calculation of the layer 3 topology is constructed so that the cost between nodes is minimized.
The calculated layer 3 topology is a network configuration that uses a large number of links and nodes and processes traffic at a high speed, for example, when the amount of traffic handled by the network is large in accordance with the use status of the network. In addition, when the traffic volume is small, the flood information that is the basis for calculating the layer 3 topology is calculated so that the network configuration is such that traffic is processed using only some links and nodes. Update.

Each node executes the functions 3, 5, and 6 to exchange the flooding information updated by the flooding information management unit 13 with other nodes, so that the flooding information database 12 and the flooding information history database 22 are updated to the latest state. (Step S104). The flooding information history database 22 includes information on nodes and links that are stopped, while the flooding information database 12 manages the current topology (information on nodes and links that are stopped is deleted).
Each node uses the flooding information history database 22 to determine what layer 3 topology is to be used (execution of function 7). Here, by using the flooding information history database 22, it is possible to generate a layer 3 topology including a link or a node that is stopped (step S105).

Next, it is determined whether or not to change from the current layer 3 topology (step S106).
That is, when the topology calculated in step S105 (changed according to the network usage status in order to realize power saving) and the current topology (obtained by execution of steps S101 and S104 and the flooding information database 12) are the same: The process ends.
If there is a topology change, it is determined whether to use the stopped node and link (step S107).

  The topology calculated in step S105 is compared with the current topology, and when the calculated topology is smaller than the number of nodes and links currently in use (when stopped nodes and links are not used), the layer 3 topology is changed. Then, the links and nodes that are no longer necessary as a result of the topology change are stopped by the function 8 (step S113), and the process is terminated. Compare the calculated topology with the current topology, and the calculated topology is greater than the number of nodes and links currently in use (if you need to re-operate them because they use stopped nodes and links) Then, the stopped node or link is restarted by executing the function 8 (step S108).

  When the stopped node or link is restarted in step S108, it is determined whether or not all nodes and links have been restarted successfully (step S109). If all nodes and links have been successfully restarted, the layer 3 topology is changed (step S112). If some of the nodes or links fail to operate again, an update operation is performed to delete the failed node and link information from the flooding information history database 22 by executing function 6 (step S110). ).

Then, the layer 3 topology is calculated again by the function 7 using the updated flooding information history database 22 (the same processing as step S105), and the process proceeds to step S106 for comparing the calculated topology with the current topology (step S11). ). After step S106, the same processing as described above is performed.
Thereafter, the operations in steps S101 to S113 are continuously performed in each node.

Subsequently, with respect to the procedure for changing the cost assigned to each link according to the network usage status information performed by the flooding information updating unit 17 (cost change in step S2), the flowchart of FIG. 4 and the model diagram of FIG. Will be described with reference to FIG.
The cost assigned to each link of the network is generally set so that the load of network traffic is distributed by the operator. As a result, the link cost set by the operator uses all the routers 1 to process the traffic as shown in the link (layer 2 topology) 2 in the “high utilization period” in FIG. (A cost value is set to “40” or “50” in the example in the figure). In FIG. 5, the numbers on the link 2 connecting the routers 1 indicate the costs. In addition, although it is possible to assign a different cost for each direction (for example, a different cost can be assigned to Router A → Router B and Router B → Router A), here it is bidirectional for simplicity. Assume that the same cost is assigned. And the network load is distributed most efficiently.

In the management system of the present invention, the cost value of the link (layer 2 topology) 2 is changed in each node according to the use status of the network. That is, the autonomous distributed power saving protocol calculates a “trend” from the network usage status database 16 as to whether the utilization rate of the entire network is increasing or decreasing (step S201). The calculation method can be calculated, for example, by obtaining two moving averages having different usage rate periods and considering the intersection as a trend change. That is, the usage rate tends to increase when “the moving average of the short period exceeds the moving average of the long period”, and the usage rate increases when “the moving average of the short period rotates around the moving average of the long period”. Recognize that it tends to decrease (Reference URL: http://www.nsjournal.jp/stock/tech_nyumon/1-1.php).

  Next, it is determined whether the node is an “edge router” or a “core router” (step S202). The “edge router” is a router directly connected to the user, and corresponds to the routers GN in FIG. The “core router” means a router other than the “edge router”.

Subsequently, when the node is an “edge router”, it is determined whether the “trend” calculated in step S201 is “a tendency to increase the utilization rate” or “a tendency to decrease the utilization rate” ( Step S203) If the “utilization rate tends to increase”, the cost of the link is changed to a direction to be smoothed (step S204).
Smoothing the cost of a link refers to changing the cost of each link so that all routers in the network are used, as shown in the “time zone with high utilization” in FIG.
That is, assume that the cost of the link set by the operator is the cost of the link that distributes the network load most efficiently (uses all routers to handle traffic) (referred to here as the default link cost). If it is possible, the cost of the link changed by the process described later is returned to the default link cost. Moreover, it is good also as a process which reduces the difference of the link cost between the interfaces in a certain router.

  If the node is an “edge router” and the “trend” calculated in step S201 is “a tendency to reduce the utilization rate” (step S203), the cost of the link is changed to be aggregated in the link. (Step S205). Specifically, it refers to the act of lowering the cost of links to some routers and raising the cost of links to other routers.

The link that lowers (or raises) the cost may reduce the cost of the link with the smaller router ID of the OSPF router, and the operator prioritizes the priority of the routers to be aggregated (priority to lower the cost of the link). May be specified.
In addition, as the cost reduction (increase) range, a value calculated by a mathematical formula set in advance using the network usage status as a parameter is used.
5 When attention is paid to the router G, this process lowers the cost of the link to the router C (40 to 20) and raises the cost of the link to the router D (50 to 80) during the low utilization period. )

  When the node is a “core router”, it is determined whether the “trend” calculated in step S201 is “a tendency to increase the utilization rate” or “a tendency to decrease the utilization rate” (step S206). If “the usage rate tends to increase”, the cost of the link is changed in a direction to be smoothed (step S207). The process of step S207 indicates changing the cost of the link so that all routers in the network are used, as in the process of step S204.

  If the node is a “core router” and the “trend” calculated in step S201 is “a tendency to reduce the utilization rate” (step S206), the usage rate of the router itself is “increased”. Or “decreasing” (step S208). If “rising”, the cost of the links of all the interfaces is lowered so that the traffic is concentrated on the node (step S209).

When the node is a “core router” and the “trend” calculated in step S201 is “a tendency to decrease the usage rate” (step S206), the usage rate of the router itself is “decreasing (of the node) The sum of the cost of the interface increases.The process of steps S204 and S205 increases the sum of the cost of the interface. For example, in the case of router D in FIG. Thus, the cost of all links is increased (step S210). In the case of “rising (the sum of the costs of the interface of the node is reduced. The sum of the costs of the interface is reduced by the processing in steps S204 and S205. For example, the router D in FIG. 5)” The cost of all interface links is lowered (step S209).
By performing the above processing for each node constituting the network, the flooding information updating unit 17 changes each link cost according to the usage status of the link network. As a result, a route is determined based on the cost of each link changed accordingly, and as a result, a layer 3 topology in consideration of power consumption is calculated.

Figure 5 shows an example of operation when the autonomous distributed power-saving protocol by extending OSPF is applied. In the case where the time period when the utilization rate is high (the upper position model diagram in FIG. 5) becomes the time period when the utilization rate is low (the lower model diagram in FIG. 5), the sequence processing described in FIG. The topology is changed. In the case of the model diagram of FIG. 5, the link costs in the time zone where the usage rate is high (model diagram in the upper position in FIG. 5) are set to “40” and “50”, and the time zone where the usage rate is low (in FIG. The link cost in the middle model diagram) is set to “20” and “80”, so that it is possible to construct a topology (layer 3 topology 3) that concentrates data on a link with a low cost.
As a result, traffic does not flow through the router D and the router F in the “time period when the usage rate is low” in FIG. 5 (the user A does not pass through any communication), so It is possible to significantly reduce the amount of power consumption by stopping the power supply of the router D and the router F by the node stop means of the power management unit 22.

  In this state (the state of the model diagram in the middle position of FIG. 5), when it is desired to restart the stopped router D and router F, the power management unit 22 causes the router J and the router J adjacent to the router D and the router F to The router N transmits a start instruction packet for instructing the stopped node to restart (model diagram in the lower position in FIG. 5). The router D and the router F that have stopped power supply receive the activation instruction packet, transmit a packet instructing re-operation to the power management unit 22, and activate the router D and the router F by the node return means of the power management unit 22. Thus, it is possible to return to the layer 3 topology in a time zone with a high utilization rate (upper position model diagram of FIG. 5).

  A general network is designed according to a time zone in which the amount of data handled by the network is the largest, and its configuration is used in a fixed manner. For this reason, the number of routers, switches, or links between them does not change even when the network usage rate is low, so the amount of power consumption is also high when the usage rate is low and when the usage rate is low. Then, the situation did not change greatly.

By adopting the network management system and management method as described above, the network configuration is dynamically changed according to the network usage status, and the routers and switches that are no longer necessary for communication, or the links between them are stopped. In this state , the power consumption of the entire network can be changed according to the data amount.

  In addition, by using the network management system and management method as described above, the layer 3 topology generated by the autonomous distributed power saving protocol allows the node itself to discover a network configuration that reduces the power consumption of the entire network, It becomes possible to reduce while ensuring the quality of the user's viewpoint.

CAIDA (http://www.caida.org/home/) publishes changes in the traffic volume of a network. For example, if we focus on the network traffic volume in Chicago, USA for the past 24 hours until September 1, 2008 at 10:03 (UTC), the amount of data in the time zone where the utilization rate is the maximum, and the utilization rate When comparing the data amount in the average time zone, it was “(data amount at the time of average utilization rate) / (data amount at the maximum utilization rate) ≈0.67 (67%)”. For this reason, if the energy saving by the configuration of the present invention functions most ideally, it can be expected that the power consumption of the entire network is also about 67% (about 33% reduction).
Actually, it is necessary to consider the overhead due to the introduction of the network management system of the present invention and the power consumption due to surplus links and nodes that should be secured in case of fluctuations in usage rate or failures in the above values. .

  DESCRIPTION OF SYMBOLS 1 ... Node, 2 ... Link (layer 2 topology), 3 ... Layer 3 topology, 11 ... Link state monitoring module, 12 ... Flooding information database, 13 ... Flooding information management part, 14 ... Route determination part, 15 ... Network use condition Management unit, 16 ... Network usage status database, 17 ... Flooding information update unit, 19 ... Control unit, 21 ... Power management unit (node stop means, node return means), 22 ... Flooding information history database.

Claims (3)

In at least several nodes of the network where many nodes are connected by each link,
A connection form grasping means for grasping the current connection form of the network and the history of the connection form by obtaining interface information of each node including those in a stopped state to be described later ;
Information sharing means for exchanging information about the interface usage status information of each node including those in a stopped state, which will be described later, and information on the cost of each link between the nodes;
Topology acquisition means for newly obtaining an IP routing table (layer 3 topology) from information of the connection form grasping means and information sharing means;
Topology change means for dynamically changing the layer 3 topology based on the result of the topology acquisition means;
Due to the configuration change of the layer 3 topology by the topology change means, the power to the unused links and nodes other than those corresponding to WakeOnLAN are shut off and all functions (all functions including the OSPF extension part) are stopped. A node stop means for setting the stop state;
A node return means for returning a link or a node from a stopped state according to an instruction from another node;
The node stop unit and the node return unit are a power management unit provided to perform power management of the node or link, and the power management unit restarts the stopped link or node so that an adjacent node A function to send a packet instructing restart to a stopped node, and a function to start these when a packet instructing restart of a link or node whose power is stopped,
A network management system comprising:   The network management system according to claim 1, wherein the topology acquisition unit reconstructs the IP routing table (layer 3 topology) by changing the cost of each link from information obtained by the information sharing unit. . A network management method performed by at least a plurality of nodes of a network in which a large number of nodes are connected by links.
A procedure for grasping the current connection form of the network and the history of the connection form by obtaining interface information of each node including those in a stopped state to be described later ,
A procedure for obtaining information by exchanging information on interface usage status of each node including those in a stopped state, which will be described later, and information on the cost of each link;
A procedure for newly obtaining an IP routing table (layer 3 topology) from the information of the connection form grasping means and the information sharing means;
When the newly obtained layer 3 topology increases the link, a procedure for returning the node from the stopped state to the operating state by instructing the node where the adjacent node has stopped to restart,
Dynamically changing to the newly determined layer 3 topology;
Procedures for shutting off all functions by shutting off the power except for the part corresponding to WakeOnLAN for links and nodes that are not used due to the topology change,
A network management method comprising:

Images