JP3781369B2 - Storage subsystem - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a subsystem transition technique, and more particularly to a technique that is effective when applied to a subsystem transition operation under a central processing unit in an information processing system or the like that assumes non-stop operation.
[0002]
[Prior art]
Data migration is to transfer data in the old disk subsystem functioning as an external storage device in the information processing system to the new disk subsystem. Conventionally, as a general method of data migration between disk subsystems, access from a central processing unit to a device to be migrated is stopped, the CPU reads data from the old disk subsystem, and the CPU reads the data from the new disk subsystem. A method using CPU intervention such as writing to a system is known. In this method, during the data migration, the customer's work for the disk subsystem is stopped for a long time.
[0003]
On the other hand, Hitachi's HODM (Hitachi Online Data Migration) function, IBM's extended remote copy function (hereinafter referred to as the “Hitachi Online Data Migration” function) XRC) or peer remote copy function (hereinafter PPRC) (literature example IBM 3990 Model 6 Enhancements), EMC's Symmetric Data Migration Service (SDMS) (literature example SYMTRIX IDAFUMT is there.
[0004]
In the HODM method, first, access to the old disk subsystem from the CPU is stopped. Thereafter, the connection path from the CPU to the old disk subsystem is changed from the CPU to the new disk subsystem, and a new access path is provided between the new and old disk subsystems. After that, by reading the data of the old disk subsystem from the new disk subsystem through the new access path, the migration is started and the access from the CPU is resumed. When the CPU is accessed for the migrated area, processing is performed by both the old and new disk subsystems. Further, when the CPU is accessed to the migration incomplete area, the data read from the old disk subsystem is reflected in the new disk subsystem for processing. This enables data migration during access from the CPU. A major feature of this function is that the old disk subsystem does not need to have a data migration function.
[0005]
In the XRC method, the old disk subsystem has a function of securing write data from the CPU in the disk controller, and the CPU has a function of reading the secured data. By writing this data to the new disk subsystem, data migration during access from the CPU is enabled. This method is characterized in that after the migration is completed, the customer's work for the old disk subsystem is stopped and switched to the new disk subsystem.
[0006]
In the PPRC method, the old disk subsystem and the new disk subsystem are connected to each other and have a communication function. By writing the CPU write data for the old disk subsystem to the new disk subsystem via this communication function, data migration during access from the CPU is enabled. In addition, like the XRC, this method has a feature that the customer's business is stopped and switched after the migration is completed.
[0007]
On the other hand, the SDMS first stops access to the old disk subsystem from the CPU. Thereafter, the connection path from the CPU to the old disk subsystem is changed from the CPU to the new disk subsystem, and a new access path is provided between the new and old disk subsystems. Thereafter, the migration is started by reading the data of the old disk subsystem from the new disk subsystem through the new access path. In addition, after the start of migration, the access from the CPU is resumed. When the CPU access is to the migrated area, it is processed directly by the new disk subsystem. When accessing the incomplete migration area, after reading the data of the track from the old disk subsystem, the new disk subsystem performs normal processing. This enables data migration during access from the CPU.
[0008]
[Problems to be solved by the invention]
In the conventional technology as described above, the access from the data stored in the old disk subsystem is switched from the old disk subsystem to the new disk subsystem by enabling access from the CPU even during data migration. You can keep it in time. However, in the case of system control data such as an OS, even if the access is temporarily stopped, it is a customer service stop, and the influence of the migration work is large. In particular, there is a technical problem that this is not permitted for customers who need online business for 24 hours, which is increasing, and that migration is possible only when the system is stopped, such as the year-end and New Year holidays.
[0009]
In general, one subsystem can be used by being connected to a plurality of CPUs. At this time, in the subsystem, the CPU is divided and processed for each received access route or for each access route group. This must also be distinguished for equivalent access to the peer subsystem.
[0010]
Further, when the access path from the CPU is switched to the new subsystem while the CPU access is continued, it is recognized that the CPU continues to access the same device. After the data migration is completed and the old subsystem is removed, a request for inputting device information may be issued from the CPU for the purpose of device confirmation. In the CPU that confirms the device and the access route by matching / mismatching the device information read in the past and the device information read this time, when the information of the new subsystem is sent at this time, the CPU does not match the device information and the access route There is a concern that the access path is cut off and the subsystem goes down.
[0011]
An object of the present invention is to provide a subsystem migration technique capable of continuing access from a host device to a subsystem side even during switching from an old subsystem to a new subsystem.
[0012]
Another object of the present invention is to provide a subsystem migration technique that does not require any stoppage of access from the host device to the subsystem side in accordance with the data migration procedure, and that allows data migration without interruption. .
[0013]
Another object of the present invention is to provide a subsystem migration technique that can smoothly migrate an old subsystem operating under the control of a plurality of host devices to a new subsystem under non-stop operation. It is in.
[0014]
Another object of the present invention is to provide a subsystem that enables smooth subsystem transition by avoiding the occurrence of failures due to environmental changes such as device information associated with transition from the old subsystem to the new subsystem. To provide transition technology.
[0015]
[Means for Solving the Problems]
In general, a plurality of access paths are provided from a higher-level device such as a CPU and a channel to a subordinate subsystem, and the higher-level device freely selects and switches the access path and accesses the subsystem. For example, even when restarting after an interruption that occurs in a series of instructions issued by a certain input / output processing request, an access path different from the original access path may be selected and used. Since the instruction before and after the interruption is a series of processes, it goes without saying that the instruction after the interruption cannot be executed unless the instruction before the interruption is executed in the subsystem. Therefore, even if the access path is changed, the subsystem can recognize it and process it as a series of instructions.
[0016]
The present invention is a storage system having a host device, a first subsystem that is a data migration source, and a second subsystem to which data is migrated from the first subsystem, in a first state To the third state through at least the second state, and the first state is a state in which the host device and the first subsystem are connected by the first access path. In the second state, the host device and the first subsystem are connected by the first access path, and the host device and the second subsystem are connected by the second access path. And the second subsystem are connected by a third access route, and the third state is a state in which a host device and the second subsystem are connected by a second access route, First In this state, the host device serves as a path for making an input / output request from the host device to the data stored in the first storage area to be accessed by the host device in the storage area in the first subsystem. Route from the first device via the first access route and at least two routes from the upper device via the second access route and the second subsystem via the third access route It is configured to be capable of input / output processing via any of the two routes, and starts data migration from the first subsystem to the second subsystem in the third state. When the second subsystem receives a command for data for which data migration has not been completed from the first subsystem to the second subsystem, the second subsystem The blanking system after data migration, the process for the command is executed. That is, when there are a plurality of first access paths from the host device to the old subsystem and a third access path between the old and new subsystems, the first access path from the host device to the old subsystem is divided into multiple times. Thus, the connection is changed to the second access path between the host device and the new subsystem. While the connection is being changed, the first and second access paths are connected to the old and new subsystems from the host device. At this time, if the old and new subsystems receive access from the host device, they are equivalent. The access request is relayed to the partner subsystem via the third access path. By doing so, the partner subsystem executes the instruction before the interruption, so that the instruction after the interruption can also be executed. This equivalent access needs to be performed by both the old and new subsystems. However, if a subsystem to be mainly processed during connection change is determined, it may be performed only by the opposite subsystem. Also, when the processing request from the host device is not interrupted, and when the processing request from the host device comes with a fixed access path, the partner subsystem does not receive the next command from the host device. In some cases, it may not be necessary to make equivalent access to the other subsystem. In addition, by making access to the other subsystem to the other subsystem with the access path fixed, it prevents the other subsystem from receiving the next sequential command from the upper device, and received from the upper device. You can also access differently. In this way, it is possible to switch the connection from the old subsystem to the new subsystem by dividing the access path into a plurality of times without stopping the access from the host device.
[0017]
For example, in the data migration in the disk subsystem, if the old disk subsystem is mainly processed during the connection change, the access request via the third access path of the present invention is sent to the new disk subsystem. If relaying is performed, connection switching can be performed without stopping access from the host device. However, if switching is performed while performing data migration, the old disk subsystem may receive direct access from the host device during the switching, and only the old disk subsystem may be updated. If the data of the part that has already been migrated is updated, the data of that part will be migrated.
[0018]
Therefore, in the present invention, the access request from the host device via the second access path is relayed to the new disk subsystem from the old disk subsystem to the new disk subsystem by relaying to the old subsystem via the third access path. When switching the connection to the system is realized, the data migration is started by starting the migration of data from the old subsystem to the new subsystem after the connection switching of all the first access paths to the second access path is completed. Updates that do not go through the new disk subsystem do not occur in the later part so that data migration is not required again.
On the other hand, when the connection is changed, the new disk subsystem performs processing mainly, that is, during the connection change from the first access path to the second access path, via the first access path. By switching the access request from the host device to the new subsystem via the third access path, connection switching can be performed without stopping access from the host device. However, before or during data migration, processing cannot be performed when data that has not been migrated to the new disk subsystem is accessed from the host device.
[0019]
Therefore, according to the present invention, in this case, prior to switching the first access path to the old subsystem to the second access path to the new subsystem, data migration from the old subsystem to the new subsystem ( Copy) is completed in advance, and by causing the old disk subsystem to perform an operation of relaying an access request coming from a host device via the first access path to the new subsystem via the third access path, The connection switching from the old disk subsystem to the new disk subsystem is realized.
[0020]
In order to enable connection switching during data migration from the old subsystem to the new subsystem, the data of both the old and new disk subsystems only need to be updated during the switching.
[0021]
Therefore, in the present invention, in such a case, in both the new and old disk subsystems, access requests coming from the host device via the first and second access paths are sent to each other via the third access path. Relaying to other subsystems enables connection switching during data migration.
[0022]
Further, in the present invention, in order to distinguish and transmit accesses of a plurality of higher-level devices to the partner subsystem, at least the same number of third access paths as the number of higher-level devices connected to the old disk subsystem are set between the new and old disk subsystems. It is intended to provide. Access via each third access path between the disk subsystems is made to correspond to access from each host device, and data migration is possible when the old disk subsystem is connected to a plurality of host devices. In addition, the number of third access paths includes not only the number of physical access paths but also the number of logical access paths.
[0023]
Further, in the present invention, the new disk subsystem issues a device information input request to the old disk subsystem in advance, and device information responded from the old disk subsystem at this time is read and stored. In response to a request for inputting device information from the host device, the device information of the stored old disk subsystem is sent instead of the device information of the new disk subsystem.
[0024]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0025]
(Embodiment 1)
FIG. 1 is a conceptual diagram showing an example of the configuration and operation of a general-purpose computer system that is an embodiment of an information processing system in which the subsystem migration method of the present invention is implemented.
[0026]
The configuration of this embodiment is an old sub comprising a CPU 10 as a central processing unit, a new disk controller unit 11 (hereinafter referred to as a new CU 11) as a data migration destination, and a new disk volume 12 (hereinafter referred to as a new VOL 12). The system includes an old subsystem consisting of an old disk controller unit 13 (hereinafter referred to as an old CU 13) and an old disk volume 14 (hereinafter referred to as an old VOL 14) as a data migration source.
[0027]
The old VOL 14 is a storage medium that operates under the old CU 13 and stores data exchanged with the CPU 10 via the old CU 13.
[0028]
The new VOL 12 is a storage medium that operates under the new CU 11, and stores data exchanged with the CPU 10 through the new CU 11 and data transferred from the old VOL 14.
[0029]
Further, the new CU 11 relays the access request coming from the CPU 10 via the second access path 20 ′ and the second access path 21 ′ to the old CU 13 via the third access path 30 and the third access path 31. As will be described later, a path migration control unit 111 that enables path migration between the old and new subsystems without stopping CPU access and a data migration control unit 112 that controls data migration are provided. The path migration control unit 111 performs an operation equivalent to the CPU 10 in accessing the old CU 13 using the third access path 30 and the third access path 31.
[0030]
In the data migration process of the present embodiment, the first access path 20 and the first access path 21 connected between the CPU 10 and the old CU 13 originally used are connected between the CPU 10 and the new CU 11 that is the data migration destination. The connection is switched as the second access path 20 ′ and the second access path 21 ′, and the new CU 11 and the old CU 13 are connected by the new third access path 30 and the third access path 31. Further, the new CU 11 and the new VOL 12 and the old CU 13 and the old VOL 14 are connected via a device path 12a and a device path 14a, respectively.
[0031]
First, an example of data migration processing in the information processing system having the configuration illustrated in FIG. 1 will be described with reference to the flowchart of FIG. FIG. 6 shows the operation of the maintenance personnel 5a, 5b,... That performs the inter-subsystem path transition according to the first invention in the embodiment of FIG. . . . . , 5i, and the CPU access path 50 indicating the access path used by the CPU 10 that changes accordingly, the path transfer control unit designation 51 indicating the state specified in the path transfer control unit 111, and the CU that processes the access from the CPU 10 The process CU52 indicating the above is shown in time series.
[0032]
First, since normal processing is being performed before the start of operation by the maintenance personnel, the first access paths 20 and 21 are used from the CPU 10, and the CPU access path 50 indicates the first access paths 20 and 21. Since the path migration control unit designation 51 is not related to the path migration, the new CU 11 processes the access from the CPU 10 and the same access to the old CU 13 through the third access paths 30 and 31 (provided later) is performed. It is specified so that it does not exist (hereinafter, such specification is referred to as processing in its own CU). Further, the processing CU52 is the old CU13. First, the maintenance staff newly establishes third access paths 30 and 31 between the new CU 11 and the old CU 13 (operation 5a).
[0033]
Further, when the CPU 10 receives access to the path transition control unit 111, the new CU 11 does not process, but designates the same access to the old CU 13 (relays the access) through the third access paths 30 and 31 (hereinafter referred to as such designation). (Referred to as processing in another CU) (operation 5b).
[0034]
Next, the first access path 20 from the CPU 10 is brought into an offline state (operation 5c), and the access of the CPU 10 using the path is stopped.
[0035]
At this time, the CPU access path 50 is only the first access path 21. Subsequently, the first access path 20 is switched to the second access path 20 ′ from the original configuration (operation 5d).
[0036]
When the connection is completed, the second access path 20 ′ (original first access path 20) is brought online from the CPU 10 (operation 5e).
[0037]
As shown in the CPU access path 50, the CPU 10 starts the access to the old CU 13 using the second access path 20 ′ together with the access using the first access path 21 performed so far. Access using the second access path 20 ′ is received by the new CU 11, but the same access is made to the old CU 13 via the third access paths 30 and 31 without processing by the new CU 11 in the path transition control unit 111. The old CU 13 is operated so as to be processed. Therefore, the CPU is accessed to the old CU 13 regardless of which of the second access paths 20 ′ (first access paths 21) is used, and the processing only by the old CU 13 is continued. In the same procedure, the first access path 21 is switched to the second access path 21 ′ by the operations 5f, 5g, and 5h. Thus, the access path from the CPU 10 to the old CU 13 can be switched to the new CU 11 without stopping access. Finally, since all the switching has been completed, if the designation of the path transition control unit 111 is changed to the processing in the own CU (operation 5f), the new CU 11 processes the access request from the CPU 10 as shown in the processing CU52. To do. In this way, the processing subsystem can be switched from the old CU 13 and the old VOL 14 to the new CU 11 and the new VOL 12 without stopping the access from the CPU 10.
[0038]
Here, the timing of the start of data migration in the migration of a subsystem accompanying data migration as in the present embodiment illustrated in FIG. 1 will be considered. The data migration process can be performed as long as any one of the third access paths 30 and 31 is present. However, if the first access path 20 is switched to the new CU 11 as the second access path 20 ′, the data migration is performed (started) before the first access path 21 is switched as the second access path 21 ′. Therefore, there is a possibility that the data update from the CPU 10 to the old CU 13 is performed on the migrated portion through the first access path 21. If such an update is performed, the new CU 11 does not know this, and therefore the data is not transferred.
[0039]
Therefore, in the present embodiment, the path migration control unit 111 provided in the new CU 11 relays access from the second access paths 20 ′ and 21 ′ to the old CU 13 via the third access paths 30 and 31. After the first access path 20 and the first access path 21 from the CPU 10 are all switched to the new CU 11 as the second access path 20 ′ and the second access path 21 ′, the data migration control unit 112 starts data migration. In addition, if the path transition control unit 111 is changed to processing in the own CU in synchronization with this, all data can be transferred without interruption, and non-stop data transfer can be performed.
[0040]
The entire processing procedure for subsystem migration including data migration in this embodiment will be described with reference to the flowchart illustrated in FIG.
[0041]
That is, in steps 101 to 106, after performing a path switching operation for switching the first access paths 20 and 21 to the second access paths 20 ′ and 21 ′, the new CU 13 side through the third access paths 30 and 31 starts the new operation. Copying (migration) of data to the CU 11 side (step 107) is performed on all data that needs to be migrated in the old VOL 14 under the control of the old CU 13 (step 108), and then the third access paths 30, 31 are performed. Then, the old CU 13 and the old VOL 14 are removed (step 109).
[0042]
Here, the processing of the access request from the CPU 10 during data migration in the above-described steps 107 to 108 is as shown in the flowcharts illustrated in FIGS. 8 and 9 as an example.
[0043]
That is, the copy process illustrated in FIG. 9 is executed as the background process, and during this time, the access request process illustrated in FIG. 8 is executed as needed.
[0044]
First, in the copy process of FIG. 9, when data copying is performed in units of tracks as an example, a copy management bitmap (not shown) is referred to (step 301), and the presence or absence of an uncopied track in the old VOL 14 is checked (step 302), if there is, the one with the smallest track number among the uncopied tracks is selected (step 303), copied to the new VOL 12 side via the third access paths 30, 31 (step 304), and then copied. The operation of updating the management bitmap to the track copied (step 305) is performed for all tracks to be migrated.
[0045]
On the other hand, as illustrated in FIG. 8, when the new CU 11 receives a command from the CPU 10 (step 201), it checks whether or not the access area of the command is an uncopied area (step 202). In this case, it is checked whether or not the command is a read command (step 203). If the command is a read command, the new VOL 12 is accessed from the old VOL 14 side via the third access paths 30 and 31. After copying the track containing the data (step 205), the copy management bitmap is updated to have been copied (step 206), and then command processing is executed (step 207).
[0046]
On the other hand, if it is determined in step 202 that the access has been made to the copied area, command processing is immediately executed in step 207.
[0047]
If it is determined in step 203 that the command is not a read command (ie, a write command), it is determined whether or not the old data is a required write (step 204), and if necessary, step 205 and subsequent steps are executed. If unnecessary, the command processing in step 207 is executed.
[0048]
Such processing enables the transition from the old subsystem to the new subsystem without stopping the information processing system, and the second access path 20 ′ of the first access paths 20 and 21 during non-stop operation. , 21 'can be performed smoothly and accurately after data switching.
[0049]
(Embodiment 2)
FIG. 2 is a conceptual diagram showing another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented, and FIGS. 10 and 11 are flowcharts showing an example of the operation thereof.
[0050]
As an example of the configuration of the information processing system according to the present embodiment, the new CU 11 does not include a path migration control unit and a data migration control unit, and the old CU 13 includes a path migration control unit 131 and a data migration control unit 132. Only differs from the case of FIG.
[0051]
First, a schematic operation of a path switching operation by relaying an access request generated from the first access paths 20 and 21 to the old CU 13 to the new CU 11 side via the third access paths 30 and 31 will be described.
[0052]
Since the normal process is not related to the path transition, the path transition control unit 131 designates the process in its own CU. First, the third access paths 30 and 31 between the new CU 11 and the old CU 13 are newly established (step 401 and step 402). Furthermore, the process at the other CU is designated to the path migration control unit 131. Thus, the processing is performed only by the new CU 11 by relaying through the third access paths 30 and 31 without processing by the old CU 13.
[0053]
Next, connection switching of the first access path 20 to the second access path 20 ′ is performed. At this time, the first access path 20 from the CPU 10 is set in an offline state, and the access of the CPU 10 using the path is stopped. Further, the third access path 30 is removed (steps 405 and 406). When the connection is completed, the CPU 10 brings the second access path 20 ′ (original first access path 20) to an online state.
[0054]
As an access to the old CU 13 from the CPU 10, the access to the new CU 11 using the second access path 20 ′ is started together with the access to the old CU 13 using the first access path 21 which has been performed so far. The new CU 11 receives an access using the second access path 20 ′, but the new CU 11 only receives the access received on the third access path 30 on the second access path 20 ′, and the processing is performed as it is. Continue (step 407). Access to the old CU 13 side using the first access path 21 is relayed and processed by the new CU 11 via the third access paths 30 and 31 (step 408).
[0055]
In a similar procedure, the first access path 21 is switched to the second access path 21 ′, and the third access path 31 is removed. In this way, the connection path from the CPU 10 to the old CU 13 can be switched to the new CU 11 without stopping access (step 409). Then, the old CU 13 and the subordinate old VOL 14 are removed (step 410).
[0056]
However, here, when applied to migration of a storage subsystem that requires data migration, in the CU of the storage subsystem that controls access to VOL data, if the data has not been migrated to the new CU 11, the CPU 10 Unable to handle access. Therefore, in this embodiment, when the third access paths 30 and 31 are newly established, the data migration control unit 132 is instructed to start data migration (step 403). When all the data can be migrated and the new CU 11 can be processed even if it is directly accessed from the CPU 10 (step 404), the path switching process after the step 405 is executed, and the access from the CPU 10 is stopped. By performing connection switching without any interruption, data can be transferred without interruption.
[0057]
Here, an example of an access request (command processing) generated from the CPU 10 during the path switching operation in steps 405 to 409 after completion of data migration in steps 403 to 404 will be described with reference to FIG.
[0058]
That is, when the old CU 13 receives a command from the CPU 10 (step 501), it checks whether the access area of the command is a copied area (step 502). Whether the command is a command is checked (step 503). If it is not a read command (write command), command processing is executed by both the new CU 11 and the old CU 13 (step 504). On the other hand, if it is determined in step 503 that the command is a read command, the old CU 13 performs command processing using the data of the old VOL 14 (step 505).
[0059]
If it is determined in step 502 that the access is to an uncopied area, in step 505, the old CU 13 performs command processing using the data of the old VOL 14.
[0060]
By such processing, the path transition control unit 131 provided on the old CU 13 side relays the access request to the new CU 11 side via the third access paths 30 and 31, without stopping the information processing system. The subsystem can be migrated to the new subsystem, and the path switching operation during the data migration can be accurately executed without stopping.
[0061]
(Embodiment 3)
FIG. 3 is a conceptual diagram showing another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented. The configuration of the information processing system in the present embodiment is different from that in FIG. 1 only in that the old CU 13 includes a path transition control unit 131. Here, the characteristic outline operation in the present embodiment will be described, and in the flow of the description, the third access route 30 to the other subsystems by the path migration control unit 111 and the path migration control unit 131 with each other. , 31, the outline operation of the path switching operation by relaying the access request will be described.
[0062]
Since the normal process is not related to the path transition, the path transition control unit 111 and the path transition control unit 131 are designated by the local CU. First, the third access paths 30 and 31 between the new CU 11 and the old CU 13 are newly established. Here, when access is received from the CPU 10 to the path transfer control unit 111 and the path transfer control unit 131, the processing is performed by the own CU and the same access is made to the partner CU through the third access paths 30 and 31 (hereinafter referred to as such specification). Is referred to as processing in both CUs). An access from the old CU 13 to the new CU 11 results in an error because there is no data in the new CU 11 yet, but there is no problem because the old CU 13 is executing processing from the CPU 10.
[0063]
Next, the data migration control unit 112 is instructed to start data migration. At this time, in the new CU 11, the access from the old CU 13 is normally processed in the case of the data migration completed part, and in the case of the data non-migration part, the data is read from the migration source by the conventional data migration function, so it is processed normally Will come to be. Thereafter, during data migration, connection switching of the first access path 20 to the second access path 20 ′ is performed. At this time, the first access path 20 from the CPU 10 is set in an offline state, and the access of the CPU 10 using the path is stopped. When the connection is completed, the CPU 10 brings the second access path 20 ′ (original first access path 20) to an online state. As an access to the old CU 13 from the CPU 10, the access using the second access path 20 ′ is started together with the access using the first access path 21 which has been performed so far. The new CU 11 receives an access using the second access path 20 ′, but the new CU 11 processes the access together with the access from the old CU 13 that has been received so far. However, since the access is from the CPU 10, the same access is made to the old CU13. In this state, both the old CU 13 and the new CU 11 are accessed and processed from the CPU 10, but all the other CUs are also accessed, so the input / output process is interrupted and the second access path 20 ′ and the first access path 21 are entered. Can be processed even if they are accessed after being replaced.
[0064]
The first access path 21 is switched to the second access path 21 ′ in the same procedure. As described above, both the old CU 13 and the new CU 11 include the path transition control unit 131 and the path transition control unit 111, and the first access paths 20 and 21 from the CPU 10 to the old CU 13 are stopped without stopping the access of the CPU 10. The connection can be switched to the second access paths 20 ′ and 21 ′ to the new CU 11.
[0065]
Thereafter, data migration (copying) of all data to be migrated from the old CU 13 side to the new CU 11 side is performed by the conventional data migration function by the data migration control unit 112 provided on the new CU 11 side. In this way, data migration that can switch connection of the first access paths 20 and 21 to the second access paths 20 ′ and 21 ′ without stopping the access of the CPU 10 even during data migration can be performed.
[0066]
(Embodiment 4)
FIG. 4 is a conceptual diagram showing an example of the configuration and operation of another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented. The configuration of the information processing system in the present embodiment includes a plurality of CPUs 10a and 10b instead of a single CPU 10. The CPU 10a includes an access path 20a and an access path 21a, and the CPU 10b includes an access path 20b and an access path 21b. 1 is different from FIG. 1 in that it is connected to a path switching device 15 that dynamically switches an access route, and the path switching device 15 and the old CU 13 are connected by a first access route 20c and a first access route 21c.
[0067]
Further, in the subsystem migration processing according to the present embodiment, the first access route 20c and the first access route 21c are connected between the path switching device 15 and the new CU 11 that is the data migration destination and the second access route 20c ′ and the first access route 20c. The connection is changed as the two access path 21c ′.
[0068]
In the present embodiment, the access from the CPU 10a through the access path 20a is performed by the old CU 13 through the first access path 20c via the path switching device 15. Similarly, the access path 21a is implemented in the old CU 13 through the first access path 21c. Further, in the case of the CPU 10b, the access path 20b is implemented through the first access path 20c, and the access path 21b is implemented through the first access path 21c. In such a case, the first access path 20c is physically one path, but has two logical access paths so that the access from the access path 20a and the access from the 20b can be distinguished.
[0069]
Similarly, in the old CU 13, each of the second access paths 20c ′ (second access paths 21c ′) is an access path 20a (access path 20b) and an access path 21a (access path 21b) on the CPU 10a (CPU 10b) side. It functions as two logical access paths corresponding to each, and these two paths are recognized and processed as access paths from separate CPUs 10a and 10b.
[0070]
As illustrated in FIG. 1, a path transition control unit 111 is provided on the new CU 11 side, and an access request from the second access route is relayed to the old CU 13 side via the third access routes 30 and 31. The access route switching operation from the first to the second access route is performed in the stop state, and the first access route 20c is switched by the same procedure as that performed after the route switching is completed, and the second access route 20c ′. Is recognized in the new CU11 as well.
[0071]
That is, the access received from the two logical access paths is divided, and the process is performed by determining which of the CPUs 10a and 10b is the access. The process during the access path switching is that the path transfer control unit 111 relays the access received from the CPU to the old CU 13 and performs the same access, but of course the CPU from which the access is from the old CU 13. Must be segmented. Since this division is made according to the difference in the access route, in the present embodiment, the third access route 31 is used for access from the CPU 10a, and the third access route 30 is used for access from the CPU 10b. . Similarly, when the first access route 21c is switched to the second access route 21c ′, the same operation may be performed. As described above, when there are a plurality of CPUs connected to the old CU 13, the inter-subsystem path migration is performed by providing the third access path between the new CU 11 and the old CU 13 at least as many as the number of CPUs, and the data migration is realized. This can be achieved by the present embodiment illustrated in FIG.
[0072]
In the present embodiment, the third access path 30 is used for access from the CPU 10a and the third access path 31 is used for access from the CPU 10b. However, only the third access path 30 can be used. In other words, when the number of access paths between the new CU 11 and the old CU 13 is smaller than the number of CPUs, it is also possible to provide two logical access paths in the third access path 30 and use them separately. Needless to say, even when the number of third access paths is sufficient (when the number is the number of CPUs or more), a plurality of logical access paths can be provided and used. Furthermore, the second access path 20c ′ is associated with the third access path 30, the second access path 21c ′ is associated with the third access path 31, and the second access path 20c ′ and the second access path 21c ′ are associated with each other. A certain logical access path may be provided on the third access paths 30 and 31 as they are.
[0073]
(Embodiment 5)
Another embodiment of the subsystem migration method of the present invention will be described with reference to FIGS. FIG. 5 is a conceptual diagram showing an example of the old CU device information table 40 in which device information of the old CU 13 is stored in a storage means such as a buffer memory provided in the new CU 11 of the information processing system illustrated in FIG. is there.
[0074]
Normally, in a subsystem operating under the control of a CPU in an information processing system, information such as the device configuration of the subsystem is read to the CPU as necessary for the purpose of determining the environment and specifications of the subsystem. Command interface (device information input request).
[0075]
When the old CU 13 is connected to the device information input request from the CPU 10, if the same request is made to the old CU 13 and the returned information is input to the CPU 10, the CPU 10 does not determine an access path failure or the like. It can be used continuously as a subsystem. However, after the data migration is completed, the old CU 13 is usually removed. Therefore, in this embodiment, the new CU 11 requests all device information input requests via the third access paths 30 and 31 in advance so that the device information of the old CU 13 can be responded (input) to the CPU 10 even after the removal. Is recorded on the old CU 13 and the returned information is recorded in the old CU device information table 40 as a pair of the input request name 40a and the information 40b returned from the old CU 13 in response to the input request. In response to the device information input request, the information 40b in the old CU device information table 40 is read and responded.
[0076]
In this way, even when the CPU 10 is a CPU that performs comparison with past device information, the use of the new CU 11 can be continued after the old CU 13 is removed. That is, the old CU 13 can be removed without causing a concern such as system down. In the present embodiment, the storage means such as the old CU device information table 40 provided in the new CU 11 is used. However, the device information originally possessed by the new CU 11 may be rewritten with the information of the old CU 13.
[0077]
As described above, according to the present invention, the path can be switched from the old subsystem at the migration source to the new subsystem at the migration source without stopping the CPU access at all. Therefore, a completely non-stop system migration is possible.
[0078]
Further, even when the disk subsystem has a function that can switch paths without stopping CPU access only in the new disk subsystem that is the migration destination, it is possible to perform completely non-disruptive data migration including when switching paths.
[0079]
Further, even when the disk subsystem has a function that can perform path switching without stopping CPU access only in the old disk subsystem of the migration source, complete non-stop data migration including path switching can be performed.
[0080]
In addition, the disk subsystem has a function that can switch paths without stopping CPU access in both the new subsystem at the migration destination and the old disk subsystem at the migration source, enabling path switching during data migration. It is possible to perform complete non-disruptive data migration.
[0081]
Furthermore, even in a subsystem that operates under the control of a plurality of CPUs, a complete non-stop system migration is possible.
[0082]
Furthermore, the device information of the subsystem is stored and compared with the newly read device information of the current subsystem, and the device information of the old subsystem is stored on the new subsystem side even in the case of a CPU that detects an abnormality such as an access path. In response to the device information of the old subsystem that has been read and stored in advance in response to the device information input request, the path is switched without stopping the CPU access, and the old subsystem of the migration source is quickly restored. Can be removed.
[0083]
Although the invention made by the present inventor has been specifically described based on the embodiments, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say.
[0084]
For example, the storage subsystem with data migration has been described as an example of the subsystem. However, the present invention is not limited to this and can be widely applied to general subsystems that do not require data migration.
[0085]
【The invention's effect】
According to the subsystem migration method of the present invention, it is possible to obtain an effect that it is possible to continue the access from the host device to the subsystem side even during switching from the old subsystem to the new subsystem.
[0086]
Further, there is no need to stop access from the host device to the subsystem side in accordance with the data migration procedure, and there is an effect that data migration can be performed without stopping.
[0087]
In addition, there is an effect that the old subsystem operating under the control of a plurality of higher-level devices can be smoothly transferred to the new subsystem under non-stop operation.
[0088]
In addition, it is possible to avoid the occurrence of a failure due to an environmental change such as device information accompanying the transition from the old subsystem to the new subsystem, and to smoothly transition the subsystem.
[Brief description of the drawings]
FIG. 1 is a conceptual diagram showing an example of the configuration and operation of a general-purpose computer system that is an embodiment of an information processing system in which a subsystem migration method of the present invention is implemented.
FIG. 2 is a conceptual diagram showing another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 3 is a conceptual diagram showing another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 4 is a conceptual diagram showing an example of the configuration and operation of another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 5 is a conceptual diagram showing an example of the contents of storage means used in another embodiment of the subsystem migration method of the present invention.
FIG. 6 is a flowchart showing an example of the operation of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 7 is a flowchart showing an example of the operation of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 8 is a flowchart showing an example of the operation of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 9 is a flowchart showing an example of the operation of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 10 is a flowchart showing an example of the operation of another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented.
FIG. 11 is a flowchart showing an example of the operation of another embodiment of the information processing system in which the subsystem migration method of the present invention is implemented.
[Explanation of symbols]
5a to 5i ... operation, 10, 10a, 10b ... CPU, 11 ... new disk controller unit (new CU), 12 ... new disk volume (new VOL), 12a ... device path, 13 ... old disk controller unit (old CU) , 14 ... Old disk volume (old VOL), 14a ... Device path, 15 ... Path switching device, 20a, 20b ... Access route, 21a, 21b ... Access route, 20 ... First access route, 20 '... Second access route 20c ... first access route, 20c '... second access route, 21 ... first access route, 21' ... second access route, 21c ... first access route, 21c '... second access route, 30, 31 ... 3rd access route, 40 ... old CU device information table, 40a ... input request name, 40b ... information, 50 ... CPU access route , 51... Path migration control unit designation, 52... Processing CU, 111... Path migration control unit, 112... Data migration control unit, 131.

Claims (2)

A storage system comprising a host device, a first subsystem that is a data migration source, and a second subsystem to which data is migrated from the first subsystem,
From the first state to the third state through at least the second state,
The first state is:
The host device and the first subsystem are connected by a first access path,
The second state is:
The host device and the first subsystem are connected by the first access path, the host device and the second subsystem are connected by a second access path, and the first subsystem A state in which the second subsystem is connected by a third access path;
The third state is:
The host device and the second subsystem are connected by the second access path;
In the second state,
As a path for making an input / output request from the host device to the data stored in the first storage area to be accessed by the host device in the storage area in the first subsystem, the host device At least a path that passes through the first access path, and a path that relays the second subsystem from the higher-level device via the second access path and passes through the third access path. There are two routes, and input / output processing via either of the two routes is possible ,
In the third state, data migration is started from the first subsystem to the second subsystem, and data migration from the first subsystem to the second subsystem is completed. When the second subsystem receives a command for data that is not present, the process for the command is executed after data migration from the first subsystem to the second subsystem. Storage system. The storage system of claim 1,
The first state is:
In a state where the host device and the first subsystem are connected by a plurality of first access paths, and the first subsystem and the second subsystem are connected by the third access path. There,
The third state is:
The host device and the second subsystem are connected by the second access path, and the first subsystem and the second subsystem are connected by the third access path. A storage system characterized by that.

Images