JP2014045374A - Method of bandwidth reservation in network - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide, in a communication network in which a plurality of nodes and a plurality of communication paths connecting between nodes are included and a virtual circuit can be set on at least part of the communication paths, a method of bandwidth reservation on the communication path subjected to route searching in order to set the virtual circuit with a requested bandwidth on the communication path.SOLUTION: In each link between a node A and a node b of the virtual circuit requested to be set on the communication path: (1) when a condition that a margin bandwidth of the link obtained by subtracting a total of bandwidth of the band already reserved from a permissible band width of the link is larger than a request bandwidth of the virtual circuit is satisfied, the request bandwidth is set as a reserved bandwidth to each link between the node A and the node B; (2) when the condition is not satisfied, an operation, for each link, that the link of interest is excluded from a searching interest in the route searching is performed to execute the route searching.

Description

  The present invention relates to a bandwidth reservation method in a path network in which a plurality of nodes are connected by a flow or a path.

  As a network bandwidth reservation method, there is RSVP (Resource reSerVation Protocol) technology. This is for the transmission source node to reserve a bandwidth for the set route to the destination node, and to provide a guaranteed bandwidth for video conferencing and real-time video distribution. On the other hand, a path network represented by MPLS (Multi-Protocol Label Switching) provides a path between a source node and a destination node. The route control in the path network is generally performed in accordance with the routing control of the IP (Internet Protocol) network.

Japanese Patent Application No. 2011-285870

  Bandwidth reservation by RSVP is performed on one route (minimum cost value route) set between a source node and a destination node by a routing protocol or the like, and an overflow ( When a traffic overflow occurs, such a bandwidth request is blocked, and control for detouring to another route is not performed. In the network example shown in FIG. 1, when congestion occurs in the link between the node 14 and the node 17 on the shortest path from the node 11 to the node 18, the overflowed amount is replaced with another path, for example, the nodes 12 and 13. , Even if there is a margin in the route via 16, it is not possible to switch to a route with this margin.

  The same applies to the RSVP control over MPLS. If an overflow occurs in the path set between the source node and the destination node, the bandwidth request is blocked. Since the route is not detoured to another route, it cannot be used even if there is room in other links, resulting in waste.

  In MPLS, when a route is dynamically set, there is a problem that it takes a processing load and a processing time for setting. In addition, when a path is assigned for each communication flow, there is a problem that the number of paths to be managed increases, and consequently, the number of times of setting a path increases.

The bandwidth reservation method for a first network of the present invention includes a plurality of nodes and a plurality of communication paths connecting the nodes, and can set a virtual circuit on at least a part of the communication paths. In a telecommunication network
When a bandwidth request occurs between certain nodes,
(1) A virtual circuit has already been set up between the nodes, and in all the links on the path of the virtual circuit, the marginal bandwidth of the link obtained by subtracting the total reserved bandwidth from the allowable bandwidth of the link If the condition that the requested bandwidth is exceeded is received, the bandwidth request is accepted, and the bandwidth including the requested bandwidth is set as the reserved total bandwidth,
(2) When the above condition is not satisfied, a link that does not exceed the requested bandwidth for which the margin bandwidth is requested is excluded from a search target, and a route is searched. A virtual line is set with the requested bandwidth as the reserved total bandwidth.

The second network bandwidth reservation method of the present invention comprises a plurality of nodes and a plurality of communication paths connecting the nodes, and sets a virtual circuit on at least a part of the communication paths. In a communication network that can
Each node
(1) Functions in the case of a source node that makes a bandwidth request.
The condition that the marginal bandwidth of the link obtained by subtracting the sum of bandwidths of the already reserved bandwidth from the allowable bandwidth of the link is larger than the requested bandwidth of the virtual circuit requested to be set is satisfied. Bandwidth request means for performing processing for broadcasting a bandwidth request signal including request identification information, destination information, request bandwidth and accumulated cost value to the link;
(2) Functions when the bandwidth request signal is input.
In association with the identification information included in the bandwidth request signal, the bandwidth request management table for performing processing for managing the minimum accumulated cost value and the input port number is associated with the output port number connected to the link, and allowed. A link management table for performing bandwidth management processing and a cost value calculation for performing processing for calculating a cost value according to a predetermined procedure in accordance with the arrived bandwidth request signal and adding it to the accumulated cost value of the bandwidth request signal Means, a cumulative cost value comparing means for performing processing for comparing the accumulated cost value after addition of the arrived bandwidth request signal and the accumulated cost value indicated by the already registered bandwidth request signal, and the bandwidth request signal The bandwidth comparison means for performing processing for the output port connected to the output link, and the resource related to the output link. The marginal bandwidth of the network management table is reduced by the required bandwidth described in the bandwidth request signal, and the adjacent nodes connected to the output link are excluded from the adjacent nodes that have input the bandwidth request signal. Management table management for performing processing for broadcasting the bandwidth request signal after addition of the accumulated cost value and registering or updating the input port number and the accumulated cost value of the bandwidth request management table in association with the bandwidth request signal Means,
(3) It functions when the node is designated as a destination node by the destination information of the input bandwidth request signal.
Band securing signal sending means for sending a band securing signal including identification information of a request signal to an output link connected to an adjacent node to which an input link with the minimum accumulated cost value specified by the bandwidth request management table is connected When,
(4) It functions when the band securing signal is input.
The input port number recorded in association with the identification information with reference to the band management table and the output port number connected to the output link to the adjacent node that has transmitted the band securing signal correspond to the identification information. And a first management means for performing processing for managing the bandwidth reserved route information.
(5) The node that broadcast the band request signal functions when the band reservation signal including the same identification information as the band request signal is input.
The output port number associated with the input port number connected to the input link to which the bandwidth reservation signal is input is associated with the identification information of the bandwidth request signal included in the bandwidth reservation signal, and further related to the bandwidth request A second management means for managing the bandwidth-reserved output port;
Having
The node where the bandwidth request has occurred broadcasts the bandwidth request signal by the bandwidth request means,
The node that has received the bandwidth request signal has the cumulative result indicated by the bandwidth request signal that has already been registered in the bandwidth request management table as compared with the cumulative cost value of the bandwidth request signal that has newly arrived. When there is an output link that satisfies the condition that the cost value is larger and the comparison result in the bandwidth comparison means satisfies the condition that the marginal bandwidth is larger than the requested bandwidth indicated by the bandwidth request signal, A bandwidth request signal to which the accumulated cost value is added by the cost value calculation means is sent to the output link, and when the presence of the output link is denied, the relay to the other node of the bandwidth request signal is stopped,
The destination node specified by the destination information of the band request signal sends the band securing signal by the band securing signal sending means,
The node that has input the bandwidth reservation signal manages the bandwidth reserved route of the virtual circuit using the first management means,
The node that makes the bandwidth request and broadcasts the bandwidth request signal by the bandwidth request means manages the output port of the virtual circuit that has made the bandwidth request by the second management means.

  Further, the present invention is the first network bandwidth reservation method, wherein the route search is performed using the second network bandwidth reservation method.

Also, the network bandwidth reservation method of the present invention comprises a plurality of nodes and a plurality of communication paths connecting the nodes, and can set a virtual circuit on at least a part of the communication paths. A method for reserving a bandwidth on a communication path in a route search for setting a virtual circuit having a requested bandwidth on the communication path in a communication network,
In each link that is a candidate for the virtual circuit between the node at one end and the node at the other end of the virtual circuit that is requested to be set on the communication path,
In a network where multiple nodes are connected by a variable bandwidth path that aggregates multiple communication flows,
When a plurality of paths having different paths can be set from a certain source node to a certain destination node, and when a bandwidth reservation request is generated from the source node to the destination node,
(1) If there is a path that has already been set,
(1-1) For any link on the route that the set path follows, the requested bandwidth requested by the bandwidth reservation request is the sum of the bandwidths already reserved for the path from the allowable bandwidth of the link. The reserved bandwidth of the path of any link on the path followed by the set path is increased by the required bandwidth,
(1-2) In the route that the set path follows, if the requested bandwidth requested by the bandwidth reservation request exceeds the surplus bandwidth of any link, the requested bandwidth including the link Avoid all links until the surplus bandwidth is not exceeded, set up a new path, assign the requested bandwidth to the path,
(2) When there is no set path, a link is avoided until the requested bandwidth does not exceed the surplus bandwidth, a new path is set, and the requested bandwidth is allocated to the path. To do.

  Also, the present invention is the fourth network bandwidth reservation method, wherein the bandwidth required by the bandwidth reservation request is a value sufficient to satisfy the bandwidth reservation requests for a plurality of communication flows. It is characterized by.

Further, according to the present invention, the difference between the total reserved bandwidth in the path and the margin bandwidth of the path obtained by subtracting the total reserved bandwidth in the path from the allowable bandwidth of the path is smaller than a predetermined value. In the case where the difference is smaller than a predetermined value, a bandwidth sufficient to satisfy a plurality of flow requests is requested.

Further, the present invention has a function in which all or some of the nodes in the communication network cache content, and determines a route from the source node to the destination node using the bandwidth reservation method of the network described above. The node having the function of caching the content on the route is characterized in that the content distributed from the transmission source node is relayed and transferred along the route while caching the content distributed from the transmission source node.

When an overflow occurs in the set route, another route is set for the same destination node, so that the use efficiency of the network is improved and the throughput is expanded.
The increase in bandwidth request is handled by path bandwidth processing, so the number of paths to be managed can be suppressed.
Since paths are set for a plurality of flows, the frequency of path setting can be reduced.
Since the bandwidth is reserved in advance prior to the communication request from the flow, it is possible to process the communication request for the flow promptly even if the reservation processing takes time.

It is a figure which shows the example of a network. It is a figure which shows the example of (a) flow management table and (b) bandwidth management table. It is a figure which shows the process means of the various processes in the relay node for the route setting of a new flow. It is a figure which shows the processing flow in the relay node for the route setting of a new flow. It is a figure which shows the example of a network when the reserved bandwidth of each link has already been set. It is a figure which shows the example of (a) path management table and (b) bandwidth management table. It is a figure which shows the example of a processing flow of each relay node for the path | route setting of a new path | pass. The right direction on the horizontal axis is the direction of time progression, and the upward direction on the vertical axis is the direction of increasing bandwidth. FIG. 10 is a diagram illustrating an example in which when there is a bandwidth request from a flow, a bandwidth request sufficiently larger than the request is implemented for a path. (A) A diagram showing a simulation model network composed of 56 nodes and bi-directional 100 links, (b) a diagram showing a bandwidth securing ratio of a path or a flow against a network load, and a bandwidth securing ratio of a flow or a path is 100%. The network load (bandwidth) becomes 23 Gbps when the present invention is used, compared to 13 Gbps when the existing RSVP method based on the shortest path is used, indicating that a higher throughput can be obtained. It is a figure which shows a series of processes of the band request signal and the band ensuring signal between the nodes 11 and 18 of FIG. It is a figure which shows the example of a desirable frame structure when using this invention, when discovering the content and processing function which exist on a network.

  Embodiments of the present invention will be described below in detail with reference to the drawings. In the following description, devices having the same function or similar functions are denoted by the same reference numerals unless there is a special reason.

  First, a network bandwidth reservation method according to the present invention includes a plurality of nodes and a plurality of communication paths connecting the nodes, and is capable of setting a virtual line on at least a part of the communication paths. The present invention relates to a method for reserving a bandwidth on a communication route in a route search for setting a virtual circuit having a requested bandwidth on the communication route in a communication network. The virtual circuit is, for example, a flow in the first embodiment, and a path that aggregates a plurality of flows in the second embodiment.

Prior to the description of the embodiment, the flow and path will be described first.
A flow is a traffic flow identified and defined by IP packet or MAC (Media Access Control) frame header information. For example, a flow having a common pair of a source IP address and a destination IP address can be called an IP flow, and a flow having a common pair of a source TCP port number and a destination TCP port number can be called a TCP (Transmission Control Protocol) flow. . In the case of IP routing, different routes cannot be set for an IP flow having the same source IP address and destination IP address pair, but this is possible if managed based on the TCP flow. The present invention does not limit the flow definition method.
In the label flow switching network described in Patent Document 1 (Japanese Patent Application No. 2011-285870), a flow is defined by a label, but this also corresponds to the flow in the present invention.

  The path is generally a route through which packets are transmitted. As representatives, the levels L1, L2, and L3 of the OSI reference model are MPLS as the L3 path, VLAN (Virtual Local Area Network) as the L2 path, and L1. Examples of the path include an SDH (Synchronous Digital Hierarchy) path. An LSP (Label Switched Path) used for MPLS as a path of a packet with a label is also known.

  The destination information of the bandwidth request signal is not limited to information such as an IP address, but may be a content name or an address associated therewith. In the patent application specification, the content is specified on a label. However, using the label information as the destination information is also included. In this case, each node (1) performs relay broadcast when it does not have content corresponding to the destination information, and (2) if it has the content, the node becomes the destination node, The route search is terminated and a band securing signal is transmitted.

  This embodiment will be described with reference to FIGS. In FIG. 1, when a communication request for a flow for which a route is not set occurs, the node 11 as a transmission source sets a route and reserves a bandwidth toward the node 18 as a destination via an output port for all output links. Broadcasts a band request signal for This band request signal is relayed and arrives at the node 18. The bandwidth request signal includes a flow ID for identifying a flow, destination information, a requested bandwidth, and an accumulated cost value. Each node has the flow management table and the bandwidth management table shown in FIG. 2, and the former includes a flow ID, an input port number, a requested bandwidth, and its bandwidth for a flow related to a received bandwidth request signal. The accumulated cost value up to the node is recorded, and in the latter, the allowable bandwidth and the marginal bandwidth of the output link connected to the output port are recorded in association with all the output ports. Here, the margin bandwidth is a value obtained by subtracting the sum of reserved bandwidths from the allowable bandwidth of the link.

  A case where the number of hops is used as the accumulated cost value will be described as an example. The source node 11 sends a bandwidth request signal of a certain flow to the adjacent nodes 12 and 14. The accumulated cost value at this time is, for example, 0 as an initial value. In each of the nodes 12 and 14, the link costs of the link 11 → 12 and the link 11 → 14 from the node 11 to the node 12 are added (that is, the number of hops is incremented), and the surplus bandwidth is larger than the required bandwidth. The bandwidth request signal is broadcast to the link that satisfies the condition that is large. However, in this broadcast, the input link is excluded. If the bandwidth request signal via node 12 (solid arrow) arrives at node 15 before the bandwidth request signal via node 14 (dotted arrow), the bandwidth request signal via node 12 that arrives first is It is recorded in the flow management table and broadcasted to link 15 → 14, link 15 → 16 and link 15 → 17. On the other hand, the bandwidth request signal from the node 14 arrives later even if the accumulated cost value (in this case, the same as the number of hops) is the same, so the flow management table is not updated and the broadcast processing is not performed. In the node 14, in the link 14 → 17, the requested bandwidth is set to be larger than the surplus bandwidth. In this case, if this bandwidth request is accepted, an overflow occurs, so that the bandwidth request signal is not relayed via this link. In other words, this bandwidth request is blocked as indicated by X in the figure. As a result of these processes, the input port number and the minimum accumulated cost value are recorded in the flow management table in a position corresponding to the flow ID of the flow to be relayed, as shown in FIG.

  Eventually, band request signals arrive at node 18 from nodes 16 and 17, respectively, but the smallest accumulated cost value corresponds to the route of nodes 11 → 12 → 15 → 17 → 18 that arrived earliest. It is assumed that the band request signal to be input is input from the node 17. In this case, the node 18 transmits a band securing signal including the flow ID. Each node refers to the flow management table based on the flow ID of the band securing signal and transfers the band securing signal to the output port corresponding to the input port number. By this series of control, the band securing signal runs backward through the reserved band, and in this case, the node 18 → 17 → 15 → 12 → 11 is followed by the hop number 4 path to the source node 11. The accounting bandwidth is secured. FIG. 10 shows a series of processes.

  In RSVP, which is the prior art, bandwidth reservation processing is performed only on the shortest path (node 11 → 14 → 17) with 3 hops, so that the requested bandwidth is larger than the margin bandwidth on the link of node 14 → 17. The reservation request is blocked, but in the present invention, the next best shortest path (node 11 → 12 → 15 → 17 → 18) is set, and the link usage efficiency and thus the total throughput of the network are reduced. Expanding.

FIG. 4 shows a processing flow in the relay node that realizes the above processing.
F1: When a bandwidth request signal arrives, go to F2.
F2: Calculate the cost of the input link, add it to the accumulated cost value of the bandwidth request signal, and go to F3.
F3: It is confirmed whether or not a flow having the same flow ID as the flow ID of the arrival band request signal is recorded in the flow management table, and if it is recorded, go to F4. If not, go to F5.
F4: Check whether or not the accumulated cost value of the same flow with the registered flow ID is larger than the accumulated cost value of the arrived bandwidth request signal (Step A). To F8.
F5: Check whether there is an output port that satisfies the condition that the margin band is larger than the requested band. If there is an output port, go to F6, and if not, go to F8.
F6: Broadcast a bandwidth request signal to the port (step B), and go to F7.
F7: The flow management table is updated and this process is terminated.
F8: This process is terminated without updating the flow management table.

In this process flow,
A) Selection of a route with a minimum accumulated cost value up to the relay node,
B) A feature is that steps A and B comprise means for performing two broadcast processes for an output link that satisfies the condition of required bandwidth <margin bandwidth.

In order to realize the processing flow as described above, various means shown in FIG. 3 are used in order. These various means may be realized by hardware, or may be realized by software operating on common hardware. here,
The bandwidth requesting means has a condition that a margin bandwidth of the link obtained by subtracting a sum of bandwidths of already reserved bandwidths from an allowable bandwidth of the link is larger than a requested bandwidth of the virtual circuit requested to be set. Broadcast the bandwidth request signal to the link that satisfies it,
The cost value calculating means performs a process in which each of the nodes calculates a cost value according to a predetermined procedure according to the arrived bandwidth request signal, and adds it to the accumulated cost value of the bandwidth request signal,
The accumulated cost value comparing means performs a process of comparing the accumulated cost value after the addition of the arrived band request signal with the accumulated cost value indicated by the already registered band request signal,
The bandwidth comparison means performs a process of comparing the requested bandwidth indicated by the bandwidth request signal with a margin bandwidth of the link for the output port connected to the output link,
The management table management means reduces the margin bandwidth of the link management table related to the output link by the required bandwidth described in the bandwidth request signal, except for the adjacent node that has input the bandwidth request signal, Broadcasting the bandwidth request signal after the addition of the accumulated cost value to the adjacent node connected to the output link, and associating with the bandwidth request signal, the input port number and the accumulated cost value of the bandwidth request management table, Process to register or update
The first management means inputs the bandwidth securing signal including the identification information, and the node that is a candidate for the virtual circuit refers to the bandwidth management table and the input port number recorded in association with the identification information, The output port number connected to the output link to the adjacent node that sent the bandwidth securing signal is associated with the identification information, and further managed as bandwidth reserved route information,
The second management means, the transmission source node that has input the bandwidth reservation signal including the identification information, the output port number connected to the output link to the adjacent node that has transmitted the bandwidth reservation signal corresponds to the identification information The management process is performed as the attached bandwidth reserved route information.
Incidentally, the bandwidth request signal is a signal including request identification information, destination information, request bandwidth, and accumulated cost value.
Each node is associated with identification information included in the input bandwidth request signal, and includes a bandwidth request management table for performing processing for managing the minimum accumulated cost value and the input port number, and a link. A link management table that performs processing for managing the allowable bandwidth in correspondence with the connected output port.
Further, in the destination node specified by the destination information of the area request signal, a request is made to the output link connected to the adjacent node to which the input link having the minimum accumulated cost value specified by the band request management table is connected. A band securing signal including signal identification information is transmitted.

  This embodiment can also be used when discovering contents and processing functions existing on a network. A frame configuration used in this case is shown in FIG. It is assumed that a node for which content discovery and bandwidth reservation are to be performed is the node 11, and the requested content is held by the node 18. The band request signal shown in FIG. The broadcast address is set as the destination address, and the requested content name is included. This requested content name has both functions of destination information and flow ID. That is, the node having the requested content becomes the destination node. In this case, when the node 18 receives the bandwidth request signal, the address of the node 18 is set as the address of the sender node, and the bandwidth is secured as shown in FIG. Send a signal. Also, this series of processing is realized in the same form as in FIG.

  The present invention provides a method for reserving / reserving a bandwidth of a route, and does not limit what data is set in a router or switch in a node after the reservation. For example, the routing table can be realized by associating the input port number of the bandwidth request signal and the input port number of the bandwidth reservation signal with a common flow ID associated with the transmission / reception address pair and the content name.

  Although the present embodiment has been described in the form of node distribution, this algorithm may be implemented intensively in the center device, and an embodiment may be adopted in which the determined route and reserved bandwidth are notified to each node. Good.

  This will be described with reference to FIGS. This embodiment relates to bandwidth reservation for paths that aggregate a plurality of flows. The difference from the first embodiment is that, when a bandwidth request is generated between nodes due to the occurrence of a certain flow, when there is a bandwidth-reserved path, the required bandwidth for all links on the path of the path If the condition that the marginal bandwidth is larger than the condition is satisfied, and if this condition is satisfied, the reserved bandwidth of the set path is increased without generating a new route. . As a result, it is not necessary to perform a route search every time a bandwidth request is made at the flow level, and consumption of the number of paths (for example, the number of labels in a label path) can be suppressed.

  5, 6, and 7, the bandwidth request signal includes a path ID that specifies a path, destination information, a requested bandwidth, and an accumulated cost value. Each node has a path management table in FIG. 6 (a) and a band management table in FIG. 6 (b). In the former FIG. The ID, the input port number, the requested bandwidth, and the accumulated cost value up to the node are recorded. In the latter FIG. 6B, the allowable bandwidth and the marginal bandwidth of the output link connected to the output port ( This is equal to a value obtained by subtracting the sum of reserved bandwidths from the allowable bandwidth of the link), and is recorded in association with all output ports.

The path setting for a new path is realized by the same processing flow as in the first embodiment. The processing flow of each relay node is shown in FIG.
G1: When a bandwidth request signal arrives, go to G2.
G2: Calculate the cost of the input link, add it to the accumulated cost value of the bandwidth request signal, and go to G3.
G3: It is confirmed whether or not a path having the same path ID as the path ID of the arrival band request signal is recorded in the path management table. If it is recorded, go to G4, and if not, go to G5.
G4: Check whether or not the accumulated cost value of the same registered path ID is larger than the accumulated cost value of the arrived bandwidth request signal (step A). If it is larger, go to G5. Go to G8.
G5: It is checked whether there is an output port that satisfies the condition that the margin band is larger than the requested band. If there is an output port, the process goes to G6, and if not, the process goes to G8.
G6: Broadcast a bandwidth request signal to the port (step B), and go to G7.
G7: The path management table is updated and this process is terminated.
G8: This process is terminated without updating the path management table.

  As a path of a path in which a bandwidth from the node 11 to the node 18 in FIG. 5 is secured, for example, a path of nodes 11 → 14 → 17 → 18 has been set. The numerical values described in the square columns in FIG. 5 indicate the sum of the already reserved bandwidths of each link. Note that the allowable bandwidth of each link is, for example, 1,000 Mbps.

  Assume that a new 2 Mbps transmission request is generated from the node 11 to the node 18 in this state. First, the node 11 sends a bandwidth increase request signal for 2 Mbps to the set path. In this path route, the link with the smallest margin bandwidth is the node 14 → 17, but even if 2 Mbps is added, it is less than the allowable bandwidth, so it can be dealt with by increasing the set reserved bandwidth by 2 Mbps. Therefore, a path from the node 11 to the node 18 can be secured.

  Next, the node 18 which is the destination node reversely travels the path of the set path with the band securing signal instructing the band increase of 2 Mbps. In this case, the bandwidth can be reserved and secured without performing the broadcast process. By this processing, the reserved bandwidth of the node 14 → 17 becomes 999 Mbps.

  Furthermore, it is assumed that a 2 Mbps bandwidth request from the node 11 to the node 18 occurs. In the established path, a state occurs in which the requested bandwidth is larger than the surplus bandwidth in the link of the node 14 → 17, and the bandwidth increase request signal is blocked. This block state is notified to the node 11 using the set path route. When the node 11 recognizes the failure to increase the bandwidth using the set path route, it broadcasts a new bandwidth request signal.

  In this case, the link 12 → 15 has a situation in which the requested bandwidth is larger than the surplus bandwidth. Therefore, the use of this link is blocked, and the route is set under the condition to avoid the use of this link. The In this embodiment, prior to route setting using broadcast processing, it is checked whether a bandwidth increase request can be satisfied with a set route, so that the frequency of broadcast communication that causes an increase in traffic and processing load is suppressed. Can do.

  As a modification of the present embodiment, when the bandwidth increase request is blocked, the route search is not performed by returning to the transmission source node, but the route in which the block is generated, in this case, the node 14 is the starting point. You may search. The node 14 broadcasts the bandwidth request signal in which the new path ID is written, except for the link to the node 17. Based on this bandwidth request signal, a route having a minimum accumulated cost value satisfying the requirement bandwidth <the margin bandwidth is searched for in all links, and the node 18 sends a bandwidth securing signal to the node 14 using the route. Since it is recognized that the route from the node 11 to the node 14 is the requested bandwidth <the margin bandwidth, a new path is set using this route. Thereby, the node 11 to the node 14 share the existing path and route, and the route from the node 14 to the node 18 is searched using the broadcast processing. According to this modification method, the frequency of the broadcast processing can be further suppressed.

  When there are a plurality of paths that have been set for the same destination node, after checking whether or not the bandwidth can be increased for all the paths, the bandwidth reservation process using a new path is performed.

  This embodiment is for reducing the frequency of path setting, and will be described with reference to FIG. In FIGS. 8A and 8B, the right direction on the horizontal axis is the direction of time progression, and the upward direction on the vertical axis is the direction of increasing bandwidth. When there is a bandwidth request from the flow, a bandwidth request sufficiently larger than the request is executed for the path. For example, when the flow # 1 requests a bandwidth of 2 Mbps, for example, a bandwidth request in units of 10 Mbps is made at the path level. As a result, up to an excess of 8 Mbps, these flows can be aggregated to a set path without special control for a bandwidth request by a new flow (shared destination node). In FIG. 8A, the bandwidth requests of the flows # 2 and # 3 are dealt with without changing the path setting. For the bandwidth request of the flow # 4, the requested bandwidth is the surplus bandwidth. It becomes a larger state, generates a bandwidth request including a bandwidth increase of 10 Mbps, and starts the bandwidth reservation processing according to the second embodiment. If the same path can be used to meet the bandwidth request, it can be handled only by increasing the bandwidth of the path to 20 Mbps. If the bandwidth increase process cannot cope, a new path search is performed using the broadcast process, as in the second embodiment.

  The present embodiment is for preventing the time required for the path setting process from becoming apparent as the service processing delay time, and will be described with reference to FIG. In the third embodiment, bandwidths are reserved for the flows # 1 to # 3 at the left end of FIG. It is assumed that a new bandwidth request for flow # 4 is generated at the timing indicated by ▼ in the figure. Since the sum of the reserved bandwidth for flows # 1 to # 3 and the required bandwidth for flow # 4 exceeds 10 Mbps, a process for further increasing the 10 Mbps bandwidth is started. After the time required for the path setting process (path setting process time indicated by hatching), transmission is started at the timing indicated by ▽.

  On the other hand, in this embodiment, as shown in FIG. 8B, the process of increasing the reserved bandwidth of the path from 10 Mbps to 10 Mbps is performed when the margin bandwidth becomes smaller than a predetermined threshold. Even if it does not occur, in order to set the allowable bandwidth to 20 Mbps, the additional 10 Mbps bandwidth expansion processing is started. As the threshold value, an average flow bandwidth, for example, a value of 2 Mbps is used. This processing time is indicated by hatching. If the generation of the bandwidth request for the flow 4 (▼) is after the completion of this path setting process, the bandwidth of the flow # 4 is secured at this point, and transmission can be started immediately at the timing of ▽. Compared to FIG. 8A, the time from the generation of a bandwidth request (要求) to the start of transmission (▽) is shortened. In this case, if the band cannot be expanded on the same route, another route is set as in the above-described embodiment. As described above, by preliminarily securing the bandwidth in units of a predetermined bandwidth, it is possible to reduce the waiting time until the start of transmission.

  Note that since routing processing and switching processing are performed using packet or frame header information, the flow or path bandwidth request identification information (ID) is managed in association with these header information. It may be defined by the header information itself that defines

  The effect of the present invention is shown by simulation. FIG. 9A shows a model network, which is composed of 56 nodes and bidirectional 100 links. When the bandwidth of each link is 1 Gbps, as shown in FIG. 9B, the network load (bandwidth) at which the bandwidth reservation ratio of the flow or path is 100% as shown in FIG. When the existing RSVP method based on the shortest path is used, 13 Gbps is obtained, and when the present invention is used, 23 Gbps, which indicates that a higher throughput can be obtained.

  Using the network bandwidth reservation method of the present invention, as shown in FIG. 9, when bandwidth is reserved on the shortest path basis, bandwidth reservation is possible with a probability of 100% up to a load of 13 Gbps as in the conventional method. Naturally, when the load exceeds that, the route cost value is not the shortest, and the next-best / next-best route is set, so that the bandwidth reservation is made with 100% probability until the load becomes 23 Gbps. Is possible.

  In general, the next best / next best route has more hops than the shortest route. Therefore, by arranging a cache server that caches content for each relay node, or by configuring the relay node to have a caching function, a route with a larger number of hops is set as the load increases. Then, it is possible to create a situation in which the number of cache servers that cache content increases.

  Therefore, by increasing the number of cache servers as described above, in the content distribution service, the servers are distributed, and it is possible to receive a service from a server at a position with a smaller number of hops. And can improve the service. In this case, the cache part copies and stores the content passing through the switch part in the node.

  By using the present invention, in addition to the above, it is possible to realize a content distribution network that reserves a bandwidth for a route corresponding to the amount of content information. In addition, a virtual private network for controlling the bandwidth between nodes can be easily realized in accordance with a user bandwidth request or traffic situation.

Claims (7)

In a communication network comprising a plurality of nodes and a plurality of communication paths connecting the nodes, and capable of setting a virtual circuit on at least a part of the communication paths,
When a bandwidth request occurs between certain nodes,
(1) A virtual circuit has already been set up between the nodes, and in all the links on the path of the virtual circuit, the marginal bandwidth of the link obtained by subtracting the total reserved bandwidth from the allowable bandwidth of the link If the condition that the requested bandwidth is exceeded is received, the bandwidth request is accepted, and the bandwidth including the requested bandwidth is set as the reserved total bandwidth,
(2) When the above condition is not satisfied, a link that does not exceed the requested bandwidth for which the margin bandwidth is requested is excluded from a search target, and a route is searched. A network bandwidth reservation method, characterized in that a virtual circuit is set with the requested bandwidth as the reserved total bandwidth. In a communication network comprising a plurality of nodes and a plurality of communication paths connecting the nodes, and capable of setting a virtual circuit on at least a part of the communication paths,
Each node
(1) Functions in the case of a source node that makes a bandwidth request.
The condition that the marginal bandwidth of the link obtained by subtracting the sum of bandwidths of the already reserved bandwidth from the allowable bandwidth of the link is larger than the requested bandwidth of the virtual circuit requested to be set is satisfied. Bandwidth request means for performing processing for broadcasting a bandwidth request signal including request identification information, destination information, request bandwidth and accumulated cost value to the link;
(2) Functions when the bandwidth request signal is input.
In association with the identification information included in the bandwidth request signal, the bandwidth request management table for performing processing for managing the minimum accumulated cost value and the input port number is associated with the output port number connected to the link, and allowed. A link management table for performing bandwidth management processing and a cost value calculation for performing processing for calculating a cost value according to a predetermined procedure in accordance with the arrived bandwidth request signal and adding it to the accumulated cost value of the bandwidth request signal Means, a cumulative cost value comparing means for performing processing for comparing the accumulated cost value after addition of the arrived bandwidth request signal and the accumulated cost value indicated by the already registered bandwidth request signal, and the bandwidth request signal The bandwidth comparison means for performing processing for the output port connected to the output link, and the resource related to the output link. The marginal bandwidth of the network management table is reduced by the required bandwidth described in the bandwidth request signal, and the adjacent nodes connected to the output link are excluded from the adjacent nodes that have input the bandwidth request signal. Management table management for performing processing for broadcasting the bandwidth request signal after addition of the accumulated cost value and registering or updating the input port number and the accumulated cost value of the bandwidth request management table in association with the bandwidth request signal Means,
(3) It functions when the node is designated as a destination node by the destination information of the input bandwidth request signal.
Band securing signal sending means for sending a band securing signal including identification information of a request signal to an output link connected to an adjacent node to which an input link with the minimum accumulated cost value specified by the bandwidth request management table is connected When,
(4) It functions when the band securing signal is input.
The input port number recorded in association with the identification information with reference to the band management table and the output port number connected to the output link to the adjacent node that has transmitted the band securing signal correspond to the identification information. And a first management means for performing processing for managing the bandwidth reserved route information.
(5) The node that broadcast the band request signal functions when the band reservation signal including the same identification information as the band request signal is input.
The output port number associated with the input port number connected to the input link to which the bandwidth reservation signal is input is associated with the identification information of the bandwidth request signal included in the bandwidth reservation signal, and further related to the bandwidth request A second management means for managing the bandwidth-reserved output port;
Having
The node where the bandwidth request has occurred broadcasts the bandwidth request signal by the bandwidth request means,
The node that has received the bandwidth request signal has the cumulative result indicated by the bandwidth request signal that has already been registered in the bandwidth request management table as compared with the cumulative cost value of the bandwidth request signal that has newly arrived. When there is an output link that satisfies the condition that the cost value is larger and the comparison result in the bandwidth comparison means satisfies the condition that the marginal bandwidth is larger than the requested bandwidth indicated by the bandwidth request signal, A bandwidth request signal to which the accumulated cost value is added by the cost value calculation means is sent to the output link, and when the presence of the output link is denied, the relay to the other node of the bandwidth request signal is stopped,
The destination node specified by the destination information of the band request signal sends the band securing signal by the band securing signal sending means,
The node that has input the bandwidth reservation signal manages the bandwidth reserved route of the virtual circuit using the first management means,
The network that performs the bandwidth request and broadcasts the bandwidth request signal by the bandwidth request means manages the output port of the virtual circuit that has made the bandwidth request by the second management means. Bandwidth reservation method.   The network bandwidth reservation method according to claim 1, wherein the route search is performed using the network bandwidth reservation method according to claim 2. Bands requested on the communication path in a communication network having a plurality of nodes and a plurality of communication paths connecting the nodes, and capable of setting a virtual circuit on at least a part of the communication paths. A method for reserving bandwidth on a communication route in route search for setting a virtual circuit having a width,
In each link that is a candidate for the virtual circuit between the node at one end and the node at the other end of the virtual circuit that is requested to be set on the communication path,
In a network where multiple nodes are connected by a variable bandwidth path that aggregates multiple communication flows,
When a plurality of paths having different paths can be set from a certain source node to a certain destination node, and when a bandwidth reservation request is generated from the source node to the destination node,
(1) If there is a path that has already been set,
(1-1) For any link on the route that the set path follows, the requested bandwidth requested by the bandwidth reservation request is the sum of the bandwidths already reserved for the path from the allowable bandwidth of the link. The reserved bandwidth of the path of any link on the path followed by the set path is increased by the required bandwidth,
(1-2) In the route that the set path follows, if the requested bandwidth requested by the bandwidth reservation request exceeds the surplus bandwidth of any link, the requested bandwidth including the link Avoid all links until the surplus bandwidth is not exceeded, set up a new path, assign the requested bandwidth to the path,
(2) When there is no set path, a link is avoided until the requested bandwidth does not exceed the surplus bandwidth, a new path is set, and the requested bandwidth is allocated to the path. Network bandwidth reservation method.   5. The bandwidth reservation method for a network according to claim 4, wherein the bandwidth requested by the bandwidth reservation request is a value sufficient to satisfy a bandwidth reservation request for a plurality of communication flows.   The difference between the sum of the reserved bandwidths in the path and the marginal bandwidth of the path obtained by subtracting the sum of the reserved bandwidths in the path from the allowable bandwidth of the path is less than the specified value 6. The network bandwidth reservation method according to claim 5, wherein a bandwidth sufficient for satisfying a plurality of flow requests is requested when the value becomes smaller than a predetermined value.   In the communication network, all or some of the nodes have a function of caching content, and the network bandwidth reservation method according to any one of claims 1 to 6 is used to send a destination from a transmission source node. A node having a function of determining a route between nodes and caching contents on the route relays and forwards the content distributed from the transmission source node along the route while caching the content distributed from the transmission source node. Bandwidth reservation method.

Images