JP2001202318A - Data distribution system - Google Patents

Abstract

(57) [Summary] [Problem] To provide a data distribution system in which a management device transmits data to a client device via a network by a server device selected from a plurality of server devices, using an accurate load of each server device. Automatically distribute the load on the server device. Each of the server devices (S1 to SN) includes a transmission unit for transmitting data to a client device (4) by a transmission process performed by a CPU using an internal memory.
The management device 1 is notified of the U usage rate and the internal memory usage rate at a predetermined timing. In the management device, the management unit manages the CPU usage rate and the internal memory usage rate notified from each server device, and the transmission control unit responds to a request transmitted from the client device via the network by using the CPU usage rate and the internal memory usage rate. The data is transmitted to the client device by the server device having a small load determined using the usage rate.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data distribution system in which a management device transmits data to a client device via a network by a server device selected from a plurality of server devices, and to such a management device.
The present invention relates to a technique for automatically distributing the load on a server device using the CPU usage ratio and the internal memory usage ratio of each server device.

[0002]

2. Description of the Related Art For example, in a data distribution system of the Internet, a client device on the user side and a server device on the service data supply side are provided, and data is transmitted from the server device in response to a request transmitted from the client device via the Internet. (Service Data) is transmitted to the client device via the Internet.

In the conventional data distribution system, only one server device is provided. For example, when access to the server device fails, the user performs access retry or the like to display HTML data or the like. The request had to be reissued. For this reason, for example, when access to the server device is congested or the like and the load on the server device is high, there is a problem that the timeout of the access process frequently occurs on the user side.

To avoid such a problem, there is known a data distribution system provided with a mirror server device storing the same data as the server device, for example. In this data distribution system, the user switches the access destination to the one with the lightest load between the server device and the mirror server device, so that the user accesses the device with the lightest load (server device or mirror server device). Thus, data can be received from the device, whereby the load can be distributed between the server device and the mirror server device.

[0005] FIG. 5 shows an example of an Internet data distribution system provided by connecting the server (main server) device 17 and the mirror server device 18 via the Ethernet 16 as described above. In this data distribution system, a client device 11 is connected to the Internet 14 via a modem 12 and a provider 13 and the like, and a server device 17 and a mirror server device 18 are connected to an Ethernet 16 and a router (or proxy server) 15. Is connected to the Internet 14 through the Internet.

However, in the data distribution system provided with the mirror server device as described above, the user grasps the load condition of each device unless the user actually accesses each device (server device or mirror server device). In addition, there is a trouble that the user has to change the address of the access destination according to the device (server device or mirror server device) to which the user desires access.

In order to solve such an inconvenience, for example, in an Internet server system described in Japanese Patent Application Laid-Open No. Hei 10-27148 (hereinafter referred to as Document 1), a management server periodically sends a request to a server. The load level is set according to the response time of the monitoring command sent to the server, and the server with the lightest load is selected.

In a dynamic reconfiguration of a network server described in, for example, Japanese Patent Application Laid-Open No. 10-105500 (hereinafter referred to as Document 2), in a client server system having a plurality of servers, each server The processing load is maintained in the form of the number of accesses (requests) serviced per unit time, and when a processing load of one server becomes excessive, the server redirects a client request to another server. Do what you want.

Although the load of each server is not monitored, for example, a network load balancing for a multi-computer server described in Japanese Patent Application Laid-Open No. 11-143804 will be introduced. In this network load balancing for a multi-computer server, the message dispatcher responds to parameters representative of network load (eg, cumulative count of packet length and / or average number of packets per second) from the client. It determines which server to dispatch the request from among the plurality of servers.

[0010]

However, in the conventional data distribution system as described above, the management device that manages a plurality of server devices cannot accurately detect the actual load of each server device. There was a problem that the effect of dispersing the load of the device could not be sufficiently obtained.

For example, in the method of setting the load level from the response time as in the Internet server system described in the above-mentioned document 1, for example, depending on the degree of congestion of the LAN (Local Area Network) connecting the management server and the server. However, there is a problem that the response time changes, and the load level of each server cannot be accurately detected. Also, for example, in a method of maintaining the processing load in the form of a client server system described in the above-mentioned document 2, similarly, the CPU provided in each server
However, there is a problem that a processing load that matches the actual load, such as the processing load, is not always detected. In addition, in this system, since a management server is not provided and redirection is performed between servers, a delay is likely to occur in data distribution processing. There was a problem.

The present invention has been made in order to solve such a conventional problem. When a management apparatus transmits data to a client apparatus via a network by a server apparatus selected from a plurality of server apparatuses. It is an object of the present invention to provide a data distribution system capable of distributing a load on a server device based on an accurate load on a CPU and a memory of each server device, and to provide such a management device.

[0013]

In order to achieve the above object, a data distribution system according to the present invention manages a data in response to a request transmitted from a client device to a management device via a network as follows. The device causes the server device selected from the plurality of server devices to transmit data to the client device via the network.

That is, each server device has transmission means for transmitting data to the client device via the network by a transmission process performed by the CPU using the internal memory, and the notifying means uses the CPU usage rate and the internal memory usage rate. Is notified to the management device at a predetermined timing. On the management device, the management unit
And the use rate of the internal memory are managed, and the transmission control means responds to a request transmitted from the client device, and
Data is transmitted to the client device by a server device with a small load determined using the PU usage ratio and the internal memory usage ratio.

Therefore, the CPU provided in each server device
The usage rate of the server and the usage rate of the internal memory are managed, and the server apparatus that performs data transmission is selected based on the load determined using these. Therefore, an accurate load that matches the actual load of each server apparatus is selected. , The load on the server device can be distributed. The present invention provides not only such a data distribution system, but also a management device having the above-described management means and transmission control means.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment according to the present invention will be described with reference to the drawings. In this example, the data distribution system according to the present invention and the management device according to the present invention will be described together. FIG. 1 shows a schematic configuration example of a data distribution system according to the present invention. This data distribution system includes, for example, a representative port unit 1 composed of a computer and the like and a computer and the like as well. The plurality of (in this example, N) service data distribution units S1 to SN are connected to each other, and the plurality of service data distribution units S1 to SN are connected to each other.
A LAN 2 for connecting the SN and the representative port unit 1 is provided.

FIG. 1 shows a client device 4 composed of, for example, a computer and the Internet 3 connecting the representative port unit 1 and the client device 4. In this example, a plurality of service data distributions are provided. Section S1
SN and the client device 4 are also connected via the LAN 2 and the Internet 3. Although only one client device 4 is shown in FIG. 1, many client devices are connected to the Internet 3 in this example.

Here, the representative port unit 1 corresponds to an embodiment of the management device according to the present invention, and the service data distribution unit S1
To SN correspond to an embodiment of the server device according to the present invention,
The client device 4 corresponds to an embodiment of the client device according to the present invention. In the representative port unit 1, the service data distribution units S1 to SN, and the client device 4,
The respective CPUs activate and execute various application programs using the internal memory, thereby realizing various processes corresponding to the programs.

First, the functions provided in the client device 4, the representative port unit 1, and the service data distribution units S1 to SN will be schematically described. That is, the client device 4 is, for example, the Internet 3
It has a browser which is software for browsing the above data, accesses the representative port unit 1 via the Internet 3, and requests a signal (display) for requesting data to be displayed on its own display screen. Request) to the representative port unit 1. In addition, the client device 4 transmits the service data from the service data distribution units S1 to SN to the LAN2.
And a function of receiving display data transmitted via the Internet 3 and displaying the display data on its own display screen. The display request is transmitted, for example, according to an instruction from a user who operates the client device 4.

The representative port unit 1 serves as a window for receiving access from the client device 4, and based on the machine load information notified from each of the service data distribution units S1 to SN, each of the service data distribution units S1 to S1. SN
The client device 4 has a function of monitoring the load of the client device 4 by selecting the service data distribution units S1 to SN with a small load based on the load status of the monitored service data distribution units S1 to SN. Has the function of distributing the access (display request) from the service data distribution units S1 to SN.

Each of the service data distribution units S1 to SN
Has an external memory storing various data (service data) to be supplied to the client device 4 in response to an access (display request) from the client device 4, and responds to a display request from the client device 4. It has a function of transmitting the display data (for example, HTML data) to the client device 4 via the LAN 2 or the Internet 3. Here, in the present example, each of the service data distribution units S1 to SN can supply the same display data to the client device 4. Further, as described above, in this example, the display request from the client device 4 is transmitted to the service data distribution units S1 to SN via the representative port unit 1.

Each of the service data distribution units S1 to SN
Has a function of notifying the representative port unit 1 via the LAN 2 of information (machine load information) on the usage rate of the CPU provided therein and the usage rate of the internal memory, for example, at predetermined time intervals (fixed time intervals). Have. Here, the CPU usage rate indicates, for example, the ratio of the processing capacity actually used in the processing capacity of the CPU, and specifically, for example, the CPU execution rate per unit time. Shows the percentage of processes that are actually running out of the maximum possible number of processes. Also, the internal memory usage rate is
For example, it indicates the ratio of the storage capacity actually used in the storage capacity of the internal memory. Specifically, for example, the storage capacity is actually stored within the maximum data amount that can be stored in the internal memory. Shows the percentage of the data amount that has been used.

Next, referring to FIG. 2, a specific example of the load distribution processing performed by the above-described representative port unit 1 and the above-described service data distribution units S1 to SN will be described. FIG. 3 conceptually shows processes T1 to T4 performed by the service data distribution units S1 to SN and processes T11 to T13 performed by the representative port unit 1 with respect to the load distribution process of the present example. Further, in this example, since the service data distribution units S1 to SN have the same configuration and perform the same processing, in FIG. 3, only the processing performed by one service data distribution unit S1 is denoted by T1 to T4. It is attached.

First, the service data distribution units S1 to SN
5 shows a specific example of processing for acquiring machine load information and notifying the representative port unit 1 of the machine load information. That is, each of the service data distribution units S1 to SN acquires, for example, the usage rate of the CPU provided therein and the usage rate of the internal memory as machine load information at predetermined intervals (processing T).
1) The machine load information is stored and managed (process T2), and the machine load information is transmitted to the representative port unit 1 via the LAN 2 to notify the representative port unit 1 of the machine load information. (Process T3).

Next, a specific example of processing in which the representative port unit 1 manages the machine load information of each service data distribution unit S1 to SN and distributes a display request from the client device 4 to each service data distribution unit S1 to SN. Is shown. That is, in the representative port unit 1, each service data distribution unit S
1 to SN, receives the machine load information transmitted from each of the service data distribution units S1 to SN, and receives the received machine load information (that is, the CPU usage rate and the internal memory usage rate).
Is stored in the connection table and managed (process T11). Here, the connection table is created by, for example, the representative port unit 1 and held in the memory of the representative port unit 1, and stores the machine load information of each of the service data distribution units S1 to SN as described above (processing T12).

When there is an access from the client device 4 via the Internet 3, the representative port unit 1 detects the machine load information of each of the service data distribution units S1 to SN stored in the connection table, for example. And
The service data distribution units S1 to SN with the smallest load determined using the machine load information are selected, and a display request from the client device 4 is transmitted to the selected service data distribution units S1 to SN via the LAN2. (transfer)
Then, the access from the client device 4 is distributed (process T13).

Here, a specific example of how the representative port unit 1 of the present embodiment selects the service data distribution units S1 to SN with the smallest load using the machine load information of each service data distribution unit S1 to SN will be described. That is, the representative port unit 1 of the present example executes the above-described load distribution management processing and the like and the access distribution processing (processing T11 to processing T13) by executing a resident application program, for example. The load distribution evaluation value of each of the service data distribution units S1 to SN is calculated with reference to the score table of the CPU usage rate shown in FIG. 4 and the score table of the internal memory usage rate shown in FIG.

The CPU utilization rate score table shown in FIG. 3 is stored, for example, in the memory of the representative port unit 1. This score table includes a plurality of types of CPUs (in this example, "Pentium1").
00MH ”,“ Pentium133MH ”,“ PentiumII160MH ”,“ Pe
ntiumII233MH ”,“ PentiumII400MH ”,“ PentiumIII60
With respect to six types of CPUs such as “0MH”, the scores of the base evaluation of each CPU (“100” and “13
3 "," 160 "," 233 "," 400 "," 600 ") and CP
A multiplication coefficient when the U usage rate is 10% (%) or less (“0.9” common to all CPUs) and a multiplication coefficient when the CPU usage rate is 20% (all CPs
U and “0.8”, and a multiplication coefficient when the CPU usage rate is 30% (“0.
7 ") and a multiplication coefficient (" 0.6 "common to all CPUs) when the CPU usage rate is 40% or more are stored in association with each other.

The score table of the internal memory usage rate shown in FIG. 4 is stored in, for example, the memory of the representative port unit 1. The score table includes a plurality of types of internal memories (in this example, "sim64M1n"). ”,“ Sim64M2n ”,“ sim64M3n ”,“ si
6 such as “m128M1n”, “sim128M2n”, “sim128M3n”
Type of internal memory), the score of the base evaluation of each internal memory (“64 * 1.0”, “64 * 0.8”,
“64 * 0.7”, “128 * 1.0”, “128 * 0.8”, “128 * 0.
7 "), a multiplication factor when the internal memory usage rate is 10% (%) or less (" 0.9 "common to all internal memories), and a multiplication factor when the internal memory usage rate is 30% (%). “0.
7 ”) and the multiplication coefficient when the internal memory usage rate is 50% (“ 0.5 ”for all internal memories)
And a multiplication coefficient when the internal memory usage rate is 70% (common to all internal memories is “0.3”) and a multiplication coefficient when the internal memory usage rate is 80% or more (common to all internal memories) And “0.2”) are stored in association with each other.

Specifically, the representative port unit 1 calculates load distribution evaluation values of the service data distribution units S1 to SN in the following manner. That is, in the representative port unit 1,
For example, the type of the CPU and the type of the internal memory provided in each of the service data distribution units S1 to SN are stored in the memory, and the service data distribution units S1 to SN are provided in the service data distribution units S1 to SN. Multiplied by multiplying the score of the base evaluation of the CPU by a multiplication coefficient according to the CPU usage rate, and the service data distribution units S1 to S
N and a multiplication result obtained by multiplying the score of the base evaluation of the internal memory provided by N with a multiplication coefficient corresponding to the internal memory usage rate, and calculates the total value calculated in this manner in each service data distribution unit. S1 to SN are load distribution evaluation values.

As an example, the CP provided in the service data distribution units S1 to SN for which the load balancing evaluation value is calculated is provided.
The type of U is “Pentium133MH” and the usage rate of the CPU is 20%, and the type of internal memory provided in the service data distribution units S1 to SN is “sim64M2n” and the usage of the internal memory is When the rate is 50%, the load balancing evaluation value is 132 (=
133 * 0.8 + 64 * 0.8 * 0.5).

When the CPU utilization is a value other than the utilization shown in the score table of FIG. 3 or when the internal memory utilization is a value other than the utilization shown in the score table of FIG. For example, the multiplication coefficient set for the closest usage rate is used for calculating the load distribution evaluation value. Further, the score table of the CPU usage rate shown in FIG. 3 and the score table of the internal memory usage rate shown in FIG. 4 are not necessarily limited to those shown in this example, and may be various. It is also possible to use a value in which a multiplication coefficient is set while being divided into finer usage rates (percentages). Further, these score tables may be configured to be rewritable by the representative port unit 1 in accordance with, for example, the usage status of the system.

The load distribution evaluation values calculated as described above are, for example, the service data distribution units S1 to SN
Is stored in the load balancing search table held in the memory of the representative port unit 1 in association with the identification information (machine ID) of the representative port unit 1. In the representative port unit 1, for example, in a state where the load distribution evaluation values of all the service data distribution units S1 to SN are stored in the load distribution search table, the service data distribution units S1 to SN having the largest load distribution evaluation value.
And selects the service data distribution units S1 to SN as service data distribution units with the smallest load. In this way, the service data distribution unit S with the smallest load
When 1 to SN are selected, as described above, the representative port unit 1 transmits a display request from the client device 4 to the LAN 2
To the service data distribution units S1 to SN via the network.

Next, the service data distribution units S1 to S
A specific example of a process of transmitting display data to the client device 4 in response to a display request from the client device 4 that N receives via the representative port unit 1 will be described. That is,
When each of the service data distribution units S1 to SN receives a display request from the client device 4 transmitted via the representative port unit 1 via the LAN 2, for example, it analyzes the display request and responds to the display request. The data (display data) of the service data display screen for displaying the service data is created, and the created display data is transmitted to the client device 4 via the LAN 2 or the Internet 3 (process T4).

When the service data distribution units S1 to SN transmit the display data to the client device 4 as described above, the execution rate of the transmission processing changes the CPU usage rate and the internal memory usage rate. Even if such a change occurs, each of the service data distribution units S1 to SN obtains its own machine load information and updates the machine load information, and transmits the updated machine load information to the representative port unit. Notify 1. Then, in this example, the machine load information managed by the representative port unit 1 is updated by such real-time feedback of the machine load information, and the updated information is used in the next distribution process of the display request from the client device 4. Is effectively reflected in

As described above, in the data distribution system of this embodiment, in response to the display request transmitted from the client device 4 to the representative port unit 1 via the Internet 3,
The representative port unit 1 includes a plurality of service data distribution units S1 to S
The display data is transmitted to the client device 4 via the LAN 2 or the Internet 3 by the service data distribution units S1 to SN selected from N.

At this time, in the data distribution system of the present embodiment, information on the CPU usage rate and the internal memory usage rate directly connected to the performance of each service data delivery section S1 to SN is always managed by the representative port section 1, Since the service data distribution units S1 to SN for performing data transmission are selected based on the load distribution evaluation value determined using these pieces of information, an accurate load matching the actual load of each service data distribution unit S1 to SN is selected. Display requests from a large number of client devices 4 on the basis of various loads.
To SN, whereby the loads on the service data distribution units S1 to SN can be automatically and evenly distributed, for example.

In the data distribution system of this embodiment, the CP directly connected to the performance of each of the service data distribution units S1 to SN
Since the service data distribution units S1 to SN with a small load are selected based on the load distribution evaluation value determined by combining both the usage rate of the U and the usage rate of the internal memory, the load used in the selection (this example In this case, the load distribution evaluation value) is accurate, and the accuracy of load distribution can be improved. In addition, the data distribution system of this example is designed to improve the system performance, particularly when the representative port unit is constituted by a computer having a low performance, for example, or when a large number of service data distribution units are provided. In such a case, a great effect can be exhibited.

Here, each service data distribution unit S of this example
In 1 to SN, the display data is transmitted to the client device 4 via the network (in this example, the LAN 2 or the Internet 3) by the transmission processing performed by the CPU provided in the service data distribution units S1 to SN using the internal memory. These functions constitute the transmission means according to the present invention.

Each service data distribution unit S1 of this embodiment
In SN, the function of notifying the representative port unit 1 at a predetermined timing of information on the usage rate of the CPU performing the above-described transmission processing and the usage rate of the internal memory used by the CPU (machine load information). The notification means according to the invention is configured.

Each service data distribution unit S1 of this embodiment
As a preferred embodiment, the configuration in which the notification of the machine load information is performed every predetermined period and the notification is performed when the machine load information is updated is performed. There is no limitation. For example, a notification may be made when the usage rate of the CPU or the usage rate of the internal memory reaches a predetermined value.

In the present embodiment, as a preferable mode, each service data distribution unit S1 to SN voluntarily notifies the representative port unit 1 of the machine load information. Distribution units S1 to S
For example, the representative port unit 1 issues a notification request of machine load information to each of the service data distribution units S1 to SN, while each of the service data distribution units S1 to SN It is also possible to adopt a configuration in which machine load information is notified to the representative port unit 1 in response to a request.

The representative port unit 1 of the present embodiment has a function of managing the CPU usage rate and the internal memory usage rate notified from the service data distribution units S1 to SN using a connection table or the like. Is configured. In addition, in the representative port unit 1 of the present example, the service data distribution units S1 to SN having a small load determined using the CPU usage rate or the internal memory usage rate in response to a display request transmitted from the client device 4. The function of transmitting display data to the client device 4 constitutes a transmission control means according to the present invention.

In the present embodiment, as a preferable mode, the display data is transmitted by the service data distribution units S1 to SN with the smallest load. It is also possible to cause transmission by SN. As an example, a configuration may be adopted in which a predetermined threshold value is provided, and transmission is performed by the service data distribution units S1 to SN (for example, if there is a plurality of load devices) whose load is equal to or less than the threshold value.

Further, each service data distribution unit S1 of this embodiment
SN are configured to transmit display data to the client device 4 in response to receiving the display request from the client device 4, so that the representative port unit 1 of the present example By changing (transmitting) the destination address to the service data distribution units S1 to SN, the display data is transmitted by the service data distribution units S1 to SN.

Here, the configurations of the data distribution system, the management device and the server device according to the present invention are not necessarily limited to those described above, but may be various.
Similarly, various configurations may be used for the client device. As an example, the number of server devices provided in the data distribution system is not particularly limited,
Any number may be used as long as it is plural.

In the present embodiment, the case where the service data distribution units S1 to SN transmit display data in response to a display request from the client device 4 has been described. Such a request (for example, a request for text data, a request for voice data, and the like) may be made, and the data transmitted from each server device may be various data according to the request.

In this example, the display data transmitted from each of the service data distribution units S1 to SN is the representative port unit 1
Although the configuration in which the data is transmitted to the client device 4 without using the client device 4 is described above, the present invention may be configured to transmit data transmitted from each server device to the client device via the management device, for example. .

In this embodiment, each service data distribution unit S1 to SN manages the usage rate of its own CPU and the usage rate of its own internal memory, while the representative port unit 1 manages these service data distribution units S1 to SN. Although the configuration is such that the CPU usage rate and internal memory usage rate of SN are managed, in the present invention,
The CPU usage rate and the internal memory usage rate do not necessarily have to be managed as they are, for example, the CPU usage rate and the internal memory usage rate may be in other forms (for example, a predetermined operation may be performed on the usage rate value). Or a value obtained by converting the value of the usage rate into stages such as A, B, C, D, etc. according to the size of the usage rate, etc.). Includes

In this embodiment, the service data distribution units S1 to S1 are used by using the CPU utilization and the internal memory utilization.
Although the configuration is such that the load of N is managed, the present invention can also be configured to manage the load by using other information such as the file access frequency of each server device. In such a configuration, for example, each server device detects its own CPU usage rate, internal memory usage rate, file access frequency and the like and notifies the management device, and the management device uses each of the notified information to notify each server device. Determine the load on the device.

Further, in the management device and each server device according to the present invention, various processes are executed by the CPU executing the control program stored in the ROM on the hardware resources including the CPU and the internal memory. But,
The present invention can be understood as a computer-readable recording medium such as a floppy disk or a CD-ROM storing such a control program, and the control program is input to the computer from the recording medium and executed by the CPU. Thereby, the processing according to the present invention can be performed.

[0052]

As described above, according to the data distribution system of the present invention, in response to a request transmitted from a client device, a management device transmits data to a client device by a server device selected from a plurality of server devices. When each server device performs transmission processing,
While notifying the management device of the usage rate of the PU and the usage rate of the internal memory to the management device at a predetermined timing, the management device manages the usage rate of the CPU and the usage rate of the internal memory notified from each server device, and Since the data is transmitted to the client device by the server device having a small load determined using the rate and the usage rate of the internal memory, the load of the server device is determined based on the exact load that matches the actual load of each server device. Can be distributed, and efficient load distribution can be realized.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an example of a data distribution system according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a specific example of a process performed by the data distribution system.

FIG. 3 is a diagram illustrating an example of a score table of CPU usage rates.

FIG. 4 is a diagram illustrating an example of a score table of an internal memory usage rate.

FIG. 5 is a diagram showing an example of a data distribution system according to a conventional example.

[Explanation of symbols]

1. Representative port section, S1 to SN service data distribution section, 2. LAN, 3. Internet, 4.
..Client devices,

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) H04L 12/40

Claims (2)

[Claims] 1. A data that causes a management device to transmit data to a client device via a network by a server device selected from a plurality of server devices in response to a request transmitted from the client device to the management device via the network. A transmission system for transmitting data to a client device via a network by a transmission process performed by a CPU using an internal memory, a server device, and a CPU usage rate and an internal memory usage rate. Notification means for notifying the management device at the timing of (i), wherein the management device manages the usage rate of the CPU and the usage rate of the internal memory notified from each server device, and is transmitted from the client device. CPU on request
And a transmission control means for transmitting data to the client device by a server device having a small load determined using the usage rate of the internal memory and the usage rate of the internal memory. 2. A server device selected from a plurality of server devices in response to a request transmitted from the client device via a network, performs transmission processing using an internal memory to transmit data to the client device via the network. A management device to be transmitted to the device, wherein the CPU is notified from each server device at a predetermined timing.
Management means for managing the usage rate of the internal memory and the usage rate of the internal memory; and a CPU in response to a request transmitted from the client device.
And a transmission control means for transmitting data to the client device by a server device having a small load determined using the usage rate of the internal memory and the usage rate of the internal memory.