JP4472498B2 - Storage connection method of diskless boot system, system, management server, storage, and program therefor - Google Patents

Description

  The present invention relates to a storage connection method in a diskless boot system, the system, a management server therefor, a storage, and a program.

  The diskless boot system is a system in which a client terminal is connected to a storage in which an OS, applications, and user data are stored via a network, starts the OS and applications, and writes and reads user data (here. Say, storage also includes storage built by file servers).

  Therefore, when the client terminal and the storage are in different sites, or when the client terminal and the storage are connected via a WAN, network delays may cause deterioration in the performance of the OS or application, and the user on the client terminal. It causes deterioration in performance of data write processing and read processing. Therefore, in Patent Document 1, a configuration file including a map of a network server and a file server storing an OS is acquired from a file server at a remote site when the client terminal starts up, and the file server at the local site, that is, the client terminal An efficient file transfer is realized by connecting to a file server on an existing site and starting the OS.

JP 2002-123400 A

  However, the above technique assumes that the user has two sites, for example. First, the OS of the client terminal is started using the storage of the site #A at the site #A, the writing process is performed, and after the shutdown, the site When moving to #B and using the storage of site #B at site #B to start the OS of the client terminal and perform the write process, there is a problem that data consistency cannot be maintained.

  In order to maintain data consistency, it is first necessary to use the storage at the site where the client terminal is started as the active storage, and to perform remote copy of update data synchronously or asynchronously from the active storage to the storage at another site It is.

  In addition, when the user's client terminal starts up at a different site from the site where the client terminal was started last time because the user moved between sites, the active storage is changed, so the remote copy direction is It is necessary to change in conjunction with movement.

  In the case of asynchronous remote copy, if the remote copy of the update data is not completed immediately before changing the direction of remote copy, it is necessary to perform resynchronization and complete the remote copy.

  Therefore, the object of the present invention is that, even when the client terminal is activated at a remote site as the user moves, the performance of the access from the client terminal to the storage is not affected, and It is an object of the present invention to provide a diskless boot system that maintains data consistency except for interruption of remote copy processing.

  Therefore, in order to solve the above problem, in the present invention, the site name of each site, the subnet information of each site, the failure state (normal or failure) of each site, the address information of the storage to which the client terminal last connected, The management server manages the storage address information of each site assigned to the user. When a client terminal makes an inquiry request to the management server, it selects the site closest to the client terminal from the sites that have a normal failure state and that has storage allocated to the user, and the storage allocated to the user of the selected site Is the new storage. Alternatively, the storage with the lowest load among the storages allocated to the users of the site whose failure state is normal is set as the new storage.

  If the new storage is different from the address information of the previously connected storage, the management server commands the storage disk control unit to resynchronize the data of the newly connected storage and the new storage, and determine the direction of remote copy between the storages. Then, change the direction of remote copy from the new storage to the storage at another site, and update the address information of the previously connected storage to the address information of the new storage.

  However, resynchronization is not necessary if the remote copy is synchronous. In addition, when the failure state of the site with the storage connected last time is failure, resynchronization is not performed.

  From the above, the client terminal can access appropriate storage such as the storage at the nearest site or the storage with the lowest load among the storage at the normal site. Performance is not affected and data consistency can be maintained except for interruption of remote copy processing of data due to a site disaster.

  Here, for the method of selecting the site closest to the client terminal, for example, there is a monitoring server with the shortest delay time by measuring the delay time between the monitoring server of the site that starts the client terminal and the monitoring server of another site The existing technology is used such that the site to be used is the closest site, and is not mentioned in the present invention.

  In addition, a method for selecting a storage with a low load uses an existing technique such as acquisition by a function provided in the disk control unit of the storage, and is not mentioned in the present invention.

  In addition, a method for changing the direction of remote copy between storages and a method for resynchronizing data between storages that perform remote copy use the technology that is conventionally provided in the disk control unit of the storage. do not do.

  When the present invention is used, even when the site where the client terminal starts is changed with the movement of the user, the OS or application can be automatically started with an appropriate storage. Degradation of user data write processing and read processing performance on the terminal can be prevented, and data consistency is maintained except when the previously activated site is in failure.

  Next, the present invention will be described in detail with reference to the drawings. FIG. 1 shows an embodiment of the present invention, where two sites (subnets # 3 and # 4) where the client terminal # 1 can be activated and two sites where the management server and storage are installed (subnets) are shown. # 1, # 2). Management server # 1, storage # 1, download server # 1, DHCP server # 1, DNS server # 1, and monitoring server # 1 belong to site # 1 (subnet # 1) and belong to site # 2 (subnet # 2) Includes management server # 2, storage # 2, download server # 2, DHCP server # 2, DNS server # 2, and monitoring server # 2. In addition, download server # 3, DHCP server # 3, DNS server # 3, and monitoring server # 3 belong to site # 3 (subnet # 3), and download server # 4 and DHCP belong to site # 4 (subnet # 4). Server # 4, DNS server # 4, and monitoring server # 4 belong. The client terminal # 1 can be connected to any of the sites # 1 to # 4 as indicated by a dotted line in FIG. In FIG. 1, the user A uses the client terminal # 1, but the management server manages the user with the user name and password as described later, and the relationship between the user and the client terminal is variable. Sites # 1 to # 4 are connected by an IP network.

  The storages # 1 and # 2 have a disk control unit and a storage device (disk) (not shown), and hold a remote copy function for copying data between the storages # 1 and # 2.

  The management servers # 1 and # 2 manage the site name 201, the site subnet information 202, and the failure state 203 that takes normal or failure values according to the site management table 200 of FIG. In the example of FIG. 2, the subnet information of site # 1 is 192.47.10.0/24, the failure state is normal, and the subnet information of site # 2 is 192.47.200.0/24, failure. The condition is normal. In addition, the management servers # 1 and # 2 use the user management table 300 in FIG. 3 to store the user name 301, the password 302, the storage address information 303 and 304 assigned to the user at each site, and the storage previously connected (hereinafter, the current use at login). Address information 305 of the system storage). In the example of FIG. 3, reference numeral 303 denotes storage address information of the site # 1, and reference numeral 304 denotes storage address information of the site # 2. In the example of FIG. 3, as is clear from the fact that the address information 305 of the active storage at the time of login of the user # 1 is the same as the address information 304 of the storage of the site # 2, the user # 1 is stored in the storage of the previous site # 2. Connected to. Similarly, users # 2 and # 3 are connected to the storage of site # 1 last time. The management servers # 1 and # 2 store the site management table 200 and the user management table 300 in their respective storage devices.

  One management server service is running on all sites # 1 to # 4, and the data in the management server is always synchronized with the management server (failover destination) at the other site from the management server running the service. Suppose that

  The monitoring servers # 1 to # 4 at each site include a management server address information table 400 (see FIG. 4) for managing the network address information 401 of all management servers (management servers # 1 and # 2) and the site name of each site. 501, subnet information 502, priority order 503 indicating the priority order for starting the management server, priority order table 500 (see FIG. 5) for managing the failure state 504, and active system site table for managing the site name 601 of the active system site 600 (see FIG. 6) is managed. In the example of the priority order table 500 in FIG. 5, the management server # 1 of the site # 1 has the priority order 1 and the management server # 2 of the site # 2 has the priority order 2. Since the management server does not exist at the site # 3 and the site # 4, the priority order 503 column of the priority order table 500 is blank. The active site is the site with the highest priority for starting the management server among the sites that are in a normal failure state. In the example of the active site table 600 in FIG. 6, the site # 1 of the site name 501 in which the column of the failure state 504 in the priority table 500 is normal and the priority level 503 is 1, is the active site in the active site table 600. Corresponds to the site name 601. The priority order for starting the management server may be dynamically calculated from the load of the management server. The monitoring servers # 1 to # 4 store a management server address information table 400, a priority table 500, and an active site table 600 in their respective storage devices.

  The download servers # 1 to # 4 at each site have a program for starting the OS downloaded by the client terminal # 1 and a program for authenticating the user.

  The DHCP servers # 1 to # 4 at each site allocate and reply to a network address in response to a network address acquisition request from the client terminal # 1.

  The DNS servers # 1 to # 4 at each site correspond to the management server name of the management server # 1 and # 2 at the two sites # 1 and # 2, respectively. Is registered as the network address of the management server, and the network address of the registered management server is returned to the management server network address acquisition request from the client terminal # 1.

  The management server registers the site name 201, the subnet information 202 of the site, and the failure state 203 input by the administrator at the time of site registration (see the site management table 200 in FIG. 2). When the site failure notification is received from the monitoring server, the failure state 203 of the corresponding site is changed to failure. Further, when a notification of site failure recovery is received from the monitoring server, the failure state 203 of the corresponding site is changed normally.

  The management server registers the user name 301 and password 302 inputted by the administrator at the time of user registration, storage address information 303 and 304 allocated to the user at each site, and address information 305 of the active storage at the time of login (FIG. 3). User management table 300).

  The monitoring servers # 1 to # 4 of each site monitor each other for failures at other sites. When a site failure is detected, the failure status 504 of the corresponding site in the priority table is changed to failure (FIG. 5), if the site where the failure is detected is other than the active site, the management server notifies the management server of the site failure. If the active site is out of order, the management server is failed over to the management server of the site with the highest priority among the sites with normal failure status. At the same time, the address of the new management server is registered in all DNS servers. The site name 601 of the active site in the system site table 600 is changed, and the site failure is notified to the management server. When the failure recovery of the site is detected, the failure state 504 of the corresponding site in the priority order table 500 is changed to normal, and the failure recovery of the site is notified to the management server. Further, when the failure state of the site having higher priority than the active site becomes normal, the management server is switched back to the management server of the higher priority site, and at the same time, the address of the new management server is registered in all DNS servers, The site name 601 of the active site in the active site table 600 is changed.

  Next, a detailed flow of processing accompanying client terminal activation will be described.

  After starting, the client terminal obtains the network address of the client terminal and the network address of the download server from the DHCP server. Next, the client terminal downloads a program for starting the OS and a program for authenticating the user from the download server. The client terminal executes a program for starting the downloaded OS and a program for authenticating the user, and the user inputs a user name and a password in a state where the program for authentication is being executed. The terminal transmits a user name, a password, and a user authentication request to the management server registered in the DNS server.

  7 and 8 show the operation flow of the management server when the client terminal is activated. As shown in FIG. 7, when receiving the user authentication request from the client terminal (701), the management server confirms the user name 301 and password 302 from the network address of the client terminal and the user management table 300 of FIG. 3 (702). ). If the user name 301 and the password 302 do not match from the user management table 300 (No in 703), a connection rejection is transmitted to the client terminal (704).

  If the user name 301 and the password 302 match from the user management table 300 (Yes in 703), the failure status 203 is determined from the network address of the client terminal, the site management table 200 in FIG. 2, and the user management table 300 in FIG. An appropriate storage is selected from the storages allocated to the users of the normal site (705, 706). Here, the method for the management server to select an appropriate storage is a site where the storage allocated to the user exists among the normal sites according to claim 3 and the site closest to the location of the client terminal A method for extracting the address information of the storage allocated to the user at the next selected site will be described below. First, based on the network address of the client terminal and the site management table 200 in FIG. 2, the management server is a site that has a normal failure state 203 and is assigned the storage to the user and is closest to the client terminal. Select. Next, the address information of the storage allocated to the user at the selected site is extracted (707). The address information 305 of the active storage at the time of login is compared with the address information of the storage allocated to the user at the selected site (708).

  If the address information 305 of the active storage at the time of login is different from the address information of the storage allocated to the user at the selected site (No in 708), the progress of the remote copy of the active storage at the time of login is displayed. An inquiry is made to the disk controller (709). If the synchronization is not completed (No in 710), the failure state of the site where the active storage exists at the time of login is further confirmed (711). If the failure state of the site where the active storage is present at the time of login is normal (Yes in 711), a resynchronization command is transmitted to the storage disk controller (712). When a resynchronization completion is received from the storage disk controller, a change command is sent to the storage disk controller from the storage assigned to the user of the site that has selected the remote copy direction of the storage in the direction of the active storage at the time of login. (801 in FIG. 8). When a remote copy direction change completion is received from the storage control unit (802), the address information 305 of the active storage at the time of login in FIG. 3 is changed to the address information of the storage at the selected site (803), and the client The address information of the selected storage is returned to the terminal (804). When synchronization is completed (Yes in 710 in FIG. 7), the storage disk control unit moves the storage remote copy direction from the storage allocated to the user of the selected site to the direction of the active storage at the time of login. A change command is transmitted (801 in FIG. 8). When a remote copy direction change completion is received from the storage control unit (802), the address information 305 of the active storage at the time of login in FIG. 3 is changed to the address information of the storage at the selected site (803), and the client The address information of the selected storage is returned to the terminal (804).

  When the address information of the active storage at the time of login is different from the address information of the storage allocated to the user at the selected site, the failure state of the site where the active storage at the time of login is a failure (No in 711 of FIG. 7). ), A change command is sent to the storage disk controller from the storage whose storage remote copy direction has been selected in the direction of the active storage at the time of login (801 in FIG. 8). When the remote control direction change completion is received from the storage control unit (802), the address information 305 of the active storage at the time of login in FIG. 3 is changed to the address information of the selected storage (803), and the client terminal The storage address information assigned to the user of the selected site is returned (804).

  When the address information of the active storage at the time of login is the same as the address information of the storage assigned to the user at the selected site (Yes in 708 in FIG. 7), the address is assigned to the user at the site selected as the client terminal. The storage address information (the address information of the active storage at the time of login, the same as the address information of the previously connected storage) is returned (713).

  Based on the storage address information received from the management server, the client terminal connects to the storage via the network, starts the OS or application stored in the storage, and reads / writes data to the storage via the network. .

  Note that the storage disk control unit operates as follows in response to the management server described above. When an inquiry about the progress status of the remote copy of the active storage at the time of login (709 in FIG. 7) is received from the management server, the disk control unit of the storage manages the progress status of the remote copy of the active storage at the time of login. Notify the server. When the resynchronization command (712) is received from the management server, resynchronization is performed, and after the resynchronization is completed, the resynchronization completion is notified to the management server. When a remote copy direction change command (801 in FIG. 8) is received from the management server, the remote copy direction is changed, and after the change is completed, the change completion is notified to the management server.

  In the above embodiment, the method for selecting an appropriate storage by the management server has been described as the method according to claim 3. However, as a method for the management server selecting an appropriate storage, the normal site according to claim 4 can be used. The site where the storage allocated to the user exists is extracted from the list, the storage load allocated to the user at the extracted site is obtained by querying the storage, and the address information of the storage with the lowest storage load is extracted. The same applies to the method used.

  According to the present embodiment, even when the site where the client terminal is activated is changed as the user moves, the OS or application can be automatically activated with an appropriate storage. The deterioration of the performance of the user data write processing and read processing can be prevented, and the consistency of the data can be maintained except when the previously activated site is in failure.

  In addition, the consistency of user data is maintained regardless of the location of the management server site, failover between management server sites, and failback, so this technology can be used to create a disaster recovery system using a diskless boot system. When the server is built, if the site where the management server exists becomes a failure and the management server fails over to another site, the OS and applications that the user is running and the user data that the user is writing or reading are Unless the stored storage exists at the same site, data consistency is maintained without shutting down or starting up the client terminal. In addition, even when the management server is switched back after the previous site recovery, it is not necessary to coordinate the shutdown and activation of the client terminal, and the data consistency can be maintained without stopping the service of the client terminal.

  In the embodiment described above, the management servers # 1 and # 2 are provided in the sites # 1 and # 2 (subnets # 1 and # 2) having the storages # 1 and # 2, respectively. Are synchronized, the management server at one site may manage storage at another site, or there may be only one management server in a system having a plurality of sites. That is, the system according to the embodiment of the present invention is a system having a plurality of sites including a site including at least one of storage and management servers, and including a plurality of storages and one or a plurality of management servers as a whole. That's fine. However, when there is only one management server in the system, the above failover cannot be performed.

  The management server, storage control unit, download server, DHCP server, DNS server, and monitoring server described above have means for performing the operation, and the means is a program created to perform the operation. This is realized by the CPU executing.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Of course.

It is a figure which shows the system which has four sites of 1 form of this invention. It is a figure which shows the example of a site management table | surface. It is a figure which shows the example of a user management table | surface. It is a figure which shows the example of the address information table | surface of a management server. It is a figure which shows the example of a priority table. It is a figure which shows the example of an active system site table | surface. It is a figure which shows the operation | movement flow of the management server at the time of a client terminal starting. It is a figure which shows the flow of the process 1 in FIG.

Explanation of symbols

200 ... Site management table, 300 ... User management table, 400 ... Address information table of management server, 500 ... Priority order table, 600 ... Active site table, 701-712, 801-804 ... Steps of operation flow of management server

Claims (8)

A system having a plurality of sites including a site including at least one of a storage and a management server, and including a plurality of storages and one or a plurality of management servers as a whole, the storage copying data between the storages A storage connection method for a diskless boot system that has a remote copy function and storage is selected by the management server depending on a system user,
The management server is
Receiving a connection request from a user via a client terminal and recognizing the user;
A procedure to select an appropriate new storage from at least the recognized user's information,
Determining whether the selected new storage is equal to the previously connected storage of the user;
When the new storage is equal to the previously connected storage, a procedure for notifying the client terminal of the address of the previously connected storage;
If the new storage is different from the last connected storage,
A procedure for inquiring about the progress of remote copy from the previously connected storage to the new storage,
If remote copy has been completed, go to the procedure for instructing to change the direction of remote copy,
If remote copy is not completed,
The procedure for checking the failure status of the site to which the previously connected storage belongs,
If the site to which the previously connected storage belongs is normal, order re-synchronization, confirm the end of resynchronization, and proceed to the procedure to instruct the change of the remote copy direction,
If the site to which the previously connected storage belongs is not normal, proceed to the procedure for instructing to change the direction of remote copy,
As a procedure for instructing the change of the direction of remote copy, the direction of remote copy is changed to the direction of other storage allocated to the user including the previously connected storage starting from the new storage. A procedure to instruct
And a procedure for notifying the client terminal of the address of the new storage after confirming the completion of the change. In the procedure for recognizing a user in the storage connection method according to claim 1,
A storage connection method for a diskless boot system, comprising a procedure for performing user authentication by prompting for input of a user ID and a password. In the procedure for selecting an appropriate new storage in the storage connection method according to claim 1 or 2,
A procedure for specifying the location of the client terminal by the network address of the connection request from the user;
A procedure for selecting a site where the storage allocated to the user exists from among normal sites and closest to the location of the client terminal;
A storage connection method for a diskless boot system, comprising: extracting address information of storage allocated to the user at a selected site. In the procedure for selecting an appropriate new storage in the storage connection method according to claim 1 or 2,
Extracting a site where storage allocated to the user exists from normal sites;
A procedure for acquiring the storage load allocated to the user at the extracted site by inquiring to the storage;
A storage connection method for a diskless boot system, comprising: extracting address information of a storage having the lowest storage load. A system having a plurality of sites including a site including at least one of a storage and a management server, and including a plurality of storages and one or a plurality of management servers as a whole, the storage copying data between the storages A diskless boot system that has a remote copy function and storage is selected by the management server depending on a user of the system,
The management server is
A means for receiving a connection request from a user via a client terminal and recognizing the user;
Means to select an appropriate new storage from at least the recognized user's information;
Means for determining whether the selected new storage is equal to the previously connected storage of the user;
Means for notifying the client terminal of the address of the previously connected storage when the new storage is equal to the previously connected storage;
If the new storage is different from the last connected storage,
Means for inquiring about the progress of remote copy from the previously connected storage to the new storage;
If the remote copy has been completed, a means for proceeding to a means for instructing the change of the remote copy direction;
If remote copy is not completed,
A means to check the failure status of the site to which the previously connected storage belongs,
When the site to which the previously connected storage belongs is normal, means for instructing resynchronization, confirming the end of resynchronization, and proceeding to means for instructing the change of the remote copy direction;
If the site to which the previously connected storage belongs is not normal, proceed to the means to instruct the change of the remote copy direction,
As means for instructing a change of the remote copy direction, changing the relative new storage, in the direction of the remote copy orientation to another storage allocated to the user that contains a storage in which the last connecting the new storage starting Means to instruct
A diskless boot system comprising: means for notifying the client terminal of a new storage address after completion of the change. Storage in the diskless boot system according to claim 5 ,
Means for remote copy to copy data between storages;
Storage unit comprising: means for changing a remote copy direction by a remote copy direction change command from the management server, and notifying the management server of the completion of the change after the change is completed. A means for receiving a connection request from a user via a client terminal and recognizing the user;
Means to select an appropriate new storage from at least the recognized user's information;
Means for determining whether the selected new storage is equal to the previously connected storage of the user;
Means for notifying the client terminal of the address of the previously connected storage when the new storage is equal to the previously connected storage;
If the new storage is different from the last connected storage,
Means for inquiring about the progress of remote copy from the previously connected storage to the new storage;
If the remote copy has been completed, a means for proceeding to a means for instructing the change of the remote copy direction;
If remote copy is not completed,
A means to check the failure status of the site to which the previously connected storage belongs,
When the site to which the previously connected storage belongs is normal, means for instructing resynchronization, confirming the end of resynchronization, and proceeding to means for instructing the change of the remote copy direction;
If the site to which the previously connected storage belongs is not normal, proceed to the means to instruct the change of the remote copy direction,
As means for instructing a change of the remote copy direction, changing the relative new storage, in the direction of the remote copy orientation to another storage allocated to the user that contains a storage in which the last connecting the new storage starting Means to instruct
A management server for a diskless boot system, comprising means for notifying the client terminal of a new storage address after completion of the change. A program for causing a computer to execute the means according to claim 6 or 7 .

Images