JP3651782B2 - Server selection device and server selection program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a server selection device and a server selection program, and more particularly to a technique effective in selecting an information providing source server in a shorter time in an information service system using a plurality of servers.
[0002]
[Prior art]
Information converted into computer data is called a file, and means for storing the file is called a file system.
This file system holds an attribute information table for each file to be stored, and has an update time field as one field of this attribute information table.
This update time field indicates the time when the file was created and updated. However, this field can also be set to some other value by an operation or program by the file system administrator.
As an information service provided by providing a server holding a file system on the Internet, a file distribution service that distributes a file describing a home page or a digitally encoded music file is known.
As another example of an information service, a database service that notifies a result of referring to a file held by a file system by a database program is known.
[0003]
A method of storing a copy of an original file on multiple servers and providing information services using multiple servers is called mirroring, a server holding the copy is called a mirror server, and a server holding the original file is called an original server. .
Since the number of services that can be executed simultaneously by one server is limited, mirroring is effective as a countermeasure when a large number of users make service requests to a server at the same time.
In addition, by distributing the mirror servers to different bases, the data transfer load (traffic) generated in the network can be reduced as compared with the case where distribution is performed from a single server.
As described above, mirroring is an indispensable technique for achieving server load distribution and network traffic reduction necessary for providing services to many users.
[0004]
In a typical mirror server operation mode, mirroring is set in units of appropriate directories, and a directory serving as a unit of mirroring setting is called a control unit directory.
In the mirror server, the address of the original server and the path name on the original server are registered for each control unit directory.
In accordance with this registration, the mirror server receives a file from the original server and generates a directory including the same file as the replication target directory of the original server.
The replication target directory of the original server is called the original directory, and the directory generated on the mirror server is called the replication directory.
[0005]
The mirror server receives a file from each original directory using a file transfer protocol such as FTP (File Transfer Protocol) from the original server. This process is called a mirroring process, and this mirroring process is executed periodically.
An event in which a new file is created in the original directory or a file in the original directory is updated is referred to as an update of the original directory.
Here, the following points should be noted.
Even if the original directory is updated, the mirror server directory is not updated instantaneously.
In other words, the file on the original server and the file on the mirror server have different contents until the next cycle for executing the mirroring process comes and when the update file is transferred. Such a state is called update deviation.
[0006]
On the other hand, information services on the Internet are identified by service identifiers.
In particular, for file distribution services, a service identifier format called URL (uniform resource locator) is defined.
The URL is composed of an identifier of a server that distributes the file and a file path in the server.
The file provider advertises the URL to the user, and the user inputs the advertised URL to the terminal. The terminal makes a file request to the server indicated by the input URL and receives the file.
In file distribution using a mirror server, in order for a terminal to receive a file from the mirror server, it is necessary to input a URL including an identifier of the mirror server to the terminal.
However, in this mechanism, it is necessary to advertise URLs as many as the number of mirror servers for files to be mirrored.
[0007]
In order to solve this, there is a method of providing a service from a mirror server while advertising only one URL as a representative URL.
This method is called redirection, and the mechanism of this redirection will be described below.
In the following description, a time stamp of a certain file f stored in a certain server s is denoted as T (f, s), and a control unit directory to which the file f belongs is denoted as D (f).
An original server of a certain directory d is So (d), a list of servers holding the original directory of d or a replication directory is denoted L (d), and a server identifier part of a URL r is denoted r. s and the file indicated by r is r. Indicated as f.
As an explanation of redirection, a mechanism for distributing a certain file F that is mirrored will be described with reference to FIG.
[0008]
The service provider The URL R with f = F is advertised as the representative URL, and the user inputs this URL R from the terminal 4.
The terminal 4 first starts with the server identifier R.P. s and server R.R. A file request message including URL R is transmitted to s (2) (1. service request shown in FIG. 12).
In FIG. 12, the server identifier R.I. The case where the server having s is the original server So (d) is illustrated.
Accordingly, the server R.I. s (2) is obtained from the service identifier R to the request file R.P. Recognize f = F, recognize the control unit directory D (F) to which the file F belongs, and select an appropriate server from the list L (D (F)) of mirror servers that hold a copy of the directory D (F). (Information providing source server) is selected (2. server selection process shown in FIG. 12).
[0009]
Here, the identifier of the selected server is represented as Sz.
Server R. s changes the new URL P.P. s = Sz, P.I. It generates as f = F and transmits a redirect message including this URL P as a response to the terminal (3. Redirect shown in FIG. 12).
The terminal 4 that has received the redirect message receives the server identifier P.P. from the URL P included in the message. s = Sz is acquired, and a file request message is transmitted to the server Sz (3) (4. Content request shown in FIG. 12).
In response to this, the server Sz (3) returns a normal response message (5. Normal response shown in FIG. 12), and subsequently transmits the file F.
In the above procedure, the server indicated by the representative URL, that is, the server R.P. s (2) is called a representative server.
[0010]
The representative server 2 receives the representative URL R and inputs the file R.R. The server group L (D (F)) capable of distributing f = F is used as a candidate server, and includes means for selecting an appropriate server from among the candidate servers.
It is desirable that the server selection device 1 does not select a mirror server in which a file update error has occurred.
Therefore, the server selection device 1 for each server s included in the compensation server L (D (F)), the time stamp T (F, s) of the file F held in the server s, and the original server So ( The time stamp T (F, So (D (F))) of the file F held in D (F)) is compared, and T (F, s) becomes T (F, So (D (F))). The server s is not selected if it is older than ().
In the conventional server selection apparatus, the value of the update time field of the file attribute information is used as the file time stamp.
[0011]
FIG. 13 is a block diagram illustrating a schematic configuration of a general server selection device 1.
The server list table 14 is a table that stores a server list L (d) for each directory d in which the server selection device 1 performs server selection.
The server list reference unit 15 obtains the server list L (D (R.f)) corresponding to the control unit directory D (R.f) to which the file (R.f) indicated by R belongs from the input URL R as the server. It is means for obtaining a reference from the list table 16. The original time stamp acquisition means 16 stores the file R.D. stored in the original server So (D (R.f)). This is a means for acquiring the time stamp T (R.f, So (D (R.f))) of f.
In this specification, this time stamp T (R.f, So (D (R.f))) is referred to as an original time stamp, and the original time stamp obtained from the service identifier R is referred to as To (R).
The duplicate file time stamp acquisition means 13 is a means for acquiring the time stamp T (R.f, s) of the file R.f on the server s in the server list L (D (R.f)).
Server narrowing means 12 is means for excluding servers s whose T (R.f, s) is older than To (R), and 11 is decision means for selecting an arbitrary server from among a plurality of servers.
[0012]
A typical home page on the Internet is described in a format called HTML. HTML is a file for describing the structure and text of a home page.
When the home page includes an image, the HTML file includes an identifier of the image file to be displayed. Thus, describing an identifier of another file in a certain file is called a link reference.
When displaying such a home page, the user terminal first receives an HTML file, and then receives an image file that is linked to the HTML file.
A file identifier in a link reference has one of the following three formats.
The first format is a file path format starting with “/”.
The second format is a file path format that does not begin with “/”.
The third format is a URL format.
[0013]
The first form or the second form is called a relative reference form, and the third form is called an absolute reference form.
In the case of the absolute reference format, the user terminal receives a file whose link is referenced from the server indicated by the file identifier.
In the case of the relative reference format, the user terminal receives a link-referenced file from the server that has delivered the HTML file.
Therefore, when a homepage including a link reference in the relative reference format is distributed by the mirror server and the server selection device, the server selection is executed only when the HTML file is distributed.
When receiving a file whose link is referenced by an HTML file, the user terminal transmits a file request to the server notified when the HTML file is received.
That is, when a certain home page is constituted by the HTML file Fl and the image file F2 to which Fl refers to a relative link, the operation of the terminal that receives this file is as follows.
[0014]
Here, the representative service identifier R of the homepage is R.R. s = So (D (F1)), R.I. It is assumed that f = F1.
First, the terminal receives the service identifier R and receives the file F1. Here, when the home page is mirrored, the server So (D (F1)) uses the above-described redirect procedure to cause the server So (D (F1)) to add a new service identifier P (P.P.P) including the identifier of the server Sz selected by the server selection device 1. s = Sz, P.f = F1) to the terminal, and the terminal receives the file Fl from the server Sz.
Next, the terminal interprets the description of the file F1, recognizes a relative link reference to the image file F2, and receives the image file F2.
At this time, the server that distributes the image file F2 is regarded as the server that distributed the file F1.
If the home page is not mirrored, this server s. If mirrored, this server s = Sz.
[0015]
[Problems to be solved by the invention]
The conventional server selection device has the following problems.
The first problem is that the time required for server selection increases because the time stamp of the file held in the mirror server is acquired every time the server selection process is performed.
The second problem is that when a request file references another file in a relative reference format, even if the server selection device performs server exclusion processing based on the time stamp of the request file, the link is referenced by the request file. It is a point that it is not possible to recognize whether there is an update difference for a file that has been updated.
For this reason, when a user terminal receives a file whose link is referenced, an old file may be distributed.
[0016]
A description will be given by using an example in which a home page composed of the HTML file F1 and the image file F2 is distributed using mirroring.
The server selection device selects the server Sz when distributing the file F1.
At this time, it is ensured that T (F1, Sz) is not older than T (F1, So (D (F1))).
However, it is not confirmed that T (F2, Sz) is not older than T (F2, So (D (F1)) for the image file F2.
For this reason, even when the content sender updates only the image file F2 and there is an update shift for the image file F2 in the server Sz, the server selection device selects the server Sz.
As a result, the image file F2 before update is distributed to the terminal.
[0017]
The present invention has been made to solve the above-described problems of the prior art, and an object of the present invention is to provide an information provider server in an information service system using a plurality of servers in a shorter time than in the past. Is to provide a server selection device capable of selecting
Another object of the present invention is to provide the latest version of a relative-referenced file when distributing a file referring to another file in a relative reference format in an information service system using a plurality of servers. Is to provide a server selection device capable of selecting an information providing source server that holds the information.
Another object of the present invention is to provide a server selection program that causes a computer to execute the server selection process described above.
The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings.
[0018]
[Means for Solving the Problems]
Of the inventions disclosed in this application, the outline of typical ones will be briefly described as follows.
That is, according to the present invention, the attribute value of a file indicating the order in which the value between files is stored in the original directory is set as a time stamp, and the latest value of the time stamp of a file belonging to a certain directory is set to Directory time stamp acquisition means for acquiring a directory time stamp of each replication directory of each mirror server in a conventional server selection device, and a directory time stamp acquired by this means for each replication directory of each mirror server. A directory time stamp table to be stored is added, and in the server narrowing-down means, the directory time stamp table is referred to, and the directory time stamp is older than the original time stamp. Characterized in that it excludes the mirror server holding directory from the candidate server.
[0019]
According to the definition of the time stamp and the directory time stamp described above, the order of the time stamps of the files indicates the order in which the files are accumulated in the original server. If it is too old, it can be determined that the replication directory does not hold the file.
The directory time stamp is attribute information for each directory.
Therefore, when the server selection device continuously performs server selection processing for different files in the same directory, there is no need to re-acquire the directory time stamp from the replication directory.
According to the above-described means, the directory time stamp table for storing and holding the directory time stamp for each replication directory is added to the server selection device. Therefore, for each of the reasons described above, the time stamp from the replication directory is selected every time the server is selected. The information providing source server can be selected in a shorter period of time than in the past.
[0020]
Further, according to the present invention, the original time stamp acquisition means acquires the time stamp of the file indicated by the service identifier as an original time stamp, or acquires the directory time stamp of the original directory to which the file belongs as the original time stamp. The original time stamp acquisition means adds a determination means for determining the original time stamp acquisition method of the file, the time stamp of the file indicated by the service identifier stored in the original directory based on the determination result of the determination means, or the original A directory time stamp of the directory is acquired as the original time stamp.
[0021]
Between the time when the service provider updates a file and the time when the file is replicated to a certain replication directory, the directory time stamp of the replication directory is older than the directory time stamp of the original directory.
Accordingly, if the directory time stamp of the original directory is the original time stamp, the server selection device excludes from the candidate server a duplicate directory in which an update error has occurred for any file in the directory to which the file indicated by the service identifier belongs.
As a result, as long as the file F1 and the file F2 referenced by the relative link by the file F1 belong to the same directory, the server holding the latest file F2 can be selected.
[0022]
On the other hand, whether the time stamp of the file indicated by the service identifier or the directory time stamp of the original directory to which the file belongs is appropriate as the original time stamp depends on the type of file to be distributed.
According to the above-described means, since the means for determining the original time stamp acquisition method based on the service identifier is added to the server selection device, a more appropriate original time stamp can be obtained.
In one embodiment of the present invention, the time stamp of the file is defined as the time when the file is stored in the original directory.
In another embodiment of the present invention, an attribute value of the original directory that is added every time an update of the original directory occurs is defined as a version number, and the time stamp of the file indicates that the file has the original directory It is defined as the version number of the original directory immediately after being stored in. The present invention is also a server selection program that causes a computer to execute the server selection process described above.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
Note that components having the same function are denoted by the same reference symbols throughout the drawings for describing the embodiment, and the repetitive description thereof will be omitted.
[Embodiment 1]
FIG. 1 is a block diagram showing a schematic configuration of the server selection device according to the first embodiment of the present invention.
In the present embodiment, directory time stamp acquisition means 17 for acquiring a directory time stamp of each replication directory of each mirror server, and a directory for storing the directory time stamp acquired by this means 17 for each replication directory of each mirror server A time stamp table 18 is added to the conventional server selection device 1.
Further, the server screening unit 12 of the server selection device 1 according to the present embodiment refers to the directory time stamp table 18 and selects a mirror server that holds the replication directory whose directory time stamp is older than the original time stamp as a candidate server. Exclude from
[0024]
Here, a file time stamp represents a file attribute value whose value order matches the order stored in the original directory, and a directory time stamp is a time stamp of a file belonging to a certain directory. Represents the latest value.
In FIG. 1, when a certain file f belongs to a certain directory d, this is represented as fεd.
The time stamp indicates that the larger the value is, the newer the time accumulated in the original directory is.
Here, the directory time stamp of the replication directory on a mirror server s in a certain directory d is denoted as Td (d, s).
The directory time stamp acquisition means 17 acquires the directory time stamp Td (d, s) for the replication directory of each mirror server s of each mirrored directory d.
The directory time stamp table 18 stores a directory time stamp Td (d, s) for each directory d and each mirror server s.
[0025]
The server narrowing means 12 stores the file R.P indicated by the input service identifier R. For a server s (sεL (D (R.f)) belonging to a candidate server L (D (R.f)) of a directory D (R.f) to which f belongs, a replication directory D (R of the server s) . F) of the directory time stamp Td (D (R · f), s) is obtained by referring to the directory time stamp table 18 and more than the original time stamp To (R) obtained by the original time stamp obtaining means 16. When the directory time stamp Td (D (R.f), s) is old (that is, when Td (D (R.f), s) <To (R)), the server s is selected as a candidate server. Exclude from
The list of remaining candidate servers as a result of the exclusion process by the server narrowing means 12 is denoted as L ′ (R).
L ′ (R) satisfies {s | sεL (D (R.f)) and Td (D (R.f), s) ≧ To (R)}.
The server determination unit 11 determines an information provider server from the candidate server list L ′ (R).
[0026]
According to the definition of the time stamp and the directory time stamp described above, the order of the time stamps of the files indicates the order in which the files are accumulated in the original server. If it is too old, it can be determined that the replication directory does not hold the file. The directory time stamp is attribute information for each directory.
Therefore, when the server selection device 1 performs server selection processing continuously for different files in the same directory, there is no need to re-acquire the directory time stamp from the replication directory.
By using this, a directory time stamp table 18 for storing and holding a directory time stamp for each replication directory is added to the server selection device 1 so that it is not necessary to acquire a time stamp from the replication directory each time a server is selected. , Server selection can be executed in a shorter time.
[0027]
Embodiment 2
FIG. 2 is a block diagram showing a schematic configuration of the server selection device according to the second embodiment of the present invention.
In the server selection device 1 according to the second embodiment of the present invention, the original time stamp acquisition unit 16 acquires the time stamp of the file indicated by the service identifier as the original time stamp, or the directory time stamp of the original directory to which the file belongs. Based on the determination result of the determination unit 51, the original time stamp acquisition unit 16 includes a time stamp of the file indicated by the service identifier, or a determination unit 51 for determining whether to acquire the original time stamp. The directory time stamp of the original directory to which the file belongs is acquired as the original time stamp.
That is, in the server selection device 1 of the present embodiment, the original time stamp acquisition unit 16 uses the directory time stamps Td (D (R.f), So (D (R.f)) as the original time stamp To (R). )) Or T (R.f, So (D (R.f)) is adaptively acquired.
[0028]
Also, an original time stamp acquisition method table 53 for each file type that stores the correspondence between the file type and the original time stamp acquisition method, or a directory that stores the correspondence between the directory identifier and the original time stamp acquisition method. Another original time stamp acquisition method table 54 is added to the server selection device 1.
The determination unit 51 acquires the original time stamp acquisition method corresponding to the file type indicated by the service identifier from the file type original time stamp acquisition method table 53 or the directory-specific original time stamp acquisition method table 54.
[0029]
Between the time when the service provider updates a file and the time when the file is replicated to a certain replication directory, the directory time stamp of the replication directory is older than the directory time stamp of the original directory.
Therefore, if the directory time stamp of the original directory is the original time stamp, the server selection device 1 excludes from the candidate server a duplicate directory in which an update error has occurred for any file in the directory to which the file indicated by the service identifier belongs. .
That is, when a homepage composed of the above-described HTML file F1 and the file F2 referenced by the relative link by the file F1 is distributed by mirroring, and an update shift has occurred in the file F2, the original directory 23 If the directory time stamp is the original time stamp, as long as F1 and F2 belong to the same directory, the server that holds the latest file F2 is selected.
[0030]
Whether the time stamp of the file indicated by the service identifier or the directory time stamp of the original directory to which the file belongs is appropriate as the original time stamp depends on the type of file to be distributed. Therefore, in this embodiment, a determination unit 51 that determines an original time stamp acquisition method based on the service identifier is added to the server selection device.
Also in this embodiment, the time stamp of the file is defined as the time stored in the original server.
In each of the above-described embodiments, the attribute of the original directory is defined as a version number that is incremented each time the directory is updated, and the time stamp of the file is stored in the original directory. It can also be defined as the version number of the original directory immediately after
[0031]
Embodiment 3
FIG. 3 is a block diagram showing a schematic configuration of the server device according to the third embodiment of the present invention.
The present embodiment is an embodiment in which the server selection device according to the first embodiment is realized by a computer and a computer program operating on the computer.
The server selection device 1, the original server 2, and the mirror server 3 shown in FIG. 3 are computers having communication functions based on FTP (File Transfer Protocol) and HTTP (Hyper Text Transfer Protocol).
FTP and HTTP are standardized and disclosed by IETF (Internet Engineering Task Force).
The directory timestamp acquisition program 101 and selection request processing acquisition program 102 on the server selection device, the update reception program 21 on the original server, and the mirroring program 31 on the mirror server are computer programs.
Each of these programs is provided on a storage medium such as a CD-ROM. Each of these programs is distributed using a network by a data transfer method such as FTP or HTTP.
[0032]
The directory time stamp acquisition means 17 in the server selection device 1 described above is realized by the directory time stamp acquisition program 101.
Further, the original time stamp acquisition means 16, the server list reference means 15, the server narrowing means 12, and the server determination means 11 are realized as a subroutine of the selection request processing acquisition program 102.
The directory time stamp table 18 is realized by storing a file called a directory time stamp table file in the server selection device 1. The format of the directory time stamp table file will be described later. In the original server 2, an update reception directory 21 is created separately from the original directory 23.
The update reception directory 21 is a directory for receiving a new file or an updated file from the service provider.
The service provider accumulates a new file to be transmitted or an updated file in the update reception directory using FTP.
[0033]
In the present embodiment, the time stamp of the file is the time accumulated in the original directory 23 of the original server 2.
In order to hold this time stamp, the update time field of the file attribute information is set to the time stored in the original directory 23. This setting process is performed by the update acceptance program 22.
Hereinafter, the processing content of the update reception program 22 will be described.
The update reception program 22 periodically monitors the update reception directory 21, and if a file that does not exist in the original directory 23 exists in the update reception directory 21, the file is stored in the original directory 23. At this time, the file update time field is set to the current time.
By this update reception program 22, the file update time field indicates the time stored in the original directory 23.
In the following description, the file time stamp indicates the value of the file update time field.
Whether or not the file exists in the directory is determined by whether or not a file having the same name, size (number of bytes), and update time field value exists in the directory.
[0034]
In the mirror server 3, the MirrorControl file 110 shown in FIG. 4 is created as a mirroring setting file.
Each line of the file 110 indicates a setting parameter of a certain replication directory as a first word is a name of the replication directory, and a second word is an FTP URL for accessing the original directory 23 by FTP.
The mirroring program 31 of the mirror server periodically executes the following synchronization subroutine.
The synchronization subroutine refers to the MirrorControl file 110, accesses the original directory 23 of each replication directory by FTP, obtains a directory list from the original directory 23, and does not exist in the replication directory among the files described in the list. Things are obtained from the original directory 23.
[0035]
In the server selection device 1, a server list table file (server list table 14 in FIG. 1) 111 shown in FIG. 5 and an original directory file 112 shown in FIG. 6 are created.
Each row of the server list table file 111 indicates information on a replication directory on a mirror server where a certain directory exists.
Each line of the original directory file 112 indicates information of the original directory 23 of a certain directory to be mirrored.
The first word in each line of the server list table file 111 or the original directory file 112 indicates a directory name.
The second word in each line indicates an HTTP URL for accessing the directory by HTTP. This URL is called an information distribution URL.
The third word in each line indicates an FTP URL for accessing the directory by FTP. This URL is called a control URL.
[0036]
In the following description, with respect to a certain directory d, obtaining the URL for information distribution or the control URL of the replication directory of a certain mirror server s from the server list table file means that the first word in the line of the file is This means that the second word (information distribution URL) or the third word (control URL) in a row that matches d and the second word includes s is acquired.
In addition, when the information distribution URL or control URL of the original directory 23 of a certain directory d is acquired from the original directory file 112, the first word in the line of the file 112 where the first word matches with d. It means that the second word (information distribution URL) or the third word (control URL) is acquired.
The acquisition of the control unit directory D (R.f) corresponding to a certain representative URL R from the original directory file 112 means that the file path portion (R) of the URL R from among the first words of the original directory file 112 is obtained. .F) means that the longest matching part by the forward matching is acquired.
In addition, obtaining the server list of a directory d from the server list table file 111 refers to the server list table file 111 and obtains the second word for all the rows in which the first word matches d. Means.
[0037]
The directory time stamp table 18 is realized by generating the directory time stamp table file 113 shown in FIG.
Each row of the directory time stamp table file 113 indicates a directory time stamp of a certain replication directory of a certain mirror server.
The first word in each row indicates the directory name (d), and the second word in each row indicates the information distribution URL of the mirror server s.
The third word in each line indicates the directory time stamp Td (d, s) of the replication directory on the mirror server s of the directory d.
In the following description, to acquire the directory time stamp table 18 corresponding to a certain URL for information distribution from the directory time stamp table file 113, the directory time stamp table file 113 is referred to, and the second word is the information This means that the third word in the line that matches the distribution URL is acquired.
[0038]
The directory time stamp acquisition program 101 periodically acquires each directory time stamp of each replication directory of the mirrored directory, and updates the directory time stamp table file 113 based on the acquired directory time stamp. To do.
More specifically, the directory time stamp acquisition program 101 performs processing from the server list table file 111 as processing for acquiring the directory time stamp Td (d, s) of the replication directory on a certain mirror server s for a certain directory d. The control URL of the mirror server s in the directory d is acquired, the replica directory is accessed by FTP using the acquired control URL, the list of the replica directory is acquired, and the time of the file indicated by the acquired list The latest value of the stamp is obtained, and the obtained value is set as the directory time stamp Td (d, s) of the replication directory.
Then, the value of the third word in the row where the first word of the directory time stamp matches d and the second word includes s is updated to the acquired directory time stamp Td (d, s). To do.
[0039]
The selection request processing acquisition program 102 is a program that receives the representative URL R as an input and outputs a URL P that includes the information distribution URL of the selected server, and is activated by the HTTP server program of the representative server.
When receiving the service request from the terminal, the HTTP server program starts the selection request processing acquisition program 102 with the service identifier R included in the service request as an input, and acquires the information distribution URL output by the selection request processing acquisition program 102. R. for the acquired URL. A part of f after D (R.f) is added to generate URL P, and an HTTP redirect message including this URL P is transmitted to the terminal.
The processing contents of the selection request processing acquisition program 102 will be described below.
The selection request processing acquisition program 102 includes a server list reference subroutine, an original time stamp acquisition subroutine, a server narrowing subroutine, and a server determination subroutine.
[0040]
The server list reference subroutine receives the URL R, acquires the control unit directory D (R.f) corresponding to the URL R from the original directory file 112 by the above-described method, and the server list table file 111 by the above-described method. The server list L (D (R.f)) of the control unit directory D (R.f) is acquired from the server, and the acquired server list is output.
For example, when the input URL R is http://www.ntt.co.jp/News/index.html, the control unit directory D (“News / index.html”) = “News” is acquired first. Next, News server list L ("News") = {http://tokyo-server.ntt.co.jp/News, http://osaka-server.ntt.co.jp/News, http: Get //nagoya-server.ntt.co.jp/News}.
[0041]
The original time stamp acquisition subroutine receives the URL R, acquires the control unit directory D (R.f) corresponding to the URL R from the original directory file 112 by the method described above, and acquires the control unit directory D (R.f) from the original directory file 112 by the method described above. The URL for information distribution in the original directory 23 is acquired, and a URL added to the URL for information distribution acquired after D (R.f) of the R file path is generated. The HEAD request message is transmitted, the HEAD response message is received, the time stamp of the file included in the HEAD response message is acquired as the original time stamp To (R), and the acquired original time stamp To (R) is output.
For example, when the input URL R is http://www.ntt.co.jp/News/index.html, first, the control unit directory D (“News / index.html”) = “News” is acquired. Next, http://www.ntt.co.jp/News is acquired as the information distribution URL of the News original directory 23, and /index.html is added to this URL, and the URL http: / /www.ntt.co.jp/News/index.html is generated, a HEAD request message is transmitted to this URL, and as a time stamp included in the HEAD response message, 2001/1/1/11: 30: 00 is output.
[0042]
The server narrowing subroutine receives the server list L (D (R.f)) and the original time stamp To (R), and outputs the server list L ′ (R) obtained as a result of the narrowing.
More specifically, the original time stamp acquisition subroutine performs the following narrowing process on the information distribution URL that is each element of the input server list.
That is, the directory time stamp corresponding to the information distribution URL is acquired from the directory time stamp table file 113 by the method described above, and if the acquired directory time stamp is older than the original time stamp To (R), the information distribution is performed. The URL is excluded from the server list L (D (R.f)).
A server list obtained after narrowing down all information distribution URLs in the input server list is acquired as L ′ (R).
For example, in the above example of R = http: //www.ntt.co.jp/News/index.html, L ′ (R) = {http: //tokyo-server.ntt·co·jp/News, Output http://osaka-server.ntt.co.jp/News}.
[0043]
The server determination subroutine receives the server list L ′ (R) as an input and outputs a URL for information distribution of the determined server.
More specifically, a variable N is held as a variable (static variable) in which a value at the time of the previous subroutine execution is held without being initialized every time the subroutine is started, and each time the subroutine is executed, The value is incremented by 1 and N is divided by the number of information distribution URLs included in the input server list to be K, and the (K + 1) th information distribution URL in the input server list is determined by the server determined. Output as information distribution URL.
For example, if L ′ (R) = {http://tokyo-server.ntt.co.jp/News, http://osaka-server.ntt.co.jp/News} is input, the server determination subroutine Outputs http://tokyo-server.ntt.co.jp/News and http://osaka-server.ntt.co.jp/News alternately.
[0044]
The overall operation of the selection request processing acquisition program will be described.
First, the selection request processing acquisition program 102 calls the server list reference subroutine with the representative URL R as an input, and acquires the server list L (D (R.f)).
Next, the selection request process acquisition program 102 receives the service identifier R, calls an original time stamp acquisition subroutine, and acquires the original time stamp To (R).
Next, the selection request processing acquisition program 102 inputs the acquired server list L (D (R.f)) and the original time stamp To (R) as an input, calls a server narrowing subroutine, and performs the narrowing processing. The list L ′ (R) is acquired.
Next, the selection request processing acquisition program 102 receives the acquired server list L ′ (R) as an input, calls a server determination subroutine, acquires the information distribution URL of the selected server, and outputs it.
[0045]
Hereinafter, as a modification of the present embodiment, an embodiment in which the time stamp of a file is used as the version number of the directory instead of the time when the file is stored in the original directory 23 will be described.
In the original server, a version number file 114 shown in FIG. 8 and a time stamp file 115 shown in FIG. 9 are created for each original directory 23.
Further, the time stamp file shown in FIG. 9 is generated for each replication directory in the mirror server.
These files are automatically updated by the update reception program 22 and the mirroring program 31.
The version number file 114 is a file indicating the version number of the directory.
Each line of the time stamp file 115 indicates the time stamp of the file included in the directory, with the first word being the path of the file in the directory and the second word being the time stamp of the directory.
[0046]
If there is a new file in the update reception directory 21, the update reception program 22 stores the file in the original directory 23.
At this time, the version number file 114 is updated by adding 1 to the value of the version number in the version number file.
Further, the time stamp of the file is entered in the time stamp file 115 using the added version number as the time stamp of the file.
When a line indicating the same file path as the file exists in the time stamp file 115, the time stamp value of the line is rewritten.
[0047]
The mirroring program 31 accesses the original server by FTP by the above-described method, and acquires the time stamp file 115 of the original server.
Among the files indicated by the time stamp file 115 of the original server, those not included in the time stamp file 115 of the replication directory, or the time stamp indicated by the time stamp file 115 of the replication directory is the time stamp file acquired from the original server. A file older than the time stamp indicated by 115 is acquired by FTP, and the acquired file is accumulated in the replication directory.
When the accumulation is completed, the time stamp indicated by the time stamp file 115 acquired from the original server is entered in the time stamp file 115 of the replication directory as the time stamp of the file.
[0048]
When a line indicating the same file path as the file exists in the time stamp file 115, the time stamp value of the line is rewritten.
The directory time stamp acquisition subroutine accesses the replication directory by FTP using the method described above, acquires the time stamp file 115 of the replication directory by FTP, and obtains the latest value of the time stamp included in the time stamp file 115, This value is set as the directory time stamp of the directory.
In the original time stamp acquisition subroutine, the control URL of the original directory 23 is acquired by the method described above, the acquired URL is accessed by FTP, and the time stamp file 115 of the original directory 23 is acquired by FTP. The time stamp of the file indicated by URL R is acquired from the time stamp file 115, and the acquired value is set as the original time stamp.
[0049]
Embodiment 4
In the present embodiment, the server selection apparatus according to the second embodiment is realized by a computer and a computer program that operates on the computer.
Hereinafter, in the present embodiment, a method for determining an original time stamp acquisition method according to a URL will be described.
By saving the file type original time stamp acquisition method table file 116 shown in FIG. 10 in the server selection device 1, the file type original time stamp acquisition method table 53 is realized.
This file 116 lists the extensions of files to be acquired using the directory time stamp of the original directory 23 as the original time stamp.
The original time stamp acquisition subroutine refers to this file 116 and determines whether or not the extension of the file indicated by URL R is described in this file. If it is described, the directory time stamp of the original directory 23 is described. Is obtained as an original time stamp.
[0050]
When acquiring the directory time stamp of the original directory 23 as the original time stamp, the original time stamp acquisition subroutine refers to the original directory file 112 to acquire a control unit directory corresponding to the URL R, and performs control corresponding to the directory. A URL for the file is acquired from the original directory file 112, the original directory 23 is accessed by FTP using the control URL, a list of the directory is acquired, and the latest value of the time stamp of the file indicated by the acquired list is obtained. The obtained value is used as the original time stamp.
[0051]
The following describes an embodiment in which the original time stamp acquisition method is registered for each directory as a modification of the present embodiment.
The directory-specific original time stamp acquisition method table 54 shown in FIG. 11 is stored in the server selection device 1 to realize the directory-specific original time stamp acquisition method table 54.
Each line of the file 117 indicates information of a directory, the first word indicates a directory identifier, and the second word indicates an original time stamp acquisition method.
The original time stamp acquisition subroutine acquires the control unit directory corresponding to the service identifier R by referring to the original directory file 112 by the method described above, and acquires the original time stamp acquisition method corresponding to the directory as the original time stamp acquisition for each directory. If it is obtained from the method table file 117 and the original time stamp obtaining method is “Directory”, the directory time stamp of the original directory 23 is obtained as the original time stamp.
[0052]
If the original time stamp acquisition method is “File”, the time stamp of the file stored in the original directory 23 and indicated by URL R is acquired as the original time stamp.
As described above, according to the present embodiment, the original time stamp acquisition method by file type table file 116 or the original time stamp acquisition method by directory in which the correspondence between the directory identifier and the original time stamp acquisition method is stored. Since the table file 117 is added to the server selection device 1, a more appropriate original time stamp can be obtained.
In the present embodiment, instead of the time when the file is stored in the original directory 23, the version number of the directory can be used as the time stamp of the file.
Although the invention made by the present inventor has been specifically described based on the above-described embodiment, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention. Of course.
[0053]
【The invention's effect】
The effects obtained by the representative ones of the inventions disclosed in the present application will be briefly described as follows.
(1) According to the present invention, as an information providing source server, it is possible to perform a process of selecting an information providing source server in a shorter time while avoiding a mirror server in which an update error has occurred.
(2) According to the present invention, it is possible to avoid a mirror server in which an update error has occurred as an information providing source server even in a form in which a homepage such as HTML is distributed.
(3) According to the present invention, it is possible to more reliably deliver the latest version of information at a higher speed, and it is possible to realize an information service that is more comfortable for the user.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a schematic configuration of a server selection device according to a first embodiment of the present invention.
FIG. 2 is a block diagram showing a schematic configuration of a server selection device according to a second embodiment of the present invention.
FIG. 3 is a block diagram showing a schematic configuration of a server selection device according to a third embodiment of the present invention.
FIG. 4 is a diagram showing the contents of a MirrorControl file created on the mirror server.
FIG. 5 is a diagram showing the contents of a server list table file created on the server selection device.
FIG. 6 is a diagram showing the contents of an original directory file created on the server selection device.
FIG. 7 is a diagram showing the contents of a directory timestamp table file created on the server selection device.
FIG. 8 is a diagram showing the contents of a version number file created on the original server.
FIG. 9 is a diagram showing the contents of a time stamp file created on the original server and the mirror server.
FIG. 10 is a diagram showing the contents of a file type original time stamp acquisition method table file created on the server selection device.
FIG. 11 is a diagram showing the contents of a directory-specific original time stamp acquisition method table file created on the server selection device.
FIG. 12 is a diagram for explaining conventional redirection.
FIG. 13 is a block diagram showing a schematic configuration of a conventional server selection device.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Server selection apparatus, 2 ... Original server, 3 ... Mirror server, 4 ... Terminal, 11 ... Determination means, 12 ... Server narrowing-down means, 13 ... Duplicate file time stamp acquisition means, 14 ... Server list table, 15 ... Server list Reference means, 16 ... original time stamp acquisition means, 17 ... directory time stamp acquisition means, 18 ... directory time stamp table, 21 ... update reception directory, 22 ... update reception program, 23 ... original directory, 31 ... mirroring program, 32 ... Mirror directory, 51... Judgment means, 53... Original time stamp acquisition method table by file type, 54... Original time stamp acquisition method table by directory, 101. ... Selection request processing acquisition program, 110 ... MirrorControl file, 111 ... Server list table file, 112 ... Original directory file, 113 ... Directory time stamp table file, 114 ... Version number file, 115 ... Time stamp file, 116 ... By file type Original time stamp acquisition method table file, 117... Original directory time stamp acquisition method table file.

Claims (12)

In an information service system that creates a duplicate directory of an original directory on a plurality of mirror servers and provides information using a plurality of servers, information including a service identifier is input and information is provided from the mirror server group. A server selection device for selecting an information provider server to be
The attribute value of a file indicates the order in which the value between files is stored in the original directory as a time stamp, the latest value of the time stamp of a file belonging to a certain directory as the directory time stamp of the directory,
Directory time stamp acquisition means for acquiring a directory time stamp of the replication directory of each mirror server;
A directory time stamp table for storing the directory time stamp acquired by the directory time stamp acquisition means for each of the replication directories of each mirror server;
An original time stamp acquisition means for acquiring the time stamp of the file stored in the original directory and indicated by the service identifier as an original time stamp;
Server narrowing means for referencing the directory time stamp table and excluding the mirror server holding the replication directory whose directory time stamp is older than the original time stamp from the server to be selected;
A server selection device comprising: server determination means for determining the information providing source server from servers other than the servers excluded by the server narrowing means. The server selection apparatus according to claim 1, wherein the time stamp of the file represents a time when the file is stored in the original directory. When the version number is an attribute value of the original directory that is added every time the original directory is updated, the time stamp of the file is the version number of the original directory immediately after the file is stored in the original directory. The server selection device according to claim 1, characterized in that: The original time stamp obtaining means obtains an original time stamp obtaining method for obtaining a time stamp of a file stored in an original directory and indicated by a service identifier, or obtaining a directory time stamp of an original directory. It has a judgment means to judge based on,
The original time stamp acquisition means acquires, as the original time stamp, the time stamp of the file stored in the original directory and indicated by the service identifier or the directory time stamp of the original directory based on the determination result of the determination means. The server selection device according to any one of claims 1 to 3, wherein the server selection device. A file type original time stamp acquisition method table storing the correspondence between the file type and the original time stamp acquisition method;
5. The server selection according to claim 4, wherein the determination unit determines an original time stamp acquisition method corresponding to a file type indicated by the service identifier, based on the file type original time stamp acquisition method table. apparatus. A directory-specific original time stamp acquisition method table storing a correspondence relationship between a directory identifier and the original time stamp acquisition method;
5. The server selection apparatus according to claim 4, wherein the determination unit determines an original time stamp acquisition method corresponding to the directory indicated by the service identifier based on the directory-specific original time stamp acquisition method table. In an information service system in which a replica directory of a certain original directory is created on a plurality of mirror servers and information is provided using a plurality of servers, information including a service identifier is input to a computer, and the information is stored in the mirror server group. A server selection program for selecting an information provider server that provides information,
The attribute value of a file indicates the order in which the value between files is stored in the original directory as a time stamp, the latest value of the time stamp of a file belonging to a certain directory as the directory time stamp of the directory,
Procedure 1 for obtaining a directory time stamp of the replication directory of each mirror server;
Storing the directory timestamp acquired in the procedure 1 in a directory timestamp table for each of the replication directories of each mirror server; and
Procedure 3 for acquiring the time stamp of the file stored in the original directory and indicated by the service identifier as the original time stamp;
A step 4 of referring to the directory timestamp table and excluding the mirror server holding the replica directory whose directory timestamp is older than the original timestamp from the servers to be selected;
A server selection program that causes a computer to execute a procedure 5 for determining a server that provides information from servers other than the servers excluded in the procedure 4. 8. The server selection program according to claim 7, wherein the time stamp of the file represents a time when the file is stored in the original directory. When the version number is an attribute value of the original directory that is added every time the original directory is updated, the time stamp of the file is the version number of the original directory immediately after the file is stored in the original directory. The server selection program according to claim 7, wherein the server selection program is expressed. The procedure 3 determines based on the service identifier whether or not to acquire the time stamp of the file indicated by the service identifier stored in the original directory or the directory time stamp of the original directory. Procedure 3A to
And a step 3B of acquiring a time stamp of the file indicated by the service identifier stored in the original directory or a directory time stamp of the original directory as the original time stamp based on the determination result in the step 3A. The server selection program according to any one of claims 7 to 9, wherein the server selection program is any one of claims 7 to 9. A file type original time stamp acquisition method table storing the correspondence between the file type and the original time stamp acquisition method;
The procedure 3A is a procedure for referring to the file type original time stamp acquisition method table and determining an original time stamp acquisition method corresponding to a file type indicated by the service identifier. The server selection program described. A directory-specific original time stamp acquisition method table storing a correspondence relationship between a directory identifier and the original time stamp acquisition method;
11. The server according to claim 10, wherein the procedure 3A is a procedure for referring to the directory-specific original timestamp acquisition method table and determining an original timestamp acquisition method corresponding to the directory indicated by the service identifier. Selection program.

Images