JP2015201029A - Load distribution device, load distribution method and load distribution system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make loads applied to respective servers approximately equal, and distribute a client terminal group of terminals which are located geographically close to each other to a same server as possible.SOLUTION: A load distribution device 20 counts a request amount from a client terminal 10 in a lattice in a predetermined period, for every lattice formed by dividing a predetermined plane, and creates a request amount information holding table. Then the load distribution device 20 creates a partial area information holding table in which partial areas in which the request amount in every partial area is approximately equal are set as an aggregation of the lattice. When receiving the request from the client terminal 10, the load distribution device 20 refers to the partial area information holding table, and determines a server 30 of a distribution destination of the request, with the partial area information according to the position of the client terminal 10 as a key.

Description

  The present invention relates to a load distribution apparatus, a load distribution method, and a load distribution system.

  In a server system that processes requests from client terminals, if there is a large amount of requests from client terminals that cannot be processed by a single server, a load balancer is installed between the client terminal and the server system, and multiple load balancers It is widely used to distribute requests to other servers.

  As a load distribution method in the load distribution apparatus, for example, by determining the distribution destination server using the attribute information included in the request as a consistent hash method described in Non-Patent Document 1, There is known a method for distributing requests including the same attribute information to the same server.

  A technique is also known in which, when location information is included in a request from a client terminal, a server that is close to the client terminal on the network is identified from the location information, and the request is distributed to the server.

  Patent Document 1 describes a technique related to a load distribution method in which a key with a high request frequency is distributed to a plurality of servers and a key with a low request frequency is distributed to a single server according to the key request frequency. Thus, according to the present technology, for example, even when the amount of requests transmitted by each client terminal is not uniform, the load on the server that is the distribution destination can be approximately uniform.

JP2013-15269A

D. Karger et al., "Consistent Hashing and Random Trees: Distributed Caching Protocols for Relieving Hot Spots on the World Wide Web" in Proceedings of the 29th ACM Symposium on Theory of Computing (STOC'97), pp.654-663, May 1997 .

  Here, for example, in a server system or the like that processes sensor information transmitted by a client terminal, processing from a plurality of servers is necessary from the viewpoint of request amount, while sensor information from a geographically close client terminal group is required. There is a requirement to process with a single server as much as possible.

  To meet these requirements, for example, the load distribution method described in Non-Patent Document 1 is applied, and load distribution is performed using the location information included in the request as a key, so that client terminals that are geographically close to each other can be identical. It can be accommodated in other servers. However, the client terminals are not necessarily distributed uniformly, and if the request amount transmitted by each client terminal is not uniform, even if an area of the same area is allocated to each server, the server load is significantly biased. There is.

  Thus, if the server load is significantly biased, the computing resources of the server cannot be used effectively, and as a result, there is a problem that the number of servers more than originally necessary is required.

  On the other hand, in the load balancing method disclosed in Patent Document 1, the load on the server can be made almost uniform. In this case, requests for client terminals that are geographically close to each other are distributed to a plurality of servers. Therefore, there is a problem that the above requirements cannot be satisfied.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to solve the above-described problems and distribute client terminals that are geographically close to the same server as much as possible while making the load of each server substantially equal.

  In order to solve the above-described problems and achieve the object, the present invention provides a load distribution apparatus that distributes requests from client terminals to any server in a distributed processing system that performs load distribution by a plurality of servers. A request receiving unit that receives a request from the client terminal, a request amount counting unit that counts a request amount from a client terminal located in the region for a predetermined period, for each region that divides the predetermined plane, A partial area that is a set of areas is determined so that the required amount in each partial area is as uniform as possible, and a partial area information holding table that associates the determined partial area with identification information of the partial area When a partial area determination unit to be generated and a request from the client terminal are received, the partial area information holding table is received. Referring to Le, characterized in that it and a request distribution part for distributing the request to the corresponding server to the identification information of the partial areas belonging of said client terminal.

  According to the present invention, it is possible to distribute geographically close client groups to the same server as much as possible while making the load on each server substantially equal.

FIG. 1 is a diagram illustrating an example of a configuration of a load distribution system. FIG. 2 is a diagram for explaining the outline of the processing of the load distribution system. FIG. 3 is a diagram for explaining the outline of processing of the load distribution apparatus. FIG. 4 is a functional block diagram of the load balancer. FIG. 5 is a diagram illustrating an example of a request amount information holding table. FIG. 6 is a diagram illustrating an example of the partial area information holding table. FIG. 7 is a sequence diagram showing a processing procedure from counting the requested amount to determining a partial area. FIG. 8 is a flowchart showing details of the partial region determination processing. FIG. 9 is a sequence diagram illustrating a processing procedure for request distribution. FIG. 10 is a flowchart showing details of the request processing server determination process. FIG. 11 is a diagram illustrating an example of the partial area number holding table. FIG. 12 is a sequence diagram showing a procedure for increasing the number of partial areas. FIG. 13 is a diagram illustrating a computer that executes a load distribution program.

  Hereinafter, embodiments (embodiments) for carrying out the present invention will be described with reference to the drawings.

(System configuration)
First, the configuration of the load distribution system of this embodiment will be described with reference to FIG. The load distribution system includes client terminals 10 (10A to 10D), a load distribution device 20, and servers 30 (30A to 30D). In the load distribution system, the client terminal 10 and the load distribution apparatus 20 are connected by, for example, a network 41, and the load distribution apparatus 20 and the server 30 are connected by a network 42. The networks 41 and 42 are, for example, IP (Internet Protocol) networks. The number of each is not limited to the number shown in FIG.

  The client terminal 10 transmits a service request (hereinafter abbreviated as a request as appropriate) to the server 30 via the load balancer 20. The client terminal 10 transmits a request including, for example, sensor information included in the client terminal 10 and its own position information acquired by, for example, GPS (Global Positioning System) to the load balancer 20.

  When receiving a request from the client terminal 10, the load distribution apparatus 20 distributes the request to one of the servers 30 based on the location information of the client terminal 10.

  When the server 30 receives a request from the client terminal 10 via the load balancer 20, the server 30 executes processing based on the request. For example, the server 30 performs application logic such as statistical processing on the sensor information included in the received request based on the partial area information of the client terminal 10 included in the request. As a specific example, the server 30 collects sensor information included in a request having the same partial area information (for example, a request from the geographically close client terminal 10), and performs statistical processing or the like on the sensor information. Note that the number of servers 30 installed is, for example, such that the average load of each server 30 is about 80% in accordance with the total required amount reaching the load balancer 20.

(Outline of processing)
The outline of the processing of the load distribution system will be described with reference to FIG. When receiving a request from the client terminal 10, the load distribution apparatus 20 of the load distribution system distributes the request to any one of the servers 30. Then, the server 30 executes processing based on the received request. The request from the client terminal 10 includes, for example, identification information of the client terminal 10 (for example, an IP (Internet Protocol) address), position information of the client terminal 10, and the like. The load balancer 20 determines the server 30 to which the request is distributed based on the identification information and the position information included in the request from the client terminal 10, and the determination of the server 30 as the distribution destination here is as follows. Do it.

  That is, first, the load distribution apparatus 20 creates information indicating the distribution of the client terminals 10 based on the position information of the client terminals 10. For example, as shown by reference numeral 101, information indicating the distribution of the client terminals 10 on a predetermined plane is created. After that, the load distribution apparatus 20 divides the plane into a predetermined number of grids as shown in FIG. 2, and counts the required amount in a predetermined period for each grid unit (see reference numeral 102). As a result, the required amount from each client terminal 10 in the grid in a predetermined period can be obtained. Information indicating the result of counting the request amount in units of grids is referred to as a request amount information holding table (see FIG. 5).

  Thereafter, the load distribution apparatus 20 designs a partial area in which the request amounts in the request amount information holding table are substantially uniform as a set of lattices. For example, the load distribution apparatus 20 designs nine partial areas indicated by reference numeral 103 as a set of lattices whose request amounts are substantially uniform. A1, A2, A3, B1, B2, B3, C1, C2, and C3 attached to the partial area are identification information (partial area information) of the partial area.

  The load balancer 20 refers to the information (partial area information holding table) obtained in this way to determine a server 30 as a distribution destination.

  For example, when receiving a request from the client terminal 10, the load distribution apparatus 20 refers to the partial area information holding table and determines a partial area to which the received request belongs. For example, the load distribution apparatus 20 determines that the partial area to which the received request belongs is “A2”. Then, the load balancer 20 selects the server 30 corresponding to the partial area by the consistent hash method based on the determination result of the partial area (for example, “A2”), and assigns the selected server 30 to the distribution destination. Server 30 is determined. As a result, requests belonging to the same partial area are distributed to the same server. Thereafter, the load balancer 20 forwards the request to the distribution destination server 30. Note that the load balancer 20 may add and transfer the partial area information (partial area information) corresponding to the client terminal 10 to the request from the client terminal 10. Thereafter, the server 30 performs application logic on the transferred request, for example, in units of partial areas.

(Details of processing)
The process in the load distribution apparatus 20 will be described in detail with reference to FIG. Here, it is assumed that the client terminal 10A and the client terminal 10B are geographically close, and the client terminal 10C and the client terminal 10D are geographically close.

  The load balancer 20 counts the request amount of the client terminal 10 in the area within each grid during a predetermined period, and records the count result in the request amount information holding table. Then, the load distribution apparatus 20 determines a partial area in which the request amounts in the request amount information holding table are approximately equal, and obtains, for example, the partial area information holding table shown in FIG.

  Thereafter, when receiving a request from the client terminal 10, the load distribution apparatus 20 reads partial area information corresponding to the position of the grid to which the client terminal 10 belongs and distributes the request. For example, when the load balancer 20 receives a request from the client terminal 10A ((X, Y) = (2, 2)), the partial area information “(X, Y) = (2, 2) corresponding to“ X, Y) = (2, 2) ”. “A1” is read, and the request is distributed to the server 30 (for example, server 30B) corresponding to “A1”. When the load balancer 20 receives a request from the client terminal 10B ((X, Y) = (2, 3)), the partial area information “X” corresponding to (X, Y) = (2, 3) “ “A1” is read, and the request is distributed to the server 30 (for example, server 30B) corresponding to “A1”. As a result, requests from geographically close client terminals 10A and 10B are distributed to the same server 30B.

  Similarly, for example, when the load balancer 20 receives a request from the client terminal 10C ((X, Y) = (5, 4)), this corresponds to (X, Y) = (5, 4). The partial area information “B2” is read out, and the request is distributed to the server 30 (for example, server 30C) corresponding to “B2”. When the load balancer 20 receives a request from the client terminal 10D ((X, Y) = (5, 5)), the partial area information “B2” corresponding to (X, Y) = (5, 5) ”And distributes the request to the server 30 (for example, server 30C) corresponding to“ B2 ”. As a result, requests from geographically close client terminals 10C and 10D are distributed to the same server 30C.

(Constitution)
Next, the configuration of the load distribution apparatus 20 will be described with reference to FIG. The load distribution device 20 includes an input / output unit 21, a control unit 22, and a storage unit 23.

  The input / output unit 21 inputs / outputs various information to / from an external device. For example, the input / output unit 21 receives a request from the client terminal 10 or transfers the request to the server 30. The input / output unit 21 is realized by a communication interface for performing communication via the networks 41 and 42, for example.

  The control unit 22 controls the entire load distribution apparatus 20, and includes a request reception unit 221, a request amount counting unit 222, a partial area determination unit 223, and a request distribution unit 224. The partial area number updating unit 225 indicated by the broken line may be equipped or not equipped, and the case of equipped will be described later.

  The request receiving unit 221 receives a request from the client terminal 10 via the input / output unit 21.

  The request amount counting unit 222 counts the request amount from each client terminal 10 in a predetermined period for each region based on the position information of the client terminal 10 that is the transmission source of the request. That is, the request amount counting unit 222 determines which region each client terminal 10 has on a region (for example, a lattice region) that divides a predetermined plane based on the position information of each client terminal 10 that is a transmission source of the request. It is judged whether it belongs to. Then, the request amount counting unit 222 counts the request amount from the client terminal 10 belonging to the region for a predetermined period for each region, and writes the count result in the request amount information holding table (see FIG. 5).

  In the following description, the shape of the region is described as an example of a lattice, but may be a shape other than the lattice. In addition, it is assumed that the number of areas into which the plane is divided is set in advance according to the processing performance of the load distribution apparatus 20. For example, if the processing capacity of the load balancer 20 is relatively high, a large number of areas are set.

  The partial area determination unit 223 determines a partial area that is a set of areas (for example, grid-like areas) on the predetermined plane so that required amounts in the partial areas are as uniform as possible. Then, the partial region determination unit 223 generates a partial region information holding table (see FIG. 6) in which the determined partial region is associated with the identification information of the partial region. The generated request amount information holding table is stored in the storage unit 23. Note that the number of partial areas determined by the partial area determination unit 223 is at least the number of servers 30 in the load distribution system. Details of the processing of the partial area determination unit 223 will be described later with specific examples.

  The request distribution unit 224 distributes the request from the client terminal 10 to one of the servers 30. For example, when receiving a request from the client terminal 10, the request distribution unit 224 reads partial area information corresponding to the position of the client terminal 10 from the partial area information holding table. Then, the request distribution unit 224 determines the server 30 corresponding to the key by the consistent hash method or the like using the partial area information as a key. Then, the request distribution unit 224 transmits (transfers) the request from the client terminal 10 to the determined server 30. Note that when the request distribution unit 224 transmits a request to the server 30, the request distribution unit 224 may transmit the partial area information read from the partial area information holding table.

  The storage unit 23 stores the request amount information holding table and the partial area information holding table in a predetermined area. The partial area number holding table indicated by the broken line will be described later.

  The request amount information holding table is information indicating the count result of the request amount from the client terminal 10 in the region for a predetermined period for each region obtained by dividing the predetermined plane. An example of the request amount information holding table is shown in FIG. The required amount information holding table shown in FIG. 5 is composed of a total of 64 regions (lattices) by dividing the coordinate plane into 8 for each of the X axis and the Y axis. The numerical value in each grid is the total number of requests from each client terminal 10 in the grid (area) in a predetermined period. The request amount information holding table is updated based on the count result by the request amount count unit 222. The required amount information holding table is stored as a two-dimensional array on the storage unit 23.

  The partial area information holding table is information in which the partial area determined by the partial area determination unit 223 is associated with the identification information of the partial area. An example of this partial area information holding table is shown in FIG. The partial area information holding table shown in FIG. 6 shows nine partial areas obtained by dividing the entire lattice of the requested amount information holding table shown in FIG. 5 into three for each of the X axis and the Y axis. A1, A2, A3, B1, B2, B3, C1, C2, and C3 attached to the partial area are identification information (partial area information) of the partial area. This partial area information holding table is referred to when the request distribution unit 224 determines a server 30 to which a request is distributed.

  According to such a load balancer 20, it is possible to distribute the geographically close client terminals 10 to the same server as much as possible while roughly equalizing the load of each server 30 of the load balancing system.

(Processing procedure)
Next, the processing procedure of the load distribution system will be described. It is assumed that the processing from counting the request amount to determining the partial area and the request distribution processing described below are executed in parallel. First, a processing procedure from counting the requested amount to determining a partial area will be described with reference to FIG.

  When the request receiving unit 221 of the load distribution apparatus 20 receives a request from the client terminal 10 (S1), the request amount counting unit 222 performs a request amount counting process (S2). The processes of S1 and S2 are repeated. That is, the request amount counting unit 222 continuously counts the number of requests transmitted from the client terminal 10. Then, the request amount counting unit 222 writes the request amount from the client terminal 10 in the request amount information holding table. The request amount counting unit 222 divides the coordinate plane in which the client terminal 10 is located in advance, for example, in a lattice shape, and the position information of the client terminal 10 included in the request (for example, latitude / longitude information acquired by GPS) Etc.), it is determined which grid the position of the client terminal 10 corresponds to, and the required amount for each grid is counted. After S2, the partial area determination unit 223 of the load balancer 20 performs a partial area determination process based on the information in the request amount information holding table (S3).

  Next, the details of the partial region determination processing in S3 of FIG. 7 will be described with reference to FIG. Here, the entire lattice shown in the request amount information holding table is divided into three parts for each of the X axis and the Y axis (that is, two boundaries are placed on each of the X axis and the Y axis), and a total of nine partial areas are formed. A case of creating will be described as an example.

  First, the partial area determination unit 223 determines the boundary of the partial area by dynamic programming for the X axis of the plane shown in the request amount information holding table (see FIG. 5) (S301). Specifically, the partial region determination unit 223 determines the boundary that minimizes the variance of the required amount of each partial region by dynamic programming on the premise that the Y axis is not divided.

  Next, the partial region determination unit 223 fixes the boundary of the X axis determined in S301, and determines the boundary of the partial region by dynamic programming for the Y axis (S302). Specifically, the partial region determination unit 223 determines a boundary for minimizing the variance of the required amount of each partial region with respect to the Y axis by dynamic programming.

  Next, the partial region determination unit 223 fixes the boundary of the Y axis determined in S302, and again determines a boundary for minimizing the variance of the required amount of each partial region for the X axis by dynamic programming. (S303).

  Next, the partial area determination unit 223 determines whether or not the variance value of the requested amount of each partial area has been improved by the process of S303 (S304). That is, the partial area determination unit 223 determines whether or not the variance value of the request amount of each partial area calculated by the process of S303 is lower than the previously calculated value.

Here, if the variance value of the requested amount of each partial area has not been improved (No in S304),
The partial area determination unit 223 updates the partial area information holding table based on the X-axis and Y-axis boundaries determined in S302 and S303 (S305). On the other hand, if the variance value has been improved in S304 (Yes in S304), the process returns to S302. That is, the partial region determination unit 223 searches for a partial region in which the required amount of each partial region becomes as uniform as possible while changing the boundary of the partial region. When the partial area determination unit 223 finds a partial area in which the required amount of each partial area is as uniform as possible, the partial area information holding table is based on the X-axis and Y-axis boundaries determined in S302 and S303. Is stored with the partial area information of each grid (see FIG. 6).

  The partial region determination unit 223 performs the above processing, for example, every predetermined period. Further, the value of the request amount shown in the request amount information holding table is reset when the partial area is determined.

  The partial area determination unit 223 performs the above processing, so that it is possible to determine a partial area that makes the required amount of each partial area as uniform as possible. In addition, since the partial area determination unit 223 performs the above processing every predetermined period, when there is a change in the distribution of the requested amount due to increase / decrease or movement of the client terminal 10, the part reflecting such a change An area can be determined.

  Next, a procedure in which the load balancer 20 distributes requests from the client terminals 10 will be described with reference to FIG.

  When the request receiving unit 221 of the load distribution apparatus 20 receives a request from the client terminal 10 (S501), the request distribution unit 224 of the load distribution apparatus 20 refers to the partial area information holding table and processes the request. The server 30 is determined (S502: request processing server determination processing). Then, the request distribution unit 224 transmits a request to the determined server 30 (S503). Upon receiving this request, the server 30 executes processing based on the received request (S504: service request processing). For example, the server 30 collects sensor information included in the received request for each partial region information included in the request, and performs statistical processing or the like.

  Hereinafter, the details of the request processing server determination process in S502 of FIG. 9 will be described with reference to FIG.

  First, the request distribution unit 224 specifies a lattice (area) corresponding to the position of the client terminal 10 that is the transmission source of the request from the position information included in the request transmitted from the client terminal 10 (S601). That is, the request distribution unit 224 identifies a lattice (region) on the partial region information holding table corresponding to the position of the client terminal 10 that is the transmission source of the request.

  Thereafter, the partial area determination unit 223 acquires partial area information of the lattice (area) to which the client terminal 10 belongs from the partial area information holding table (S602).

  For example, when the position on the partial area information holding table (see FIG. 6) corresponding to the position of the client terminal 10 is (X, Y) = (5, 4), the request distribution unit 224 sets the partial area information holding table. "B2" is acquired as the partial area information.

  Next, the partial region determination unit 223 determines the server 30 that is the transmission destination of the request by the consistent hash method using the acquired partial region information as a key (for example, “B2”) (S603).

  In this way, the request distribution unit 224 can determine the request distribution destination server 30 with reference to the partial area information holding table.

(Other embodiments)
When the load balancer 20 receives a notification from one of the servers 30 in the load balance system that the load exceeds a predetermined threshold (for example, 95% of the processing performance of the server 30), the load balancer 20 holds the partial area information. The number of partial areas in the table may be increased.

  In this case, the load distribution apparatus 20 further includes a partial area number updating unit 225 shown in FIG. The storage unit 23 of the load balancer 20 further stores a partial area number holding table.

  The partial area number updating unit 225 increases the number of partial areas in the partial area number holding table when receiving a notification that the load has exceeded a predetermined threshold (server load threshold excess notification) from the server 30.

  The partial area number holding table is information indicating the number of partial areas. The number of partial areas is indicated by, for example, a combination of the number of boundaries between the X axis and the Y axis that serve as a break between partial areas in the partial area information holding table (see FIG. 11). The partial area number holding table illustrated in FIG. 11 indicates that the number of X-axis and Y-axis boundaries serving as the breaks between the partial areas of the partial area information holding table is “8”. The partial area number holding table is referred to when the partial area determination unit 223 determines a partial area. That is, when the partial area number holding table is updated, the partial area determining unit 223 resets the current partial area information holding table, and sets the partial area again so that the partial area number shown in the partial area number holding table is reached. decide.

  The procedure for increasing the number of partial areas in the load balancer 20 will be described with reference to FIG. When receiving the server load threshold excess notification from the server 30 (S401), the partial area number updating unit 225 of the load distribution apparatus 20 performs the partial area number addition process (S402). That is, the partial area number updating unit 225 increases the value of the partial area number holding table.

  For example, as shown in the partial area count holding table of FIG. 11, when the number of partial area boundaries is “8” for each of the X axis and the Y axis, the partial area number updating unit 225 displays the number of X axis boundaries. Is updated from “8” to “9”. Then, the partial region determination unit 223 resets the current partial region information holding table again so that the number of partial regions (the number of X-axis boundaries is “9”) shown in the updated partial region number holding table is reached. Determine the partial area.

  When the partial area number updating unit 225 continues to receive the server load threshold value excess notification, the partial area number addition process is further performed. For example, the partial region number updating unit 225 updates the number of Y-axis boundaries from “8” to “9”. Then, the partial area determination unit 223 resets the current partial area information holding table, and becomes the number of partial areas (the number of X-axis and Y-axis boundaries is “9”) shown in the updated partial area number holding table. Then, the partial area is determined again. Thereafter, every time a server load threshold excess notification is received, the number of boundaries between the X axis and the Y axis is added alternately.

  In this way, by increasing the number of partial areas each time the load distribution apparatus 20 receives a server load threshold excess notification, the possibility of reducing the load bias among the servers 30 can be increased.

  When the server 30 of the load distribution system is added or removed, the load distribution apparatus 20 resets the number of partial areas to the minimum value (the number of servers 30 of the load distribution system).

  In the above-described embodiment, the request transmitted from the client terminal 10 has been described as including the location information of the client terminal 10, but the present invention is not limited to this. For example, when the client terminal 10 is fixedly installed and does not move, it is not necessary to include position information in each request. In this case, the load balancer 20 holds information indicating the correspondence between the identification information (for example, IP address) of the client terminal 10 and the position information in advance, and the position of the client terminal 10 is determined based on this information. What is necessary is just to specify.

  Further, although the load balancer 20 detects that the load on the server 30 is high by receiving the server load threshold excess notification, the load balancer 20 records the number of requests distributed to each server 30 in advance. The load height of each server 30 may be estimated based on the number of distributions.

  Further, although the load distribution apparatus 20 has been described as determining a rectangular partial area when determining the partial area, the shape of the partial area may be other than that.

  Although the load balancer 20 has been described as using dynamic programming when determining a partial region that minimizes the variance of the requested amount, it should be understood that algorithms other than dynamic programming may be used. Good.

(program)
It is also possible to create a program in which the processing executed by the load distribution apparatus 20 according to the above embodiment is described in a language that can be executed by a computer. In this case, the same effect as the above-described embodiment can be obtained by the computer executing the program. Further, such a program may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read by a computer and executed to execute the same processing as in the above embodiment. Hereinafter, an example of a computer that executes a load distribution program that realizes the same function as the load distribution apparatus 20 will be described.

  FIG. 13 is a diagram illustrating a computer that executes a load distribution program. As illustrated in FIG. 13, the computer 1000 includes, for example, a memory 1010, a CPU 1020, a hard disk drive interface 1030, a disk drive interface 1040, a serial port interface 1050, a video adapter 1060, and a network interface 1070. These units are connected by a bus 1080.

  The memory 1010 includes a ROM (Read Only Memory) 1011 and a RAM (Random Access Memory) 1012. The ROM 1011 stores a boot program such as BIOS (Basic Input Output System). A hard disk drive 1090 is connected to the hard disk drive interface 1030. A disk drive 1100 is connected to the disk drive interface 1040. A removable storage medium such as a magnetic disk or an optical disk is inserted into the disk drive 1100, for example. For example, a mouse 1110 and a keyboard 1120 are connected to the serial port interface 1050. For example, a display 1130 is connected to the video adapter 1060.

  Here, as shown in FIG. 13, the hard disk drive 1090 stores, for example, an OS 1091, an application program 1092, a program module 1093, and program data 1094. Each table described in the above embodiment is stored in the hard disk drive 1090 or the memory 1010, for example.

  Further, the load distribution program is stored in the hard disk drive 1090 as a program module in which a command executed by the computer 1000 is described, for example. Specifically, a program module describing each process executed by the load distribution apparatus 20 described in the above embodiment is stored in the hard disk drive 1090.

  Further, data used for information processing by the load balancing program is stored as program data in, for example, the hard disk drive 1090. Then, the CPU 1020 reads out the program module 1093 and the program data 1094 stored in the hard disk drive 1090 to the RAM 1012 as necessary, and executes the above-described procedures.

  Note that the program module 1093 and the program data 1094 related to the load balancing program are not limited to being stored in the hard disk drive 1090. For example, the program module 1093 and the program data 1094 are stored in a removable storage medium and read by the CPU 1020 via the disk drive 1100 or the like. May be issued. Alternatively, the program module 1093 and the program data 1094 related to the load distribution program are stored in another computer connected via a network such as a LAN (Local Area Network) or a WAN (Wide Area Network), and are transmitted via the network interface 1070. May be read by the CPU 1020.

10, 10A to 10D Client terminal 20 Load distribution device 21 Input / output unit 22 Control unit 23 Storage unit 30, 30A to 30D Server 221 Request receiving unit 222 Request amount counting unit 223 Partial region determining unit 224 Request distribution unit 225 Updating the number of partial regions Part

Claims (8)

In a distributed processing system that performs load distribution by a plurality of servers, a load distribution apparatus that distributes requests from client terminals to any of the servers,
A request receiving unit that receives a request from the client terminal;
A request amount counting unit that counts a request amount from a client terminal located in the region for a predetermined period for each region that divides the predetermined plane;
A partial area that is a set of areas on the plane is determined so that the required amount in each partial area is as uniform as possible, and the determined partial area is associated with the identification information of the partial area A partial area determination unit for generating an information holding table;
When receiving a request from the client terminal, a request distribution unit that distributes the request to a server corresponding to identification information of the partial area to which the client terminal belongs, with reference to the partial area information holding table;
A load balancer comprising: The partial region determination unit
The load distribution apparatus according to claim 1, wherein when the load on the server increases, the number of partial areas is increased. The request distribution unit
3. The load distribution apparatus according to claim 1, wherein when the request is distributed to the server, the request includes identification information of the read partial area. The request distribution unit
The load distribution according to any one of claims 1 to 3, wherein when determining a server to which the request is distributed, the determination is performed by a consistent hash method using identification information of the partial area as a key. apparatus.   The load distribution apparatus according to claim 1, wherein the area is an area obtained by dividing a predetermined plane into a lattice shape. The request amount counting unit
6. The load distribution apparatus according to claim 1, wherein after the predetermined period has elapsed, the request amount from the client terminal for each region is counted again. In a distributed processing system that distributes load by a plurality of servers, a load distribution device that distributes requests from client terminals to any of the servers,
Receiving a request from the client terminal;
For each area that divides a predetermined plane, counting a request amount from a client terminal located in the area for a predetermined period;
A partial area that is a set of areas on the plane is determined so that the required amount in each partial area is as uniform as possible, and the determined partial area is associated with the identification information of the partial area Generating an information holding table;
When receiving a request from the client terminal, referring to the partial area information holding table, and distributing the request to a server corresponding to identification information of the partial area to which the client terminal belongs;
The load distribution method characterized by performing. A load distribution system comprising: a load distribution device that distributes requests from client terminals to any server; and a server that is a distribution destination of the requests,
The load balancer is:
A request receiving unit that receives a request from the client terminal;
A request amount counting unit that counts a request amount from a client terminal located in the region for a predetermined period for each region that divides the predetermined plane;
A partial area that is a set of areas on the plane is determined so that the required amount in each partial area is as uniform as possible, and the determined partial area is associated with the identification information of the partial area A partial area determination unit for generating an information holding table;
When receiving a request from the client terminal, a request distribution unit that distributes the request to a server corresponding to identification information of the partial area to which the client terminal belongs, with reference to the partial area information holding table;
A load distribution system comprising:

Images