JP4200522B2 - Resource calculation apparatus and resource calculation method - Google Patents

Description

  The present invention relates to a resource calculation device and a resource calculation method.

  A network operator that provides a virtual network service such as VPN (Virtual Private Network), etc., has a contract model as a precondition for the calculation method of network resources (bandwidth) allocated to each virtual network according to the contract contents with the user. There are a hose model and a pipe model, and there is a resource allocation method for each model (Non-Patent Document 1, Non-Patent Document 2).

  First, the hose model is a model in which only the upper limit of the inflow / outflow traffic to the provider edge device is contracted with the customer. In the resource allocation method for the hose model, only the upper limit value of the inflow / outflow traffic to the edge device (provider edge device) that accommodates the user side device (customer edge device) is specified in the contract, and the bandwidth is allocated. It is a calculation method. Specifically, this is a method of creating a topology of a tree structure and calculating resources to be allocated at each link on the tree. The traffic between any provider edge devices is allowed as long as it is within the upper limit of the inflow / outflow.

On the other hand, the pipe model is a model in which only the upper limit of traffic between provider edge devices is contracted with a customer. As a resource allocation calculation method for this model, there is a calculation method in which only the upper limit of traffic between provider edge devices is specified in a contract and a bandwidth is allocated. That is, it is a method of calculating the shortest path between provider edge devices and allocating resources on the shortest path.
A. Juttner at al., "On bandwidth efficiency of the hose resource management model in virtual private networks", IEEE INFOCOM 2003. A. Kumar et al., “Algorithms of provisioning virtual private networks in the hose model”, IEEE Trans. Networking, Aug. 2002.

  By the way, when several traffic patterns can be assumed (for example, day and night), a traffic contract expressed by combining two types of upper limits, that is, an inflow / outflow traffic upper limit to the provider edge device and an upper limit traffic between provider edge devices. It is assumed that

  However, the conventional model has an extreme premise that the traffic pattern between provider edge devices is fixed (pipe model) or completely arbitrary (hose model). That is, since the conventional model and the resource allocation method therefor are all assumed to be expressed by one type of upper limit, they are not suitable for making a contract expressed by two types of upper limit.

  Specifically, the resource allocation method for the hose model does not consider the traffic upper limit between provider edge devices for the contract expressed above. Therefore, in the hose model, the traffic between the provider edge devices does not define an upper limit value, and network resources are secured by assuming the maximum traffic that may flow from the provider edge device. Therefore, if the maximum value exceeds the actual contract value, the excess network resources become extra resources that do not need to be secured. In addition, the resource allocation method for the conventional hose model is based on a tree structure, and it is not always possible to obtain an optimal solution (a minimum resource while keeping a contract).

  On the contrary, in the pipe model, inflow / outflow traffic to the provider edge device is not taken into consideration, the inflow / outflow traffic is assumed to be the maximum value between the provider edge devices, and network resources are secured. Even in such a case, the difference between the maximum value and the contract value is also an additionally secured resource. For this reason, resources that are not used in the contract are reserved unnecessarily, and there is a problem that the utilization efficiency of network resources is low.

  Accordingly, the main object of the present invention is to calculate a network resource allocation that solves the above-described problems and enhances the utilization efficiency of network resources.

In order to solve the above problems, the present invention provides a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic volume flowing into and out of each provider edge device. A resource calculation device for calculating a resource to be allocated, the storage means for storing a network topology of the predetermined network and a profile indicating the contract, and a route calculation means for calculating a route between the provider edge devices from the network topology; And a link resource calculation means for calculating a resource to be allocated to a link on the route by solving any of a maximum flow problem and a linear programming problem to be described later .

  Accordingly, the network is calculated by calculating the resource so that the consumption resource of the entire network is reduced as much as possible in consideration of both the upper limit of traffic between provider edge devices and the upper limit of inflow / outflow traffic to the provider edge device. Resources allocated to can be reduced.

The present invention relates to the evaluation value of the previous time with respect to the evaluation value of the route that the route calculation means evaluates higher as the total value of all the links of the predetermined network is smaller for the link resource obtained by the link resource calculation means. If the candidate of the route that becomes the current evaluation value is generated from the route that becomes and the evaluation value of the candidate exceeds the previous evaluation value, the process of adopting the candidate as the route that becomes the current evaluation value is repeated. The evaluation value is compared with the current evaluation value, and the repetition is terminated when the evaluation value does not increase.

  As a result, the evaluation value may gradually increase with each calculation, but does not decrease. Thereby, since the calculation converges quickly, the number of times of repeating the calculation is reduced before the calculation is completed, and the calculation result can be obtained quickly.

The present invention relates to the evaluation value of the previous time with respect to the evaluation value of the route that the route calculation means evaluates higher as the total value of all the links of the predetermined network is smaller for the link resource obtained by the link resource calculation means. If a candidate for the route that becomes the current evaluation value is generated from the route that becomes the current evaluation value and the evaluation value of the candidate exceeds the previous evaluation value, the candidate is adopted as the route that becomes the current evaluation value. If the evaluation value of the candidate does not exceed the previous evaluation value, whether or not to adopt the candidate as a route that becomes the current evaluation value is repeatedly determined by a predetermined probability, If a given route is not continuously changed a predetermined number of times, it is determined that there is no room for further improvement of the evaluation value, and the repetition is terminated.

  As a result, the evaluation value may rise or fall every time it is calculated. As a result, even if an optimal solution (local optimal solution) in a narrow range is found, the calculation can be continued, so that there is a possibility of finding a solution that is totally optimal, and a calculation result with a high evaluation value can be obtained. .

  In the present invention, the link resource calculation means solves the maximum flow problem of (d) described later in a predetermined graph created from the network topology by the creation procedures of (a), (b), and (c) described later. The solution obtained by the above is calculated as a resource. (A) When N is the number of provider edge devices, a graph of 2N + 2 nodes connected in the order of start point node → incoming provider edge device group → outgoing provider edge device group → end point node is constructed (b) start point The bandwidth between the node and the incoming provider edge device group is the upper limit of the inflow traffic specified as the contract, and the bandwidth between the outgoing provider edge device group and the end node is the outflow specified as the contract. (C) The bandwidth of the incoming provider edge device group → the outgoing provider edge device group is set as the upper limit of the traffic amount between the provider edge devices specified as the contract. (D) Start point node → End point Maximum flow problem for nodes

  Thereby, even when the network route is not a tree configuration, the resource of each link can be calculated. Further, it is possible to calculate the resources of each link in consideration of both the traffic upper limit between provider edge devices and the upper limit of inflow / outflow amount to the provider edge device.

  In the present invention, the link resource calculation means executes an algorithm for solving a linear programming problem that optimizes a predetermined objective function indicating a resource to be allocated while satisfying a conditional expression generated based on the profile, Assuming that the number of provider edge devices is N and F (x, y) is a variable as the amount of resources to be allocated to the virtual network with i → j, the optimization of the predetermined objective function is as follows. For (x, y) passing through the link i → j, the sum of F (x, y) is maximized, and the conditional expressions are defined as (1) to (3) below. . (1) 0 ≦ F (x, y) ≦ (Upper limit traffic volume between provider edge device x and provider edge device y designated as a contract of the virtual network) (2) For any provider edge device a, F ( a, z) (0 ≦ z ≦ N) with respect to z being equal to or less than the upper limit of the inflow traffic amount to the provider edge device a designated as the contract of the virtual network (3) for any provider edge device a , F (z, a) with respect to z is less than or equal to the upper limit of the outflow traffic amount from the provider edge device a designated as the contract of the virtual network.

  Thereby, even when the network route is not a tree configuration, the resource of each link can be calculated. Further, it is possible to calculate the resources of each link in consideration of both the traffic upper limit between provider edge devices and the upper limit of inflow / outflow amount to the provider edge device.

  The present invention is characterized in that the resource calculation device further includes resource allocation means for controlling the resource calculated by the link resource calculation means to be allocated to a predetermined network.

  As a result, efficient resources calculated based on the contract can be reflected in the operation of the network.

  In the present invention, the resource calculation device determines whether the traffic related to the contract target person observes the traffic volume upper limit according to the contract, and suppresses traffic exceeding the traffic volume upper limit from being communicated. It is characterized by further including admission control means for controlling a predetermined network.

  As a result, traffic that violates the contract can be excluded from the network, so that smooth network operation is possible.

The present invention provides a resource calculation for calculating a resource allocated to a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic amount flowing into and out of each provider edge device. A method in which a computer stores a network topology of the predetermined network and a profile indicating the contract in a storage unit, a procedure for calculating a path between the provider edge devices from the network topology, and the maximum And calculating a resource to be allocated to the link on the route by solving either a flow problem or a linear programming problem .

  Accordingly, the network is calculated by calculating the resource so that the consumption resource of the entire network is reduced as much as possible in consideration of both the upper limit of traffic between provider edge devices and the upper limit of inflow / outflow traffic to the provider edge device. Resources allocated to can be reduced.

  The present invention is characterized in that resources are calculated so as to satisfy the constraints generated from two types of contracts simultaneously. Accordingly, the network is calculated by calculating the resource so that the consumption resource of the entire network is reduced as much as possible in consideration of both the upper limit of traffic between provider edge devices and the upper limit of inflow / outflow traffic to the provider edge device. It is possible to reduce the useless amount while securing the minimum necessary resources to be allocated to the. Therefore, it is possible to calculate network resource allocation that enhances network resource utilization efficiency.

  Next, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 shows a network configuration to which an embodiment of the present invention is applied. This network enables communication only between the customer edge devices CE of the same customer while accommodating one or more customers in the same network. Here, the customer may be an administrator different from the network administrator or the same administrator.

  Therefore, the network shown in FIG. 1 is one in which one provider network and two virtual networks (# 1, # 2) are connected. A device belonging to one virtual network can communicate with a device belonging to the same virtual network, but cannot communicate with a device belonging to another virtual network. This means that the virtual network is VPN. When devices belonging to the same virtual network exist at positions physically separated from each other, communication between these devices is performed via the provider network. The main feature of the present embodiment is to propose a virtual network resource allocation method that keeps the contract for each virtual network and reduces the resources to be secured in the entire network as much as possible.

  The CE (Customer Edge) in FIG. 1 means the customer edge device CE. The customer edge device CE is owned by a customer and is connected to one or more PEs (Provider Edge). This PE means a provider edge device PE. The provider edge device PE is owned by the provider and is connected to one or more customer edge devices CE. P (Provider) means the provider device P. The provider device P is owned by the provider and is connected to the provider edge device PE or another provider device P. The link exists between the customer edge device CE-provider edge device PE, provider edge device PE-provider device P, provider edge device PE-provider edge device PE, and serves as an information transfer medium. The provider network composed of the provider edge device PE, the provider device P, and the link enables communication only between the customer edge devices CE in the same virtual network.

  As the customer edge device CE, provider edge device PE, and provider device P, for example, router, L2SW (Layer 2 Switch), TDM (Time Division Multiplexing) cross-connect, wavelength cross-connect, MPLS (MultiProtocol Label Switching) LSR (Label) Switch Router), etc.

  FIG. 2 shows the configuration of the resource calculation device. Note that the resource calculation device is configured as a computer including at least a memory serving as a storage unit used when performing arithmetic processing, and an arithmetic processing device that performs the arithmetic processing. The memory is constituted by a RAM (Random Access Memory) or the like. Arithmetic processing is realized by an arithmetic processing unit constituted by a CPU (Central Processing Unit) executing a program on a memory.

  As an example of the contract for each virtual network, the upper limit of the inflow traffic amount indicates the upper limit of the inflow traffic amount to the provider edge device PE. The outflow traffic amount upper limit indicates the upper limit of the outflow traffic amount from the provider edge device PE. The inflow traffic amount indicates the current inflow traffic amount to the provider edge device PE.

  As another example of the contract for each virtual network, the inter-PE traffic volume upper limit indicates the upper limit of the traffic volume between the provider edge devices PE. Note that PE-ID → PE-ID indicates a pair of provider edge apparatuses PE in which resources should be secured, and → indicates a direction in which traffic flows. For example, when A → B, traffic flows from A to B.

  Further, as the current traffic volume, the outflow traffic volume indicates the current outflow traffic volume from the provider edge apparatus PE. The inter-PE traffic volume indicates the current traffic volume between the provider edge apparatuses PE.

  The network topology DB 22 stores provider network topology information. The link ID is information that uniquely identifies the link. The upstream device ID is information that uniquely indicates a device on the upstream side of the link (a device that is a partner into which traffic flows in as viewed from a certain device). The downstream device ID is information that uniquely indicates a device on the downstream side of the link (a device that is a partner that flows out traffic as viewed from a certain device). In this network topology DB 22, all devices constituting the network and all links between these devices are defined. Based on the defined information, the network topology shown in FIG. 1 is illustrated. be able to.

  The path DB 24 stores paths between provider edge devices PE for each virtual network. VPN-ID is information that uniquely identifies a virtual network. Like the profile DB 26, PE-ID → PE-ID indicates a pair of provider edge apparatuses PE in which resources should be secured, and → indicates the direction of traffic flow. The link ID list is a list of link IDs and indicates a route. In the example of FIG. 1, when the route of PE1 → PE2 is PE1 → P1 → PE2, the link ID list is <PE1, P1, PE2>.

  The profile DB 26 stores a contract for each virtual network and the current traffic volume. VPN-ID is information that uniquely identifies a virtual network. The PE-ID is information for uniquely identifying the provider edge device PE, and corresponds to, for example, an IP address.

  The link resource DB 28 stores the allocated resource amount in each link for each virtual network. VPN-ID is information that uniquely identifies a virtual network. The link ID is information that uniquely identifies the link. The resource amount indicates an allocated resource amount in the link of the virtual network.

  When given a VPN-ID indicating a virtual network to be calculated, the route calculation means 12 calculates a route between provider edge devices PE in the virtual network. When given a VPN-ID indicating a virtual network to be calculated, the resource allocation means 16 instructs to calculate a route between provider edge devices PE in the virtual network, stores a value in the route DB 24, and An instruction is given to calculate the allocated resource amount in each link, and a value is stored in the link resource DB 28.

  When there is a resource increase request between provider edge devices PE for a certain virtual network, the reception control means 18 is within the upper limit of the inflow / outflow traffic amount to the provider edge device PE and between the provider edge devices PE If the traffic amount is within the upper limit of the traffic amount, the increase of the resource is permitted, the traffic amount of the profile DB 26 is updated, and if there is a resource reduction request between the provider edge devices PE, the traffic amount of the profile DB 26 is updated. By doing so, admission control is performed according to the contract. The updating of the traffic volume is a process of rewriting the contents (traffic volume) of the profile DB 26 database. As a specific process of the reception processing means 18, the process of excluding traffic that violates the contract is, for example, when a user whose contract is 10 Mbps flows 15 Mbps of traffic, traffic for 5 Mbps that violates the contract is networked. In this process, the network transfer apparatus is instructed to be excluded from communication targets.

  The resource allocation unit 16 is a unit that allocates the link resource calculated by the link resource calculation unit 14 to the network, and is realized by, for example, SNMP (Simple Network Management Protocol).

  The link resource calculation means 14 calculates the allocated resource amount in the link of the virtual network, given the VPN-ID indicating the virtual network to be calculated and the link ID indicating the link to be calculated. The configuration of the link resource calculation unit 14 will be described in detail later.

  Hereinafter, two examples of the resource amount calculation method performed by the link resource calculation unit 14 will be described. For the sake of explanation, the contract of the virtual network # 1 of FIG. 1 is shown in FIGS. 7 (a) and 7 (b).

  FIG. 3 shows a first resource amount calculation method. That is, the network configuration of the number of nodes 2N + 2 obtained by the link resource calculation unit 14 with respect to the contract shown in FIGS. 7A and 7B is shown. The link symbols (I1, I2, I3, O1, O2, and O3) correspond to the resource amounts shown in FIG. For the node 2-6, node 2-7, node 3-6, node 3-7, node 4-5, and node 4-6, when the path between the provider edge devices PE passes the attention link, If a link exists but does not pass, it does not exist. For example, if the link of interest is PE1-P1 in FIG. 1, when the route PE1 → PE2 passes through the link of interest, a link exists between the nodes 2-6 in FIG. The correspondence between the nodes in FIG. 3 and the nodes in FIG. 1 is as shown in FIG. For example, the node 2-3 in FIG. 3 corresponds to PE1 → PE2 in FIG.

  In the first resource amount calculation method, the resource amount in the link is calculated by solving the maximum flow problem in the network thus obtained. As a method of solving the maximum flow problem, there are a labeling method and a Preflow-Push method.

  FIG. 8 shows a second resource amount calculation method. In the second method, the amount of resources in the link is calculated by solving the linear programming problem shown in FIG. Note that the linear programming problem can be solved by, for example, the simplex method (single method). The resource amount (for example, bandwidth) to be calculated is the value of the objective function in FIG.

  Next, a route calculation method performed by the route calculation means 12 will be described with reference to the flowcharts shown in FIGS. 4, 5, and 6. These flowcharts show an example of a heuristic method.

  FIG. 4 shows a flowchart in which a procedure for calculating a route between provider edge devices PE is performed by a greedy algorithm. This greedy algorithm is an improved algorithm that does not calculate the data with the highest evaluation value from the beginning, but repeats the calculation so as to gradually increase the evaluation value from the data with the first evaluation value. Then, from the data that becomes the previous evaluation value, a candidate for the data that becomes the current evaluation value is generated, and when the candidate's evaluation value exceeds the previous evaluation value, the candidate is adopted as the data that becomes the current evaluation value To do. Furthermore, when the previous evaluation value and the current evaluation value are compared and the evaluation value does not increase, the calculation is terminated because there is no room for further improvement of the evaluation value.

  For this reason, the evaluation value may gradually increase with each calculation, but does not decrease. Thereby, since the calculation converges quickly, the number of times of repeating the calculation is reduced until the calculation is completed, and the calculation result can be obtained quickly.

  The link resource calculation means 14 calculates necessary link resources for each link. This value is R (1) (S101). The link resource calculation unit 14 obtains the sum of all the link resources obtained in S101. This value is R (S102). The route calculation means 12 inputs a large value for the variable MIN (S103).

  The link resource calculation means 14 pays attention to one of the paths between the provider edge devices PEsrc → dst whose traffic upper limit between the provider edge devices PE is larger than 0 (S104). The link resource calculation means 14 calculates the necessary link resource when the traffic volume upper limit between the provider edge devices PE is set to 0 for each link. This value is R ′ (l) (S105). The route calculation means 12 sets [R (l) −R ′ (l)] × X + 1 (X: constant) as the link cost for each link (S106). The route calculation means 12 calculates the shortest route between the provider edge device PEsrc → dst based on the link cost (S107). The link resource calculation unit 14 calculates necessary resources for each link assuming that the path between the provider edge devices PEsrc → dst is replaced with the path obtained in S107 (S108). The link resource calculation means 14 calculates the sum of the obtained link resources for all links. This value is set as R ′ (S109).

  The route calculation means 12 confirms whether R′−R <0 and R ′ <MIN (S110). The route calculation means 12 sets MIN = R ′ when S110 is YES (S111). The route calculation means 12 confirms whether the calculation has been completed for all provider edge devices PE (S120). The route calculation means 12 returns to S104 if NO in S120. The route calculation means 12 confirms whether the MIN remains the large value set in S103 (S130). When S130 is NO (S130, N), the route calculation means 12 replaces the route between the provider edge devices PE to be MIN, and returns to S101 (S131). When S130 is YES (S130, Y), the route calculation unit 12 ends the route calculation process.

  FIG. 5 and FIG. 6 show flowcharts in which the procedure for calculating the path between the provider edge devices PE is performed by simulated annealing. This method is an improved algorithm that does not calculate the data with the highest evaluation value from the beginning, but repeats the calculation so as to gradually increase the evaluation value from the data with the first evaluation value. Then, a candidate for data that becomes the current evaluation value is generated from the data that becomes the previous evaluation value. If the evaluation value of the candidate is higher than the previous evaluation value, the candidate is adopted as data that becomes the current evaluation value (see S220 and S221). On the other hand, if the evaluation value of the candidate does not exceed the previous evaluation value, whether or not to adopt the candidate as data that becomes the current evaluation value is determined by a predetermined probability (see S224 and S225). . Further, when the data that becomes the evaluation value is not continuously changed a predetermined number of times, the calculation is terminated assuming that there is no room for further improvement of the evaluation value (see S240).

  Therefore, the evaluation value may increase or decrease every time calculation is performed. As a result, even if an optimal solution in a narrow range (local optimal solution) is found, calculation can be continued, so that there is a possibility of finding an overall optimal solution, and a calculation result with a high evaluation value can be obtained. .

  The route calculation means 12 sets c_a = 0, c_r = 0, c_c = 0, c_s = 0, and T = T0 (T0: constant) (S201). The link resource calculation means 14 calculates a necessary link resource for each link, and sets this value as R (1) (S202). The link resource calculation means 14 obtains the sum of the link resources obtained in S202 for all links, and sets this value as R (S203).

  The route calculation means 12 randomly selects the link i → j (S204). The route calculation means 12 randomly selects between the provider edge devices PEsrc → dst whose route passes the link i → j (S205). The link resource calculation means 14 calculates the link resource of each necessary link when the traffic upper limit between the provider edge devices PEsrc → dst is set to zero. This value is R ′ (l) (S206). The route calculation means 12 sets the link cost for each link as [R (l) −R ′ (l)] × X + 1. However, the link cost of the link i → j selected in S204 is set to a large value (S207). The route calculation means 12 calculates the shortest route between the provider edge device PEsrc → dst selected in S205 based on the link cost obtained in S207 (S208). The link resource calculation means 14 calculates necessary resources for each link assuming that the path between the provider edge devices PEsrc → dst has been replaced with the path obtained in S208 (S209).

  The route calculation means 12 increases the value of c_c by 1, and obtains the sum of all link resources obtained in S209. This value is set as R ′ (S210). When the value of R′−R is negative (S220, Y), the route calculation unit 12 replaces the route, increases the value of c_a by 1, and sets the values of c_s and c_r to 0 (S221). When the value of R′−R is 0 (S222, Y), the route calculation unit 12 replaces the route, increases the values of c_a and c_s by 1, and sets the value of c_r to 0 (S223). When the value of R′−R is positive, the route calculation unit 12 determines whether to replace the route with a probability e (= (− R ′ + R) / T) (S224). When replacing the route (S224, Y), the route calculation unit 12 increments the value of c_a by 1, and sets the values of c_s and c_r to 0 (S225). If the route is not replaced (S224, N), the route calculation means 12 increases the value of c_r by 1 (S226). In the case of c_a = C_A (C_A: constant) or c_c = C_C (C_C: constant) (S230, Y), the route calculation means 12 sets T = α × T (α: constant) and c_a = 0, c_c = 0 is set (S231). The route calculation means 12 returns to S202 if c_r = C_R (C_R: constant) or c_s = C_S (C_S: constant) is not satisfied (S240, N), but ends if it is (S240, Y). As a result, the calculated route may be totally optimized.

  The present invention described above can be widely modified without departing from the spirit thereof as follows.

  For example, the resource calculation device shown in FIG. 2 may be configured as an independent device, or may be accommodated in PEs and Ps. Furthermore, the resource calculation device may have a function to transfer traffic (router function) in the network to be calculated, or control to perform resource calculation and control the network based on the result. It may be configured as a device.

  Furthermore, in the embodiment shown in FIG. 1, the network is divided into two parts, the provider network and the virtual network, but this is only an example. In other words, the present invention is not limited to VPN, and can be applied to any network in which a traffic volume upper limit is specified by a contract.

1 shows a network configuration according to an embodiment of the present invention. 1 shows a configuration of a resource calculation apparatus according to an embodiment of the present invention. The link resource calculation means concerning one Embodiment of this invention shows the structure of the network of the number of nodes 2N + 2 used when calculating a link resource. The flowchart which performs the procedure which calculates the path | route between the provider edge apparatuses regarding one Embodiment of this invention by a greedy algorithm is shown. The flowchart which performs the procedure which calculates the path | route between the provider edge apparatuses regarding one Embodiment of this invention by simulated annealing is shown. The flowchart which performs the procedure which calculates the path | route between the provider edge apparatuses regarding one Embodiment of this invention by simulated annealing is shown. It is a figure explaining the maximum flow problem regarding one Embodiment of this invention. FIG. 4 illustrates a linear programming problem for one embodiment of the present invention.

Explanation of symbols

PE provider edge device 12 route calculation means 14 link resource calculation means 16 resource allocation means 18 admission control means 22 network topology DB
24 Route DB
26 Profile DB
28 Link Resource DB

Claims (8)

A resource calculation device for calculating resources allocated to a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic amount flowing into and out of each provider edge device. ,
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. When the candidate for the route that becomes the current evaluation value is generated and the evaluation value of the candidate exceeds the previous evaluation value, the process of adopting the candidate as the route that becomes the current evaluation value is repeated, When the evaluation value does not increase by comparing with the evaluation value this time,
The link resource calculation means obtains a solution obtained by solving a maximum flow problem (d) described later in a predetermined graph created from the network topology according to a procedure (a) (b) (c) described later. Is calculated as a resource.
(A) When N is the number of provider edge devices, a graph of 2N + 2 nodes connected in the order of start point node → incoming provider edge device group → outgoing provider edge device group → end point node is constructed.
(B) The bandwidth between the starting node → the incoming provider edge device group is set as the upper limit of the inflow traffic specified as the contract, and the bandwidth between the outgoing provider edge device group → the end node is set as the contract. Specified spill traffic upper limit
(C) The bandwidth of the provider edge device group on the input side → the band of the provider edge device group on the output side is set as the traffic volume upper limit between the provider edge devices designated as the contract.
(D) Maximum flow problem from start node to end node A resource calculation device for calculating resources allocated to a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic amount flowing into and out of each provider edge device. ,
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to the links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. When the candidate for the route that becomes the current evaluation value is generated and the evaluation value of the candidate exceeds the previous evaluation value, the process of adopting the candidate as the route that becomes the current evaluation value is repeated, When the evaluation value does not increase compared with the evaluation value this time, the repetition is terminated.
The link resource calculation means executes an algorithm for solving a linear programming problem that optimizes a predetermined objective function indicating a resource to be allocated while satisfying a conditional expression generated based on the profile,
Assuming that the number of provider edge devices is N and F (x, y) is a variable as the amount of resources to be allocated to the virtual network by link i → j, the optimization of the predetermined objective function is the path between provider edge devices x → y. , The total of F (x, y) is maximized for (x, y) passing through the link i → j, and the conditional expressions are the following (1) to (3). Resource calculation device.
(1) 0 ≦ F (x, y) ≦ (Upper limit traffic amount between provider edge device x and provider edge device y designated as a contract of the virtual network)
(2) For any provider edge device a, the sum of z of F (a, z) (0 ≦ z ≦ N) is less than or equal to the upper limit of the inflow traffic amount to the provider edge device a designated as the contract of the virtual network To become
(3) For any provider edge device a, the sum of F (z, a) with respect to z is less than or equal to the upper limit of outflow traffic from the provider edge device a designated as the contract of the virtual network. A resource calculation device for calculating resources allocated to a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic amount flowing into and out of each provider edge device. ,
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. If a candidate for the route that will be the current evaluation value is generated and the evaluation value of the candidate is higher than the previous evaluation value, the candidate is adopted as the route that will be the current evaluation value. If the evaluation value does not exceed the previous evaluation value, whether or not the candidate is adopted as the route that becomes the current evaluation value is determined by repeating a process determined by a predetermined probability, and the route that becomes the evaluation value is predetermined. If it is not changed continuously for the number of times, it is assumed that there is no room for further improvement of the evaluation value, and the iteration is terminated.
The link resource calculation means obtains a solution obtained by solving a maximum flow problem (d) described later in a predetermined graph created from the network topology according to a procedure (a) (b) (c) described later. Is calculated as a resource.
(A) When N is the number of provider edge devices, a graph of 2N + 2 nodes connected in the order of start point node → incoming provider edge device group → outgoing provider edge device group → end point node is constructed.
(B) The bandwidth between the starting node → the incoming provider edge device group is set as the upper limit of the inflow traffic specified as the contract, and the bandwidth between the outgoing provider edge device group → the end node is set as the contract. Use specified spill traffic upper limit
(C) The bandwidth of the provider edge device group on the input side → the band of the provider edge device group on the output side is set as the traffic volume upper limit between the provider edge devices designated as the contract.
(D) Maximum flow problem from start node to end node A resource calculation device for calculating resources allocated to a network in a predetermined network subject to a contract defined by an upper limit of traffic volume between provider edge devices and an upper limit of traffic amount flowing into and out of each provider edge device. ,
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. If a candidate for the route that will be the current evaluation value is generated and the evaluation value of the candidate is higher than the previous evaluation value, the candidate is adopted as the route that will be the current evaluation value. If the evaluation value does not exceed the previous evaluation value, whether or not the candidate is adopted as the route that becomes the current evaluation value is determined by repeating a process determined by a predetermined probability, and the route that becomes the evaluation value is predetermined. If it is not changed continuously for the number of times, it is assumed that there is no room for further improvement of the evaluation value, and the iteration is terminated.
The link resource calculation means executes an algorithm for solving a linear programming problem that optimizes a predetermined objective function indicating a resource to be allocated while satisfying a conditional expression generated based on the profile,
Assuming that the number of provider edge devices is N and F (x, y) is a variable as the amount of resources to be allocated to the virtual network by link i → j, the optimization of the predetermined objective function is the path between provider edge devices x → y. , The total of F (x, y) is maximized for (x, y) passing through the link i → j, and the conditional expressions are the following (1) to (3). Resource calculation device.
(1) 0 ≦ F (x, y) ≦ (Upper limit traffic amount between provider edge device x and provider edge device y designated as a contract of the virtual network)
(2) For any provider edge device a, the sum of z of F (a, z) (0 ≦ z ≦ N) is less than or equal to the upper limit of the inflow traffic amount to the provider edge device a designated as the contract of the virtual network To become
(3) For any provider edge device a, the sum of F (z, a) with respect to z is less than or equal to the upper limit of outflow traffic from the provider edge device a designated as the contract of the virtual network. Resource calculation by a resource calculation device that calculates resources allocated to a network in a predetermined network subject to a contract defined by the upper limit of traffic volume between provider edge devices and the upper limit of traffic amount flowing into and out of each provider edge device A method,
The resource calculation device includes:
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. When the candidate for the route that becomes the current evaluation value is generated and the evaluation value of the candidate exceeds the previous evaluation value, the process of adopting the candidate as the route that becomes the current evaluation value is repeated, When the evaluation value does not increase by comparing with the evaluation value this time,
The link resource calculation means obtains a solution obtained by solving a maximum flow problem (d) described later in a predetermined graph created from the network topology according to a procedure (a) (b) (c) described later. A resource calculation method characterized by calculating as a resource.
(A) When N is the number of provider edge devices, a graph of 2N + 2 nodes connected in the order of start point node → incoming provider edge device group → outgoing provider edge device group → end point node is constructed.
(B) The bandwidth between the starting node → the incoming provider edge device group is set as the upper limit of the inflow traffic specified as the contract, and the bandwidth between the outgoing provider edge device group → the end node is set as the contract. Specified spill traffic upper limit
(C) The bandwidth of the provider edge device group on the input side → the band of the provider edge device group on the output side is set as the traffic volume upper limit between the provider edge devices designated as the contract.
(D) Maximum flow problem from start node to end node Resource calculation by a resource calculation device that calculates resources allocated to a network in a predetermined network subject to a contract defined by the upper limit of traffic volume between provider edge devices and the upper limit of traffic amount flowing into and out of each provider edge device A method,
The resource calculation device includes:
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. When the candidate for the route that becomes the current evaluation value is generated and the evaluation value of the candidate exceeds the previous evaluation value, the process of adopting the candidate as the route that becomes the current evaluation value is repeated, When the evaluation value does not increase compared with the evaluation value this time, the repetition is terminated.
The link resource calculation means executes an algorithm for solving a linear programming problem that optimizes a predetermined objective function indicating a resource to be allocated while satisfying a conditional expression generated based on the profile,
Assuming that the number of provider edge devices is N and F (x, y) is a variable as the amount of resources to be allocated to the virtual network by link i → j, the optimization of the predetermined objective function is the path between provider edge devices x → y. , The total of F (x, y) is maximized for (x, y) passing through the link i → j, and the conditional expressions are the following (1) to (3). Resource calculation method.
(1) 0 ≦ F (x, y) ≦ (Upper limit traffic amount between provider edge device x and provider edge device y designated as a contract of the virtual network)
(2) For any provider edge device a, the sum of z of F (a, z) (0 ≦ z ≦ N) is less than or equal to the upper limit of the inflow traffic amount to the provider edge device a designated as the contract of the virtual network To become
(3) For any provider edge device a, the sum of F (z, a) with respect to z is less than or equal to the upper limit of outflow traffic from the provider edge device a designated as the contract of the virtual network. Resource calculation by a resource calculation device that calculates resources allocated to a network in a predetermined network subject to a contract defined by the upper limit of traffic volume between provider edge devices and the upper limit of traffic amount flowing into and out of each provider edge device A method,
The resource calculation device includes:
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. If a candidate for the route that will be the current evaluation value is generated and the evaluation value of the candidate is higher than the previous evaluation value, the candidate is adopted as the route that will be the current evaluation value. If the evaluation value does not exceed the previous evaluation value, whether or not the candidate is adopted as the route that becomes the current evaluation value is determined by repeating a process determined by a predetermined probability, and the route that becomes the evaluation value is predetermined. If it is not changed continuously for the number of times, it is assumed that there is no room for further improvement of the evaluation value, and the iteration is terminated.
The link resource calculation means obtains a solution obtained by solving a maximum flow problem (d) described later in a predetermined graph created from the network topology according to a procedure (a) (b) (c) described later. A resource calculation method characterized by calculating as a resource.
(A) When N is the number of provider edge devices, a graph of 2N + 2 nodes connected in the order of start point node → incoming provider edge device group → outgoing provider edge device group → end point node is constructed.
(B) The bandwidth between the starting node → the incoming provider edge device group is set as the upper limit of the inflow traffic specified as the contract, and the bandwidth between the outgoing provider edge device group → the end node is set as the contract. Specified spill traffic upper limit
(C) The bandwidth of the provider edge device group on the input side → the band of the provider edge device group on the output side is set as the traffic volume upper limit between the provider edge devices designated as the contract.
(D) Maximum flow problem from start node to end node Resource calculation by a resource calculation device that calculates resources allocated to a network in a predetermined network subject to a contract defined by the upper limit of traffic volume between provider edge devices and the upper limit of traffic amount flowing into and out of each provider edge device A method,
The resource calculation device includes:
Storage means for storing a network topology of the predetermined network and a profile indicating the contract;
A route calculation means for calculating a route between the provider edge devices from the network topology;
Link resource calculation means for calculating resources to be allocated to links on the route so as to satisfy the constraint defined by the contract ,
For the link resource obtained by the link resource calculation unit, the route calculation unit determines the evaluation value of a route that is highly evaluated as the total value of all links of the predetermined network is small. If a candidate for the route that will be the current evaluation value is generated and the evaluation value of the candidate is higher than the previous evaluation value, the candidate is adopted as the route that will be the current evaluation value. If the evaluation value does not exceed the previous evaluation value, whether or not the candidate is adopted as the route that becomes the current evaluation value is determined by repeating a process determined by a predetermined probability, and the route that becomes the evaluation value is predetermined. If it is not changed continuously for the number of times, it is assumed that there is no room for further improvement of the evaluation value, and the iteration is terminated.
The link resource calculation means executes an algorithm for solving a linear programming problem that optimizes a predetermined objective function indicating a resource to be allocated while satisfying a conditional expression generated based on the profile,
Assuming that the number of provider edge devices is N and F (x, y) is a variable as the amount of resources to be allocated to the virtual network by link i → j, the optimization of the predetermined objective function is the path between provider edge devices x → y. , The total of F (x, y) is maximized for (x, y) passing through the link i → j, and the conditional expressions are the following (1) to (3). Resource calculation method.
(1) 0 ≦ F (x, y) ≦ (Upper limit traffic amount between provider edge device x and provider edge device y designated as a contract of the virtual network)
(2) For any provider edge device a, the sum of z of F (a, z) (0 ≦ z ≦ N) is less than or equal to the upper limit of the inflow traffic amount to the provider edge device a designated as the contract of the virtual network To become
(3) For any provider edge device a, the sum of F (z, a) with respect to z is less than or equal to the upper limit of outflow traffic from the provider edge device a designated as the contract of the virtual network.

Images