JP2001290787A - Data distribution method and storage medium storing data distribution program - Google Patents

Abstract

(57) [Summary] [Problem] To avoid storing data that is rarely accessed or not accessed from a client to a mirror server, and to prevent load of mirror server and line between mirror server and client. A data distribution method and a storage medium storing a data distribution program that allow a mirror server that distributes data to a client to be selected in consideration of congestion. The present invention does not directly distribute data from a data server to a client, but uses a cache having a small load and a small delay in communication time with the client from a plurality of cache servers at the time of data distribution. Select a server, distribute the data to the cache server,
Distribute data from the cache server to the client.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a data delivery method and a storage medium storing a data delivery program, and more particularly to a data delivery method and a data delivery program for delivering data from a server to a client via a computer network. It relates to the stored storage medium.

[0002]

2. Description of the Related Art When a relatively large amount of data is distributed from a server to a client via a computer network, the data is directly distributed from the server in response to a request from the client. Increase and take longer to distribute data.

[0003] Even if the load is not concentrated on the server, even if the communication line between the server and the client is congested, it takes time to distribute the data. Routers, which are devices that relay communications in computer networks, are equipped with a function to select a route so as to minimize the number of routers that pass through when communicating between servers and clients. It is not a mechanism based on parameters that change in real time.

In order to solve some of these problems, conventionally, a server called a mirror server, which duplicates and accumulates the same data as the server and distributes the data in place of the server, is connected to a plurality of different locations, and the client is connected to the server. for,
A method has been used in which data distribution is performed from the nearest mirror server to avoid concentration of load on the server and avoid mixed lines.

[0005]

However, in the above method using a mirror server, the mirror server must store the same amount of data as the server. Also,
There is a problem that when server data is updated, it must be reflected on all mirror servers.
In this case, if the amount of data stored in the server increases, the amount of data stored in the mirror server also increases, and depending on the mirror server, data that is accessed less frequently or never accessed by the client may be stored. Therefore, the process of transferring data from the server to the mirror server in advance and the storage area for storing the transferred data are wasted. In addition, when selecting the nearest mirror server based on the number of routers that pass through, or based on the geographical distance between the server and the client, the load state of the mirror server, the mirror server and the client The congestion status of the line between them is not considered.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and avoids storing data that is rarely accessed by a client or that is not accessed by a mirror server. A data distribution method and a storage medium storing a data distribution program that allow a server (cache server) that distributes data to a client to be selected in consideration of a line congestion state between a client and a mirror server and a client. The purpose is to:

[0007]

Means for Solving the Problems The present invention (claim 1) provides:
In a data distribution method for distributing data from a data server to a client via a computer network,
Selecting a cache server having a small load and a small delay in communication time with the client from the plurality of cache servers without directly distributing data from the data server to the client; And the data is distributed from the cache server to the client.

According to the present invention (claim 2), when selecting a cache server for distributing data, a client uses a response time of the cache server and the number of relay routers between the client and the cache server. To determine the cache server.

According to the present invention (claim 3), when selecting a cache server to which data is to be distributed, the client uses:
A primary candidate of the cache server is selected based on the response time of each cache server and the information of the number of relay routers collected in advance, and the primary candidate is transmitted to the data server.
From the primary candidates for the cache server obtained from the client, the cache server 2
The next candidate is selected and transmitted to the client. At the client, the current response time of the secondary candidate cache server obtained from the data server is measured, and the cache server having the shortest response time is transmitted to the client. It is determined as a cache server to perform.

According to the present invention (claim 4), when data requested from a client is not copied and stored in the cache server when data is distributed from the cache server to the client, the data stored in the cache server is stored in the cache server. If the number exceeds a certain number, data with a low access frequency is deleted from the data stored in the cache server based on the last access time and the number of accesses, and the data is acquired from the data server. Accumulate.

The present invention (claim 5) is a storage medium storing a data distribution program mounted on a data server for distributing data to a client via a computer network, wherein the data server directly distributes data to the client. Instead, there is a process of selecting a cache server with a small load and a small delay in communication time with the client during the data distribution from a plurality of cache servers, and distributing data to the cache server.

[0012] The present invention (claim 6) provides a cache server based on the response time of each cache server and the current load status of the primary candidate of the cache server selected by the information on the number of relay routers, obtained from the client. And transmitting the secondary candidate to the client.

[0013] The present invention (claim 7) is a storage medium storing a data distribution program mounted on a cache server that distributes data obtained from a data server to a client via a computer network. If the number of stored data exceeds a certain number when the copied data is not copied and stored, if the stored data has a low access frequency based on the last access time and the number of accesses, It has a process of deleting data, a process of acquiring and storing data from a data server, and a process of transmitting data requested by a client.

[0014] The present invention (claim 8) is a storage medium storing a data distribution program mounted on a client that acquires data acquired from a data server via a computer network. Determining a cache server based on the number of relay routers between the client and the cache server.

The present invention (claim 9) provides a process of selecting a primary candidate of a cache server based on information of the response time of each cache server and the number of relay routers collected in advance and transmitting the primary candidate to the data server; Measuring the current response time of the secondary candidate cache server obtained from the data server, and determining the cache server with the shortest response time as the cache server that performs data distribution.

As described above, the present invention provides a server for storing data excluding data stored in a mirror server that is not frequently accessed by a client or is not accessed. Multiple cache servers) in place of the mirror server at different locations in the computer network, taking into account server load conditions and line congestion with the client,
It is possible to select a server that distributes data to the client.

[0017]

FIG. 1 shows the configuration of a data distribution system according to the present invention. The data distribution system shown in FIG.
It comprises a computer network 10, a data server 20, a cache server 30, and a client 40. In the computer network 10, one data server 20 and one or more cache servers 30 are connected to different locations, and one or more clients 40 are connected.

Normally, the data server 20 is connected to the computer network 10 by a high-speed communication line. The cache server 30 is connected to the computer network 10 by a relatively high-speed communication line. The communication line connecting the client 40 to the computer network 10 may be a normal low-speed line. Further, the cache server 30 and the client 40 may be integrally constructed.

The data server 20 has data 21, a data list 22, and a cache server management table 23. In the case of a system configuration in which the cache server 30 and the client 40 are integrated, a cache server selection process described later is performed. There is no cache server management table 2
3 becomes unnecessary. Data 2 held by the data server 20
1 is the original of all data to be distributed. In the data list 22, all data names stored in the data server 20 are registered.

The cache server management table 23 is a table in which, for all the cache servers 30 connected to the computer network 10, a cache server name, a cache server address, and the number of clients to which the cache server is currently delivering data are registered. is there. As the cache server address, for example, the IP address of the cache server is registered.

The cache server 30 has data 31 and a stored data management table 32. The data of the cache server 30 is a copy of a part of the data stored in the data server 20. Stored data management table 32
Is a table in which data names, last access times, and access counts are registered for all data stored in the cache server 30. The data name indicates a data name stored in the cache server 30. Since the content of the stored data management table 32 is updated in response to a data distribution request from the client 40, the cache server 30
The contents of the accumulation management table 32 are different for each.

The client 40 has a cache server list 41. However, in the case of a system configuration in which the cache server and the client are integrated, the cache server selection table 41 described below is not required because the cache server selection process described later is not performed. Becomes The cache server list 41 includes a cache server name, a cache server address, a response time, the number of updates of the response time, and the number of relay routers.

Next, an outline of the operation in the above configuration will be described.

FIG. 2 is a diagram for explaining the outline of the operation of the present invention.

In the cache server 30, a part of the data is obtained in advance from the data server 20 (step 1), and the data is stored (step 2). Note that the cache server 30 may acquire necessary data from the data server 20 when there is a request from the client 40 without performing the process. In addition, in the process, when the number of stored data in the cache server 30 exceeds a certain number, data with a low access frequency is deleted based on the last access time and the number of accesses, and then the data is acquired from the data server 20. Shall be.

When the client 40 is the cache server 3
When 0 is selected, candidates for the cache server are narrowed down based on the response time of each cache server and the number of relay routers collected in advance (step 3), and the candidates are transmitted to the data server 20 (step 4). . In the data server 20, for the cache server candidate acquired from the client 40, the current cache server 30
The load status of the cache server 3 with a light load
0 is set as a cache server candidate (step 5) and passed to the client 40 (step 6). Thus, the client 40 measures the current response time of the cache server candidate passed from the data server 20 (step 7), and selects the cache server 30 having the shortest response time as a data acquisition target (step 8). ), And makes a data distribution request to the cache server 30 (step 9). As a result, the cache server 30 that has acquired the data distribution request searches for data corresponding to the request and provides the data to the client 40 (step 10). However, if the requested data is not stored, the data is acquired from the data server 20 as described above.

[0027]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 3 is a flowchart of the client preparation process of the present invention. This is performed at least once for each client 40 prior to data distribution described later with reference to FIG.

The client 40 acquires the contents of the data list from the data server 20 (step 101).
In the case of a system configuration in which the cache server and the client are integrated, there is no cache server selection process described later, and thus the preparation process ends here. Otherwise, the client 40 sends the cache server name and the cache server address registered in the cache server management table from the data server 20 to the data server 2.
It is obtained from 0 (step 102). Client 40
, Information on the response time of each cache server 30 and the number of relay routers up to each cache server 30 is collected and registered in the cache server list 41.
The measurement of the response time is performed, for example, by using the cache server management table 4
Using the cache server address of one, an ICMP (Internet Control Message Protocol (standardized protocol)) echo message is transmitted from the client 40 to each cache server 30, and the time until the echo response message is returned is determined. Measure at client 40. The number of relay routers up to each cache server 30 is acquired using, for example, a cache server address and RIP (route information protocol (standardized protocol)) in the cache server management table (step 103). The initial value of the number of updates of the response time in the cache server list is set to 1. If information on the response time of each cache server 30 and the number of relay routers up to each cache server 30 has been collected before (step 104, Yes), the response time is calculated by the following formula (cache server Response time registered in list x
Number of updates of the response time registered in the cache server list + currently measured response time of the cache server) ÷
A new response time is calculated and updated based on (the number of updates of the response time registered in the cache server list + 1). this is,
This is the average value of the response time (step 107). The value of the response time update count is increased by 1 (step 108). If the number of relay routers is different from the previous one, it is updated with that value.

If the information has not been collected before (step 104, No), the response time of the collected information and the number of relay routers are registered in the cache server list 41 (step 105). The initial value of the number of updates of the response time in Table 41 is set to 1 (step 10).
6).

Next, the flow of data distribution will be described.

FIG. 4 is a flowchart of the data distribution processing of the present invention.

The client 40 displays the data list acquired from the data server 20 (step 20).
1) The user of the client 40 selects data (step 202). The client 40 performs a cache server selection process (step 203), and performs a data distribution process from the determined cache server 30 to the client 40 (step 204). In the case of a system configuration in which the cache server and the client are integrated, the data delivery processing is performed from the cache server integrated with the client without performing the cache server selection processing.

Next, the process of selecting a cache server in step 203 will be described.

FIG. 5 is a sequence chart of the cache server selecting process according to the present invention.

In the client 40, from the cache server list, when there are a plurality of identical routers in the order of shorter response time, a certain number (for example,
(5) cache server names are transmitted (step 30).
1). In the data server 20, among the cache server names acquired from the clients 40, in the cache server management table 23, those having a small number of clients currently delivering data are reduced to a certain number (for example, three), and the cache server names are changed. The message is transmitted to the client 40 (step 302). In the client 40,
The current response time of the cache server 30 with the cache server name acquired from the data server 20 is measured (Step 303). The one with the shortest response time is determined as the cache server 30 that performs data distribution (step 304). In the client 40, the response time of the cache server 30 that has performed the measurement is updated to the calculated value in the cache server list 41 by using the above-described formula to calculate a new response time. This is the average of the measured values of the relevant cache server up to the present. The number of times of updating the response time in the cache server list 41 is increased by one. These are used in the next cache server selection process. The client 40 requests data transmission to the cache server 30 determined as one (step 305).

Next, a data distribution process of the cache server 30 which has received a data distribution request from the client 40 will be described.

FIG. 6 is a sequence chart of a data distribution process of the cache server according to one embodiment of the present invention.

The client 40 is the cache server 3
A data delivery request is made by designating a data name for 0 (step 401). If the data requested by the cache server 30 has already been copied and accumulated from the data server 20 (step 402, Yes), the data is currently being distributed for the corresponding cache server name in the cache server management table 23 of the data server 20. Is increased by one (step 408), the last access time of the stored data management table 23 for the requested data is updated to the current time, the number of accesses is increased by one (step 409), and the data is transmitted from the cache server 30 to the client. The data is distributed to the server 40 (step 410). At the end of distribution,
For the corresponding cache server name in the cache server management table 23 of the data server 20, the number of clients currently delivering data is reduced by 1 (step 411).

In the case of a system configuration in which the cache server and the client are integrated, the data server 20 does not have the cache server management table 23 because there is no cache server selection process. Is not performed.

If the data requested by the cache server 30 is not copied and stored from the data server (step 402, No), the data is copied and stored from the data server 20 to the cache server 30 (step 40).
6), the data name of the stored data, and the number of accesses is 0
Is registered in the stored data management table 32 (step 407),
Then, the cache server management table 2 of the data server 20
3, the number of clients currently delivering data is increased by 1 (step 408). Further, the last access time and the number of accesses in the stored data management table 32 of the cache server 30 are updated, the data is distributed to the client 40, and the data server 2
The number of clients currently delivering data in the cache server management table 23 of 0 is reduced by 1 (step 41).
1).

When the number of data stored in the cache server 30 exceeds a certain number at the time of data distribution from the data server 20 to the cas server 30, the cache server 30
According to the stored data management table 32, the last access time is in the order of chronological order, and when there are a plurality of the same ones (step 403, Yes), the accumulated data If there are multiple files, the files are deleted in ascending order of access count, registration information on the deleted data is deleted from the stored data management table 32 (step 404), and then data is obtained from the data server 20 (step 405). .

According to this method, the cache server 30 that has never been accessed by the client 40 has
The data is copied and stored from the data server 20, and is sequentially updated to a certain number of data with a high access frequency from the client 40 and stored.

According to the present invention, specifically, in the distribution of music data to a personal computer connected to the Internet, the server storing the music data corresponds to the data server. A mechanism can be applied.

Also, a user's storage medium (MD (mini disk), flash memory, etc.) is inserted into a device called a kiosk terminal installed on the street, and music, video,
In a system for distributing data such as game software, the above-described mechanism in which the cache server and the client are integrated with each other can be applied to the kiosk terminal.

The operation of each of the above devices (client, data server, cache server) is constructed as a respective program, and is connected to each device or built in a memory, a disk device, or a floppy (registered). The present invention can be easily realized by storing it in a portable storage medium such as a trademark (trademark) disk, a CD-ROM, or the like and installing it when implementing the present invention.

[0047]

As described above, according to the present invention, even when access to the same data as that of the data server from a plurality of clients is concentrated, a process for actually delivering data to the client is performed in a plurality of processes. Distributed to cache servers.

The cache server stores frequently accessed data, so that it is not necessary to store all the same data as the data server, so that the resources can be effectively used.

Further, at the time of data distribution from the cache server, the client can receive data distribution from the cache server which has a smaller load than other cache servers and has a small delay in communication time due to line congestion.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a data distribution system of the present invention.

FIG. 2 is a diagram for explaining an outline of an operation of the present invention.

FIG. 3 is a flowchart of a client preparation process according to an embodiment of the present invention.

FIG. 4 is a flowchart of a data distribution process according to an embodiment of the present invention.

FIG. 5 is a sequence chart of a cache server selection process according to an embodiment of the present invention.

FIG. 6 is a sequence chart of a data distribution process of the cache server according to the embodiment of the present invention.

[Explanation of symbols]

 10 Computer Network 20 Data Server 21 Data 22 Data List 23 Cache Server Management Table 30 Cache Server 31 Data 32 Stored Data Management Table 40 Client 41 Cache Server List

 ────────────────────────────────────────────────── ─── Continued on the front page (72) Inventor Susumu Nomura 1-6, Uchisaiwaicho, Chiyoda-ku, Tokyo NTT Communications F-term (reference) 5B045 GG02 GG04 5B082 FA12 HA05 HA08 5B089 GA11 JA33 JB14 KA06 KA07 KB03 MA03 MA07

Claims (9)

[Claims] 1. A data distribution method for distributing data from a data server to a client via a computer network, wherein the data is not directly distributed from the data server to the client, but the data is distributed from a plurality of cache servers. A data server comprising: selecting a cache server having a small load and a small delay in communication time with the client; distributing data to the cache server; and distributing the data from the cache server to the client. Method. 2. When selecting a cache server that distributes data, at the client, a cache server is selected based on a response time of the cache server and the number of relay routers between the client and the cache server. 2. The data distribution method according to claim 1, wherein the data distribution is determined. 3. When selecting a cache server to which data is to be distributed, the client selects a primary candidate of the cache server based on information on the response time of each cache server and the number of relay routers collected in advance. , Transmitting to the data server, the data server selects, from the primary candidates for the cache server obtained from the client, a secondary candidate for the cache server based on the current load condition, and transmits the secondary candidate to the client. Measuring, at the client, the current response time of the secondary candidate cache server obtained from the data server, and determining the cache server having the shortest response time;
2. The data distribution method according to claim 1, wherein the data distribution method is determined as a cache server that performs data distribution. 4. When delivering data from the cache server to the client, if the data requested by the client is not copied and stored in the cache server, the number of data stored in the cache server is If the number exceeds a certain number, of the data stored in the cache server, delete the data with a low access frequency based on the last access time and the number of accesses, and obtain the data from the data server. 2. The data distribution method according to claim 1, wherein the data is stored. 5. A storage medium storing a data distribution program mounted on a data server that distributes data to a client via a computer network, wherein the data distribution program is not directly distributed to the client, but is stored in a plurality of cache servers. A data distribution program that has a process of selecting a cache server with a small load and a small delay in communication time with the client at the time of data distribution, and having a process of distributing data to the cache server. Storage medium. 6. A cache server secondary candidate is selected based on a response time of each cache server and a current load status of a primary candidate of a cache server selected based on information on the number of relay routers, obtained from the client. do it,
6. A storage medium storing the data distribution program according to claim 5, including a process of transmitting the data distribution program to the client. 7. A storage medium storing a data distribution program mounted on a cache server that distributes data obtained from a data server to a client via a computer network, wherein the data requested by the client is copied. A process of deleting data with a low access frequency based on the last access time and the number of accesses from the stored data when the number of stored data exceeds a certain number when the data is not stored; A storage medium storing a data distribution program, comprising: a process of acquiring and accumulating data from the data server; and a process of transmitting data requested by the client. 8. A storage medium storing a data distribution program mounted on a client that acquires data acquired from a data server via a computer network, the response time of the cache server, and the client and the cache server. A storage medium storing a data distribution program, characterized by having a process of determining a cache server based on the number of relay routers between the servers. 9. A process of selecting a primary candidate of a cache server based on information of a response time of each cache server and the number of relay routers collected in advance, transmitting the primary candidate to the data server, and acquiring the primary candidate of the cache server from the data server. 9. The storage according to claim 8, further comprising: measuring a current response time of the secondary candidate cache server, and determining a cache server having the shortest response time as a cache server performing data distribution. Medium.