Network design device, network design method and network design program

Abstract

To obtain a larger network capacity by dividing intermediate switches into three or more groups to enhance the flexibility of a network configuration in a Clos network.SOLUTION: A plurality of intermediate switches are divided into S groups, and in each step of S groups, an intermediate switch allocated to the step is used. In the case that the number of terminals connected on an input side or an output side of an input / output switch requested for connection exceeds an upper limit value of the previous step to be equal to or less than an upper limit value of the step, a calculation unit 12 of a network design device 1 calculates the number n of terminals, the number k of input / output switches, the number of intermediate switches m1, m2, ..., mS in each group, and the number of links v1, v2, ..., vS between one input / output switch and one intermediate switch so as to use any intermediate switch corresponding to the step for connection.SELECTED DRAWING: Figure 5

Description

Technical Field

[0001] The present disclosure relates to a network design apparatus, a network design method, and a network design program.

Background Art

[0002] There is an MPP network with a better accommodation efficiency than a conventional MPP network composed of a Clos network and having a non-blocking network structure (Patent Document 1). Patent Document 1 discloses an MPP network having a configuration in which input / output ports m and connection ports n to an intermediate MPP are installed on both sides of an outer MPP. Patent Document 1 includes a parameter input unit for inputting the required number r of connection ports and the size N of the MPP, and uses the input r and N to perform an exhaustive search for the number k of outer MPPs, the number l of intermediate MPPs, m, and n that minimize the number (k + l) of MPPs while satisfying predetermined conditions.

[0003] In a Clos network, there is a technique for theoretically guaranteeing the blocking rate of a path, enhancing the flexibility of network configuration, and obtaining a larger network capacity (Non-Patent Document 1). Non-Patent Document 1 discloses a technique of dividing intermediate switches into two groups and sequentially using the intermediate switches in each group.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Non-Patent Documents

[0005]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] Non-Patent Document 1 discloses a case where intermediate switches are divided into two groups. However, when they are divided into three or more groups, it is expected that the degree of freedom will be further increased and a larger network capacity can be obtained.

[0007] This disclosure has been made in view of the above circumstances, and the object of this disclosure is to provide a technology capable of increasing the flexibility of network configuration and obtaining a larger network capacity by dividing intermediate switches into three or more groups in a Clos network.

Means for Solving the Problems

[0008] The network design device according to one aspect of the present disclosure includes a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, and designs a network that satisfies a predetermined network capacity. The network design device includes the number of ports of the switches, the total number of the plurality of input / output switches and the plurality of intermediate switches including the input ports of the number of ports and the output ports of the number of ports, the terminal utilization rate of the terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches to be impossible, and an acquisition unit that acquires the number of groups of intermediate switches (≧3), and in a network that satisfies the number of ports, the total number, and the terminal utilization rate, when the number of terminals is increased one by one from the maximum value of the number of terminals when the network is non-blocking, the total probability of each blocking occurrence probability does not exceed the allowable blocking rate, and a calculation unit that calculates the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one of the input / output switches and one of the intermediate switches so that the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized. The calculation unit divides the plurality of intermediate switches into S groups, and in each of the S steps, uses the intermediate switch assigned to that step, and when the number of terminals being connected on the input side or the output side of the input / output switch for which connection is requested exceeds the upper limit value of the previous step and is equal to or less than the upper limit value of the current step, uses any of the intermediate switches corresponding to the current step to make a connection, calculate the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group.

[0009] The network design method according to one aspect of the present disclosure includes a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, and designs a network that satisfies a predetermined network capacity. The network design device is a computer that obtains the number of ports of the switches, the total number of the plurality of input / output switches and the plurality of intermediate switches including the input ports and the output ports of the number of ports, the terminal utilization rate of the terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches to be unavailable, and the number of groups of intermediate switches (≧3), and in a network that satisfies the number of ports, the total number, and the terminal utilization rate, when the network becomes non-blocking, the total probability of each blocking occurrence probability when the number of terminals is increased one by one from the maximum value of the number of terminals does not exceed the allowable blocking rate, and calculates the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one of the input / output switches and one of the intermediate switches so that the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized. When calculating, the plurality of intermediate switches are divided into S groups, and in each of the S steps, the intermediate switches assigned to that step are used, and when the number of terminals being connected on the input side or the output side of the input / output switch for which a connection request has been made exceeds the upper limit value of the previous step and is equal to or less than the upper limit value of the current step, the computer uses any of the intermediate switches corresponding to the current step to make the connection, and calculates the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group.

[0010] The network design program according to one aspect of the present disclosure includes a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, and a computer for designing a network that satisfies a predetermined network capacity, the number of ports of the switches, the total number of the plurality of input / output switches and the plurality of intermediate switches each having an input port and an output port of the number of ports, the terminal utilization rate of the terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches to be blocked, and an acquisition unit that acquires the number of groups of intermediate switches (≧3); in a network that satisfies the number of ports, the total number, and the terminal utilization rate, when the network is non-blocking, the total probability of each blocking occurrence probability when the number of terminals is increased one by one from the maximum value of the number of terminals, does not exceed the allowable blocking rate, and a calculation unit that calculates the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one of the input / output switches and one of the intermediate switches so that the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized. The calculation unit divides the plurality of intermediate switches into S groups, and in each of the S steps, uses the intermediate switch assigned to that step, and when the number of terminals being connected on the input side or the output side of the input / output switch for which a connection is requested exceeds the upper limit value of the previous step and is less than or equal to the upper limit value of that step, calculates the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group so as to connect using any of the intermediate switches corresponding to that step.

Advantages of the Invention

[0011] According to the present disclosure, in a Clos network, it is possible to provide a technology that divides intermediate switches into three or more groups to enhance the flexibility of the network configuration and obtain a larger network capacity.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Embodiments for Carrying Out the Invention

[0013] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the description of the drawings, the same reference numerals are given to the same parts and the description thereof is omitted.

[0014] Referring to Figure 1, an example of path setting will be described. In a data center, a plurality of racks installed in the data center are accommodated in the network, and a communication line (path) is set by communicably connecting a specified pair of racks. A rack houses switches, servers, etc.

[0015] In a communication building, multiple transmission devices are accommodated in a network and connected to transmission devices in other buildings or other operators. Currently, packet switching technologies such as IP and Ethernet are used in transmission devices, but in the future, circuit switching technologies such as optical communication will be used to improve transmission capacity and power efficiency.

[0016] Referring to Figure 2, the path closure situation will be described. In circuit switching, if there is no unused path between the pair of terminals that have requested a connection, the path cannot be set. This situation is called blockage. Between the second port P12 of the first switch S1 and the fourth port P34 of the third switch S3, a path is to be set. However, due to the two previously set paths (dashed lines), there is no free path, so the desired path cannot be set.

[0017] This disclosure describes a network design technique that maximizes the network capacity (the number of accommodated switches and transmission devices) while satisfying the allowable blockage rate when the allowable blockage rate is given. In particular, it targets the Clos network with the Clos structure widely used in data centers.

[0018] Referring to Figure 3, a general Clos network will be described. The Clos network is composed of three layers: an input layer, an intermediate layer, and an output layer. However, the input layer and the output layer are combined into one input-output layer, and the structure is configured to fold back at the intermediate layer.

[0019] (Conventional method) Referring to Figure 4, a construction example of the Clos network according to Non-Patent Document 1 will be described. Terminals such as racks and transmission devices are connected to the input-output switches in the input-output layer. Terminal t is a transmitting terminal, and terminal r is a receiving terminal.

[0020] The Clos network shown in Fig. 4 has a switches. Each switch is an N×N switch with N ports on each of the input and output sides. Let the number of input / output switches in the input / output layer be k. To each input / output switch, n terminals are connected as inputs and outputs respectively. Let the number of intermediate switches be m. Starting from the input / output switches in order, each is given a number S1, …, S k , S k+1 …S k+m and so on.

[0021] In such a Clos network, a solution is sought to design a network with the largest possible capacity. The network capacity is taken as the total number of terminals nk in the network. Here, the inputs and outputs are counted separately. However, the blocking probability must be no more than a specified value ε. The utilization rate p of each terminal is assumed to be known. Note that the utilization rate p of the terminal is estimated from past service records and the like.

[0022] The intermediate switches are divided into two groups according to the number of links v1, v2 to the input / output switches. The first group (First-step group) has m1 intermediate switches. In the first group, each intermediate switch has v1 links to each input / output switch. The second group (Second-step group) has m2 intermediate switches. In the second group, each intermediate switch has v2 links to each input / output switch. Non-Patent Document 1 uses these two groups as follows.

[0023] In the first step, when the number of terminals n connected to either the input or output side of the input / output switch for a connection request is no more than the number of terminals n1 snb units, any one of the m1 intermediate switches included in the first group is used for the connection. The number of terminals n1 snb is obtained as part of the solution. The probability of successful connection in the first step is no more than the allowable blocking probability ε. The probability of failed connection in the first step is no more than a predetermined blocking probability ε1 (ε ≤ ε1 ≤ 1). SNB is the abbreviation of strictly nonblocking, meaning no blocking occurs.

[0024] In the second step, when the number of terminals n connected to the input / output switch on either the input side or the output side of the connection request exceeds n1 snb by "n1 snb + n2 snb " and is equal to or less than "n1 + n2", any one of the m2 intermediate switches included in the second group is further used to make the connection. The probability of being unable to make the connection even in the second step is equal to or less than the allowable blocking rate ε.

[0025] When none of the above conditions are met, the connection request is rejected. The rejection probability is equal to or less than the allowable blocking rate ε.

[0026] The ILP (Integer Linear Programming) of Non-Patent Document 1 is expressed by Formula (1a) to Formula (1m).

[0027]

Number

[0028] The following are given as problems. · The number of ports N of the switch · The total number a of input / output switches and intermediate switches · The terminal utilization rate p · The allowable blocking rate ε

[0029] The following are the decision variables of the optimization problem. · The number of terminals n · The number of input / output switches k · The number of intermediate switches m1 in the first step · The number of intermediate switches m2 in the second step · The number of links v1 from one intermediate switch in the first step to one input / output switch · The number of links v2 from one intermediate switch in the second step to one input / output switch · The upper limit value n1 of the number of terminals connected to the input / output switch for which a connection is requested in the first step snb · The upper limit value n2 of the number of terminals connected to the input / output switch for which a connection is requested in the second step snb

[0030] ε1 is an intermediate variable required only for the solution. Equations (1j)-(1m) define the domain of the variables.

[0031] Equation (1a) is the objective function, which maximizes the network capacity. Equation (1b) is a condition regarding the number of ports of the input / output switch. Equations (1c)-(1d) are conditions regarding the number of ports of the intermediate switch.

[0032] Equations (1e)-(1f) are constraints regarding the first step. From Equation (1f), the number of used ports is n1 snb If it is as follows, it is guaranteed that no congestion will occur. According to Equation (1e), the number of used ports is n1 snb If it is more, it is guaranteed that the sum of the congestion rates does not exceed ε1.

[0033] Equations (1g)-(1h) are constraints regarding the second step. It is guaranteed that the sum of the congestion rates does not exceed ε according to Equation (1g). Note that when m2 = 0, the second step is omitted, so ε is also described in Equation (1e) as the upper limit of the congestion rate corresponding to m2 = 0.

[0034] Explain the outline of the algorithm for solving Equation (1). For the integer decision variables n, n1 snb 、k、v1、v2、m1、n2 snb 、m2, perform an exhaustive search. Specifically, for these integer decision variables, all possible combinations of values are examined. Since the value of the real variable ε1 is not required for operation, only the existence of an executable ε1 is confirmed.

[0035] (The method of the present disclosure) As shown in FIGS. 5-6, the intermediate switches are divided into S groups. The number of intermediate switches in the first group is m1, the number of intermediate switches in the second group is m2,..., and the number of intermediate switches in the S-th group is m SIt is a station. The number of links from the input / output switch to the intermediate switches in the first group is v1, the number of links to the intermediate switches in the second group is v2, …, and the number of links to the intermediate switches in the S-th group is v S pieces.

[0036] The method for processing connection requests is extended as follows compared with the conventional method.

[0037] In a certain step i ∈ [1, S], when the number of terminals being connected on the input side or the output side of the input / output switch for which a connection request has been made exceeds n i-1 snb by n i snb and is less than or equal to n i at the following time, any one of the m

[0038] The ILP in this disclosure is expressed by equations (2a) to (2m).

[0039]

Number

[0040] The following are given as problems. · The number of ports N of the switch · The total number a of input / output switches and intermediate switches · The terminal utilization rate p · The allowable blocking rate ε

[0041] The number of groups S of intermediate switches is determined by the person using the network design device 1 according to this disclosure. For example, the network can be designed by increasing S = 1, 2, 3, …. However, the larger S is, the longer the calculation time will be.

[0042] The following are the decision variables of the optimization problem. · The number of terminals n · The number of input / output switches k · The number of intermediate switches m1, m2, …, m in each step S · The number of links v1, v2, …, v from one intermediate switch to one input / output switch in each step S · The upper limit value n1 of the number of terminals connected to the input / output switch for which a connection is requested in each step snb , n2 snb , …, n S snb

[0043] ε1, ε2, … ε S-1 are intermediate variables required only for the solution. Equations (2h)-(2m) are the definition domains of the variables.

[0044] Equation (2a) is the objective function. The objective function maximizes the network capacity. Equation (2b) is a condition regarding the number of ports of the input / output switch. Equation (2b) sums the number of links to the intermediate switch i. Equation (2c) is a condition regarding the number of ports of the intermediate switch. In Equation (2c), the condition is defined for each group i. Equation (2e) is a condition regarding the number of switches. Equation (2e) sums the number of intermediate switches for each group i.

[0045] Equations (2d)-(2g) are constraints for controlling blocking. These constraints are generally the same as the method described in Non-Patent Document 1. Equation (2d) ensures that no blocking occurs if the number of used ports at step i is n i snb or less. For the s < S step, according to Equation (2f), it is ensured that the sum of the blocking ratios when the number of used ports is more than n s snb does not exceed ε s . Note that the subscript representing the step number is expressed as s here. Regarding the S step, according to Equation (2g), it is ensured that the blocking ratio does not exceed ε.

[0046] Next, the calculation method in the present disclosure will be described. In the present disclosure, since the number of decision variables is larger than that of the conventional method, the following contrivance is made. · The number of input / output switches k and the number of terminals n are efficiently determined by binary search, and other integer variables (n isnb , m i , v i) is fully searched in the same way as the conventional method. · Real variable ε i assumes the monotonicity of the constraint conditions and checks the feasibility (whether the constraints are satisfied) only at the endpoints.

[0047] Referring to FIG. 7, an example of the processing of the calculation method according to the present disclosure is shown.

[0048] In the third row, using the existing method (Patent Document 1), a non-blocking network (a network with a blockage rate of zero) is obtained. In the present disclosure, since blockages of ε or less are allowed, the network capacity will not be smaller than that of the non-blocking network. Therefore, the present disclosure uses the non-blocking network as the starting point of the search. In the fourth and fifth rows, the capacity and structure of the non-blocking network are substituted.

[0049] In lines 6 - 14, for each combination of the respective values of m i , v i , and n i snb a loop is defined.

[0050] Lines 15 - 17 determine the range of k, specifically the lower limit value K D and the upper limit value K U . The range of k is then bisected by shrinking the range in half at a time to perform a binary search. The lower limit of k is derived from Equation (2b). The upper limit of k is derived from Equations (2c) and (2e).

[0051] Next, in lines 19 - 38, a binary search is performed on n. The lower limit (left) of the range of n in the binary search is, as shown in line 19, the value obtained by dividing the network capacity SW at that time by k. The upper limit (right) of the range of n is, as shown in line 20, the maximum value of n, that is, N. In lines 22 - 23, if the product of the maximum value of n and k is smaller than the previously obtained network capacity, the capacity cannot be improved, so the binary search is terminated.

[0052] In the 25th line, check the constraint conditions at the median value of n. If all the constraint conditions are satisfied and the network capacity is greater than the previously obtained network capacity, in lines 25 - 28, update the network capacity and network structure. Further, update the lower limit of n to the median value of n, and in line 29, continue the binary search. In lines 30 - 31, if the constraint conditions other than formula (2k) are satisfied at the median value of n, update the lower limit value of n to the median value of n. Specifically, change the range of n to the upper half, and continue the binary search without updating the maximum value of the network capacity.

[0053] Here, the reason for taking the upper half of the range of n as the shortest is that for n greater than the median value of n, a feasible solution may exist.

[0054] In lines 32 - 33, if the constraint conditions other than formula (2k) are not satisfied at the median value of n, update the upper limit of n to a value 1 less than the median value of n, and end the binary search of n.

[0055] In lines 34 - 38, determine the value of k for the next search. In lines 34 - 36, if the maximum capacity is updated during the search of n and the lower limit of k becomes greater than the value of k during the search, update the lower limit of k and start the loop with the updated value of k. In lines 37 - 38, otherwise, increment the value of k by 1.

[0056] In line 39, return the optimal network capacity and network structure and end the process.

[0057] In the processes of FIGS. 5 - 6, checking the constraint conditions of formulas (2f) - (2g), specifically, determining whether there exists an ε i that satisfies all of formulas (2f) - (2g) is not easy. Therefore, this disclosure updates the minimum value ε i of ε i min based on the recurrence formulas of formulas (3a) and (3b). The minimum value ε i min is used to check formulas (2f) - (2g).

[0058] [Number]

[0059] Incidentally, ε i The minimum value ε i min When the expressions (2f)-(2g) are satisfied with ε, there exists an ε that satisfies the constraint condition. However, just because it is not satisfied does not necessarily mean that there does not exist an ε that satisfies the constraint condition. i i i There is not necessarily no ε that satisfies the constraint condition.

[0060] Therefore, the determination method of the expressions (3a)-(3b) may miss the optimal solution, more specifically, may not be able to find a structure with a larger network capacity. However, since the network structure obtained by this determination method always satisfies the constraint conditions, safe use is guaranteed. This determination method is considered a device for completing the search within a realistic time while sacrificing optimality. Even with this device, in a given environment, it takes about 1 day when S = 4 and about two weeks when S = 5 for the calculation time. Since network calculations are not performed frequently and generally take several weeks, these times are within the allowable range.

[0061] (Processing method) With reference to FIG. 8, an example of the network design method by the network design device 1 according to the present disclosure will be described.

[0062] In step S101, the network design device 1 identifies a non-blocking network and the capacity and structure of the non-blocking network. Next, for each combination of the respective values of m i , v i , and n i snb , the processing from step S102 to step S156 is performed.

[0063] In step S102, the network design device 1 performs a binary search from the upper and lower limit values of k. In step S103, the network design device 1 performs a binary search from the upper and lower limit values of n.

[0064] In step S104, the network design device 1 determines whether the network capacity specified by n and k is greater than the previously calculated value. If it is not greater, m i , v i , and n i snb For other combinations of, the processes of steps S102 to S156 are performed.

[0065] If it is greater, in step S105, the constraint conditions are checked with the intermediate value of n. If it is determined in step S106 that each constraint condition is not satisfied, the process proceeds to step S152. If it is determined in step S106 that each constraint condition is satisfied, in step S107, the network capacity and structure are updated with the current k and n. In step S151, the network design device 1 changes the lower limit value of n to the intermediate value and performs a binary search.

[0066] In step S152, the network design device 1 determines whether the constraint conditions other than the formula (2k) are satisfied with the intermediate value of n. If they are satisfied, in step S153, the network design device 1 updates the lower limit value of n to the intermediate value and performs a binary search. If they are not satisfied, in step S154, the network design device 1 updates the upper limit value of n to a value 1 less than the intermediate value of n and performs a binary search. In step S155, it is determined whether the search for n has ended. If the search for n has not ended, the process returns to step S152.

[0067] When the search for n has ended, in step S156, the network design device 1 updates the value of k. Specifically, when the maximum capacity is updated during the search for n and the lower limit of k becomes greater than the value of k during the search, the network design device 1 updates the lower limit of k and starts the loop with the updated value of k. Otherwise, the network design device 1 increments the value of k by 1.

[0068] m i , v i , and n i snbFor each combination of respective values, when the processing from step S102 to step S156 is completed, the process proceeds to step S157. In step S157, the network design device 1 determines and outputs an optimal network capacity and network structure.

[0069] (Network design device) Referring to FIG. 10, the network design device 1 according to the present disclosure will be described. The network design device 1 includes a plurality of intermediate switches and a plurality of input / output switches connected via one intermediate switch, and designs a network that satisfies a predetermined network capacity. Here, the predetermined network capacity is maximized under the condition of satisfying an allowable blocking rate ε that allows communication not to be possible between two terminals connected to a plurality of input / output switches when specifications such as the number of ports N of the switch, the total number a of intermediate switches and input / output switches, and the terminal utilization rate p of the terminals are given.

[0070] The network design device 1 includes an acquisition unit 11, a calculation unit 12, and a storage unit 13.

[0071] The storage unit 13 stores data processed by the network design device 1.

[0072] The acquisition unit 11 acquires the number of ports N of the switch, the total number a of a plurality of input / output switches and a plurality of intermediate switches each having an input port and an output port of the number of ports, the terminal utilization rate p of the terminals connected to the plurality of input / output switches, the allowable blocking rate ε that allows communication not to be possible between two terminals connected to the plurality of input / output switches, and the number of groups S (≧3) of the intermediate switches. The number of groups S for dividing the intermediate switches does not define the configuration of the network to be designed, but is necessary for calculating the network design.

[0073] The calculation unit 12 calculates the number of terminals n, the number of input / output switches k, the number of intermediate switches m, and the number of links v between one input / output switch and one intermediate switch such that, in a network satisfying the number of ports N, the total number a, and the terminal utilization rate p acquired by the acquisition unit 11, the total probability of each blocking occurrence probability when the number of terminals is increased one by one from the maximum value of the number of terminals n when the network becomes non-blocking does not exceed the allowable blocking rate ε, and the network capacity obtained by multiplying the number of terminals n and the number of input / output switches k is maximized.

[0074] At this time, in the network, a plurality of intermediate switches are divided into S groups, and in each of the S steps, the intermediate switches assigned to that step are used. Here, the number of links between one input / output switch and the intermediate switch varies for each group to which the switch belongs. At this time, when the number of terminals connected on the input side or the output side of the input / output switch for which a connection request has been made exceeds the upper limit value of the previous step and is less than or equal to the upper limit value of the current step, the calculation unit 12 uses any of the intermediate switches corresponding to the current step to make a connection, and calculates the number of terminals n, the number of input / output switches k, the number of intermediate switches m1, m2,..., m S of each group, and the number of links v1, v2,..., v S between one input / output switch and one intermediate switch of each group.

[0075] More specifically, the calculation unit 12 calculates the number of terminals n, the number of input / output switches k, the number of intermediate switches m1, m2,..., m S of each group, and the number of links v1, v2,..., v S between one input / output switch and one intermediate switch of each group according to formulas (2a)-(2m).

[0076] The calculation unit 12 calculates the number of intermediate switches m1, m2,..., m S of each group, the number of links v1, v2,..., v S between one input / output switch and one intermediate switch of each group, and the upper limit value n i snbFor each combination of values, perform an exhaustive search. For the number of terminals n and the number of input / output switches k, perform a binary search to find the values that maximize the network capacity.

[0077] Referring to FIG. 11, the effect of the network design device 1 according to the present disclosure will be described. FIG. 11 shows the network capacity (switching capacity) with respect to the number of switches a constituting the network. SNB indicates the network capacity in a non-blocking network. S = 2 indicates the network capacity calculated by the method described in Non-Patent Document 1. S>3 indicates the network capacity calculated by the method according to the present disclosure. The method according to the present disclosure shows that a larger network can be configured than in Non-Patent Document 1, especially when a is large.

[0078] In the network design that theoretically guarantees the blocking rate, the network design device 1 according to the present disclosure sets the number of links between switches to S. The network design device 1 can enhance the flexibility of the network configuration and obtain a larger network capacity. Also, in order to solve an equation including many decision variables, the network design device 1 can obtain a solution in a realistic time by applying a binary search within a possible range or applying a simplification such as Equation (3).

[0079] Referring to FIG. 12, a supplement about blocking will be made.

[0080] First, the usage of the network targeted in this embodiment will be supplemented. In the target network, the terminals and the input ports or output ports are connected from the beginning to the end. For example, terminal t1 1 is connected to switch S1, and this wiring does not change from the beginning to the end. The same applies to other terminals. All terminals t, r are connected to any input / output switch from the beginning to the end.

[0081] What changes within the network are the connection configuration within the switch and the presence or absence of signals between terminals. Communication occurs between a transmitting terminal t and a receiving terminal r. One terminal cannot communicate with multiple terminals simultaneously, but it can communicate with different terminals at different time intervals. The ratio of the time for which each terminal attempts to communicate is called the "terminal utilization rate."

[0082] For example, consider the situation where two terminals t1 1 , r1 k attempt to communicate. Using only free links (links through which no signal is passing), a path is prepared between the two terminals t1 1 , r1 k .

[0083] In the example of Figure 12, assume that there are free links between the input / output switch S1 and the intermediate switch S k+1 , and there are also free links between the intermediate switch S k+1 and the input / output switch S k . Therefore, the internal configuration of the intermediate switch S k+1 is changed to connect these links. Similarly, the internal configurations of the two input / output switches S1, S k are changed to connect links with each of the two terminals t1 1 , r1 k respectively. As a result, a path consisting only of free links is formed from terminal t1 1 to terminal r1 k . When terminal t1 1 transmits a signal, it will reach terminal r1 k .

[0084] When communication starts in this way, the links on the path are in use and are no longer free links, so other communications cannot be used. After that, when the communication ends, it returns to a free link. The link between the input / output switch and the terminal is assigned to a specific terminal, but the link between the input / output switch and the intermediate switch may be used by any terminal. When the links between the idle switches decrease, there may be no path consisting of free links even if communication is attempted, so communication may not be possible. This is called "blocking". In the present disclosure, Equation (2d) represents a condition under which blocking never occurs. It has been mathematically proven that blocking never occurs no matter how the selection method and order of the destination terminal are changed.

[0085] Such a network design device 1 according to the present disclosure can divide the intermediate switches into three or more groups in a Clos network, enhance the flexibility of the network configuration, and obtain a larger network capacity.

[0086] The network design device 1 according to the present disclosure described above uses, for example, a general-purpose computer system including a CPU (Central Processing Unit), a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906. In this computer system, each function of the network design device 1 is realized by the CPU 901 executing a program loaded on the memory 902.

[0087] Note that the network design device 1 may be implemented by one computer or may be implemented by a plurality of computers. Also, the network design device 1 may be a virtual machine implemented on a computer.

[0088] The program of the network design device 1 can be stored in a computer-readable recording medium such as an HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), or DVD (Digital Versatile Disc), or can be distributed via a network. The computer-readable recording medium is, for example, a non-transitory recording medium.

[0089] Note that the present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the gist thereof.

Explanation of Reference Numerals

[0090] 1 Network design device 11 Acquisition unit 12 Calculation unit 13 Storage unit 901 CPU 902 Memory 903 Storage 904 Communication device 905 Input device 906 Output device

Claims

1. A network design device for designing a network that includes a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, and that satisfies a predetermined network capacity, an acquisition unit that acquires the number of ports of a switch, the total number of a plurality of input / output switches and a plurality of intermediate switches each having an input port of the number of ports and an output port of the number of ports, the terminal utilization rate of terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches to be blocked, and the number of groups of intermediate switches (≥3); a calculation unit that calculates the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one of the input / output switches and one of the intermediate switches such that, in a network that satisfies the number of ports, the total number, and the terminal utilization rate, the sum probability of each blocking occurrence probability when the number of terminals is increased one by one from the maximum value of the number of terminals when the network is non-blocking does not exceed the allowable blocking rate, and the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized, wherein the calculation unit divides the plurality of intermediate switches into S groups, and in each of the S steps, uses the intermediate switches assigned to that step, and when the number of terminals being connected on the input side or the output side of an input / output switch for which a connection has been requested exceeds the upper limit value of the previous step and is less than or equal to the upper limit value of the current step, connects using any of the intermediate switches corresponding to the current step, to calculate the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group; a network design device.

2. The calculation unit performs an exhaustive search for each combination of values of the number of intermediate switches in each group, the number of links between one input / output switch and one intermediate switch in each group, and the upper limit value of the number of terminals being connected on the input side or the output side of an input / output switch for which a connection has been requested in each step, and performs a binary search for the number of terminals and the number of input / output switches to search for values that maximize the network capacity. The network design device according to Claim 1.

3. A network design method for designing a network that includes a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, and that satisfies a predetermined network capacity, wherein a computer A plurality of input / output switches having the number of ports of the switch, the input ports of the number of ports, and the output ports of the number of ports, the total number of the plurality of intermediate switches, the terminal utilization rate of the terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches not to be possible, and the number of groups of intermediate switches (≥3) are obtained, In a network that satisfies the number of ports, the total number, and the terminal utilization rate, when the number of terminals is increased one by one from the maximum value of the number of terminals when the network is non-blocking, the total probability of each blocking occurrence probability does not exceed the allowable blocking rate, and the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized. Calculate the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one input / output switch and one intermediate switch. When calculating, The plurality of intermediate switches are divided into S groups, and in each of the S steps, the intermediate switches assigned to that step are used. When the number of terminals being connected on the input side or the output side of the input / output switch for which a connection is requested exceeds the upper limit value of the previous step and is less than or equal to the upper limit value of the current step, use any of the intermediate switches corresponding to the current step to make the connection. Calculate the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group. Network design method.

4. A computer for designing a network including a plurality of intermediate switches and a plurality of input / output switches connected via one of the intermediate switches, the network satisfying a predetermined network capacity. An acquisition unit that acquires the number of ports of the switch, the total number of the plurality of input / output switches having the input ports of the number of ports and the output ports of the number of ports, and the plurality of intermediate switches, the terminal utilization rate of the terminals connected to the plurality of input / output switches, the allowable blocking rate that allows communication between two terminals connected to the plurality of input / output switches not to be possible, and the number of groups of intermediate switches (≥3). In a network that satisfies the number of ports, the total number, and the terminal utilization rate, when the network becomes non-blocking, the total probability of each blocking occurrence probability when the number of terminals is increased one by one from the maximum value of the number of terminals is not exceeded the allowable blocking rate, and the network capacity obtained by multiplying the number of terminals by the number of input / output switches is maximized. A calculation unit that calculates the number of terminals, the number of input / output switches, the number of intermediate switches, and the number of links between one input / output switch and one intermediate switch, and functions as, The calculation unit is, divide the plurality of intermediate switches into S groups, and in each of the S steps, use the intermediate switches assigned to that step, when the number of terminals being connected on the input side or the output side of the input / output switch for which a connection request has been made exceeds the upper limit value of the previous step and is less than or equal to the upper limit value of the current step, connect using any of the intermediate switches corresponding to the current step, calculate the number of terminals, the number of input / output switches, the number of intermediate switches in each group, and the number of links between one input / output switch and one intermediate switch in each group Network design program.

Images