JP3612209B2 - Remote transfer method by magnetic disk controller - Google Patents

Description

[0001]
(table of contents)
TECHNICAL FIELD OF THE INVENTION
Conventional technology (FIGS. 16 to 18)
Problems to be Solved by the Invention (FIGS. 19 to 25)
Means for solving the problem
BEST MODE FOR CARRYING OUT THE INVENTION
[A] Description of the first embodiment (FIGS. 1 to 6)
[B] Description of the second embodiment (FIGS. 7 to 10)
[C] Description of the third embodiment (FIGS. 11 and 12)
[D] Description of the fourth embodiment (FIGS. 13 to 15)
[E] Other
The invention's effect
[0002]
BACKGROUND OF THE INVENTION
In general, in a system in which a secondary center (backup system) is installed at a location (remote location) different from the location of the primary center (business system) where the original business is performed, and the primary center business can be backed up by the secondary center Each center is provided with a magnetic disk control device (hereinafter referred to as a remote FCU (File Control Unit)) having a remote transfer function.
[0003]
The present invention relates to a remote transfer method for realizing a function for guaranteeing the order of transfer data (tracks) when remote transfer is performed from the primary center side to the secondary center side by the remote FCU described above.
[0004]
[Prior art]
In recent years, when the operation of the main center (business system) that performs the original business is stopped due to a disaster such as an earthquake or sudden failure, as shown in FIG. It is common to install a sub-center (backup system) 1B on the ground.
[0005]
The configuration of the system that enables backup of the work of the main center 1A by the sub-center 1B in the remote place will be described with reference to FIG. In FIG. 16, components having the same function in the primary center 1A and the sub-center 1B are given the same numerals as the symbols, and symbols indicating components belonging to the primary center 1A are given “A”. The reference numerals indicating the components belonging to the sub-center 1B are given “B”.
[0006]
The primary center 1A includes a host (CPU) 2A, a magnetic disk control unit (hereinafter referred to as FCU) 3A, a magnetic disk unit (volume) 4A, and an adapter 5A. The host 2A performs processing for the original business and is connected to the communication line 6 via the adapter 5A. The FCU 3A is interposed between the host 2A and the magnetic disk device 4A, and writes / reads to / from the magnetic disk device 4A in response to input / output signals (hereinafter sometimes simply referred to as I / O) from the host 2A. Control is performed.
[0007]
Similarly to the primary center 1A, the secondary center 1B includes a host (CPU) 2B, an FCU 3B, a magnetic disk device (volume) 4B, and an adapter 5B. The host 2B takes over processing for business from the main center 1A at the time of backup, and is connected to the communication line 6 via the adapter 5B. The FCU 3B is interposed between the host 2B and the magnetic disk device 4B, and performs write / read control for the magnetic disk device 4B in accordance with input / output signals from the host 2B.
[0008]
In the main center 1A, when the data of the magnetic disk device 4A is changed in accordance with the execution of business, information about the changed data is sent to the communication line 6 through the host 2A and the adapter 5A. It is transmitted to the sub-center 1B via the line 6. The information is received by the host 2B from the communication line 6 through the adapter 5B in the sub-center 1B. The host 2B outputs I / O to the FCU 3B, thereby correcting the data of the magnetic disk device 4B. The same change as the data of the magnetic disk device 4A of the center 1A is performed. As a result, the same data is always held in the magnetic disk device 4A of the primary center 1A and the magnetic disk device 4B of the secondary center 1B. That is, a volume (remote shadow volume) equivalent to the magnetic disk device 4A of the primary center 1A is always constructed in the magnetic disk device 4B of the secondary center 1B.
[0009]
Therefore, if the operation of either the primary center 1A or the sub-center 1B is stopped due to a disaster such as an earthquake or a sudden failure, the data is held in the other center 1B or 1A that has avoided the operation stop. Therefore, the other center 1B or 1A can immediately take over the operation of the center 1A or 1B that has stopped operating and back up the business.
[0010]
However, since the host 2A performs processing for the original business and has a heavy load from the beginning, when the host 2A of the primary center 1A transfers data to the secondary center 1B as described above, it can transfer too much data. Therefore, efficient transfer could not be performed.
Therefore, in each of the primary center 1A and the secondary center 1B, instead of the FCUs 3A and 3B shown in FIG. 16, as shown in FIG. 17, FCU (hereinafter referred to as remote FCU) 10A and 10B having a remote transfer function are provided. A remote system in which the remote FCUs 10A and 10B are connected by a communication line 6 has been proposed and used. In this remote system, data transfer (remote transfer) from the primary center 1A to the secondary center 1B is performed by the remote FCU 10A, so that the load on the host 2A is distributed and data transfer can be performed efficiently.
[0011]
For example, a high-speed digital line or an ATM line is used as the communication line 6 connecting the remote FCUs 10A and 10B. The hosts 2A and 2B and the remote FCUs 10A and 10B are connected by a channel interface, and the remote FCUs 10A and 10B and the magnetic disk units (volumes) 4A and 4B are connected by a PCI (Peripheral Component Interconnect) interface or SCSI. (Small Computer System Interface) interface.
[0012]
The configuration of the remote FCUs 10A and 10B described above will be described with reference to FIG. As shown in FIG. 18, each of the remote FCUs 10A and 10B includes a CA (Channel Adapter) 11, an RM (Resource Manager) 12, a TS (Table Storage) 13, an RA (Remote Adapter) 14, and a CFE (Cache Function Engine). 15, SS (Shared Storage) 16, DA (Device Adapter) 17, and C-BUS (Control-BUS) 18.
[0013]
Here, CA11 is a module in charge of external interface control with the hosts 2A and 2B, RM12 is a module in charge of management / control of resources in the remote FCUs 10A and 10B related to basic operations, and TS13 is A memory attached to the RM 12 and used as a resource management table in the remote FCUs 10A and 10B.
[0014]
The RA 14 is a line communication module that uses a line control device (not shown) and is responsible for transferring data to and from another remote FCU 10A or 10B via the communication line 6. The CFE 15 is a cache memory 16a in the SS 16. This module is responsible for managing (see FIG. 17). Here, the shared memory unit including the cache memory 16a and the like is collectively referred to as SS (reference numeral 16). Normally, a plurality (for example, 2 to 8) of RAs 14 are mounted on each remote FCU 10A, 10B, and a plurality (for example, 2 to 8) of RA 14 is provided between the primary center 1A and the subcenter 1B. The line is laid. However, in FIG. 17, only one RA 14 and one communication line 6 are shown.
[0015]
The DA 17 is a module in charge of interface control with the lower-level magnetic disk devices (volumes) 4A and 4B, and the C-BUS 18 is used for the above-described functional modules (reference numerals 11 to 17) to communicate with each other. The route is used for transmission / reception of control information between modules.
The data flow in the remote system configured as described above will be described with reference to FIG.
[0016]
As shown in FIG. 17, the RA 14 in the remote FCU 10A is connected to the RA 14 in the remote FCU 10B in the remote center 1B via the communication line 6, and the data from the host 2A in the primary center 1A. Is received by the CA 11 of the remote FCU 10A and once written to the cache memory 16a on the SS 16.
[0017]
The data (track) written in the cache memory 16a is divided into a plurality of packets by the RA 14 and transferred to the remote FCU 10B on the secondary center 1B side. In the primary center 1A, writing from the cache memory 16a to the magnetic disk device 4A is executed at an appropriate timing by the DA 17 of the remote FCU 10A.
[0018]
On the other hand, in the secondary center 1B, data from the remote FCU 10A on the primary center 1A side is received by the RA 14 of the remote FCU 10B, and is once written in the cache memory 16a on the SS 16. The data (track) written to the cache memory 16a is written to the magnetic disk device 4B by the DA 17 at an appropriate timing.
[0019]
[Problems to be solved by the invention]
By the way, when the host 2A on the primary center 1A side issues an I / O to the remote FCU 10A, it designates the order guarantee mode to perform data writing / transfer with guaranteed order to the remote FCU 10A. Can be made.
Here, data writing / transfer in which order is guaranteed means that, on the primary center 1A side, data to be written / transferred in the order of tracks T1, T2, T3 is sent from the host 2A to the remote FCU 10A. Data (tracks T1, T2, T3) is written / transferred in the same order to the magnetic disk unit 4A on the primary center 1A side and the remote FCU 10B on the secondary center 1B side.
[0020]
When the host 2A on the primary center 1A side designates the order guarantee mode as described above, on the primary center 1A side, the order of data sent from the host 2A to the remote FCU 10A, and from the remote FCU 10A to the magnetic disk unit 4A The order of data sent is always guaranteed for control.
However, on the secondary center 1B side, the remote FCU 10B writes the tracks from the primary center 1A side to the magnetic disk device 4B in the order of arrival, so the order may be lost (ie, the magnetic disk device 4B The order of tracks to be written may be different from the order of data transmitted from the host 2A on the primary center 1A side).
[0021]
Hereinafter, data writing / transfer in which order is maintained, data writing / transfer in which order is not maintained, and the necessity of a transfer function that guarantees order will be described with reference to FIGS. explain.
First, data writing / transfer in which order is maintained will be described with reference to FIGS. 19 and 20. FIG. Here, FIG. 19 is a diagram for explaining data writing / transfer in which order is maintained, and FIG. 20 shows data writing / transfer in which order is maintained in the time axis and FIG. 5 is a flowchart for explaining in correspondence with (1) to (4). In FIG. 19, the same reference numerals as those described above indicate the same or substantially the same parts, and the description thereof will be omitted.
[0022]
However, in the remote system shown in FIG. 19, the remote FCU 10A on the primary center 1A side and the remote FCU 10B on the secondary center 1B side are connected by two communication lines 6-1 and 6-2. That is, two RAs 14 (hereinafter referred to as RA01 and RA02, respectively) are mounted on the remote FCU 10A, and two RAs 14 (hereinafter referred to as RA03 and RA04, respectively) are mounted on the remote FCU 10B. RA01 and RA03 are connected by a communication line 6-1, and RA02 and RA04 are connected by a communication line 6-2.
[0023]
As shown in FIGS. 19 and 20, on the primary center 1A side, the host 2A sequentially writes tracks T1 and T2 into the cache memory 16a secured on the SS 16 of the remote FCU 10A ((1) and (See (3)). At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
Then, the remote FCU 10A reads the track T1 previously written in the cache memory 16a by RA01 and transfers it to the remote FCU 10B on the secondary center 1B side using the communication line 6-1 (RA01 / RA03 path) (▲ 2 ▼).
[0024]
Following this, the remote FCU 10A next transfers the track T2 written in the cache memory 16a. At this time, since the communication line 6-1 (RA01 / RA03 path) is in use, RA02 The track T2 is read and transferred to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path) (see (4)).
[0025]
Thus, in the data transfer examples shown in FIGS. 19 and 20, the tracks T1 and T2 are transferred from the remote FCU 10A to the remote FCU 10B in the order sent from the host 2A. The tracks T1 and T2 are written to the magnetic disk device 4B while maintaining the order.
[0026]
On the other hand, data writing / transfer in which order is not maintained will be described with reference to FIGS. Here, FIG. 21 is a diagram for explaining data writing / transfer in which order is not maintained, and FIG. 22 is a time axis and data writing / transfer in which order is not maintained in FIG. 7 is a flowchart for explaining in correspondence with (1) to (5). In FIG. 21, the same reference numerals as those already described indicate the same or substantially the same parts, and the description thereof will be omitted.
[0027]
As shown in FIGS. 21 and 22, on the primary center 1A side, the host 2A sequentially writes the tracks T1 and T2 to the cache memory 16a secured on the SS 16 of the remote FCU 10A ((1) and (See (3)). At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
Then, the remote FCU 10A reads the track T1 previously written in the cache memory 16a by RA01, and uses the communication line 6-1 (RA01 / RA03 path) for the track T1 to the remote FCU 10B on the secondary center 1B side. The transfer is started (see (2)).
[0028]
Subsequently, the remote FCU 10A transfers the track T2 written to the cache memory 16a next. At this time, since the communication line 6-1 (RA01 / RA03 path) is in use, the track is transmitted by RA02. T2 is read and transferred to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path) (see (4)).
[0029]
In this way, when both RA01 and RA02 are transferring, if the transfer of the track T1 fails due to a failure on the communication line 6-1 or the like, the remote FCU 10A causes the communication line 6-2 (RA02 / RA02) by RA02. At this point, the communication line 6-2 (RA02 / RA04 path) is used to transfer the track T2. The transfer of the track T1 is kept waiting until (4) is completed. When the transfer of the track T2 by RA02 is completed, the track T1 is read by RA02 and transferred to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path) (see (5)). ).
[0030]
Thus, in the data transfer examples shown in FIGS. 21 and 22, due to the occurrence of a transfer error, the track arrival order of the remote FCU 10B on the secondary center 1B side (that is, the writing order to the magnetic disk device 4B) is correct. This is different from the track sending order on the center 1A side, and the order is lost. Specifically, those sent in the order of tracks T1 and T2 arrive at the remote FCU 10B in the order of tracks T2 and T1 and are written to the magnetic disk device 4B.
[0031]
Next, the necessity of the transfer function that guarantees the order will be described with reference to FIGS. Here, FIG. 23A and FIG. 23B are diagrams showing specific examples of the two tracks T1 and T2 to be transferred while guaranteeing the order, respectively, and FIG. 24 shows FIG. FIG. 23B is a diagram for explaining the operation of the remote FCU 10B on the side of the sub center 1B when the two tracks T1 and T2 shown in FIG. 23B are transferred while guaranteeing the order, and FIG. FIG. 24 is a diagram for explaining the operation of the remote FCU 10B on the secondary center 1B side when the transfer order of the two tracks T1 and T2 shown in FIG. 23A and FIG. 23B is reversed.
[0032]
The order guarantee function is necessary, for example, when a plurality of data blocks are transferred with one I / O [one SIO (Start I / O) request from the host 2A].
When transferring a large number of data blocks from the primary center 1A side to the remote FCU 10B on the secondary center 1B side by one I / O, the data block to be transferred cannot be transferred with only the capacity of one record. May be divided and transferred. For example, as shown in FIG. 23A, a data block DB1 including an index for managing each subsequent data block and a number of data blocks DB2,..., DBm including update data are transferred, and FIG. ), The data block DB (m + 1),... Containing update data may be transferred by the track T2.
[0033]
At this time, when one set of update data is included in the data blocks DBm and DB (m + 1), ON / OFF of the recovery flag is set in the previous data block DBm. The role of the recovery flag and the operation of the remote FCU 10B on the secondary center 1B side when the two tracks T1 and T2 are transferred while guaranteeing the order will be described with reference to FIG.
[0034]
When the two tracks T1 and T2 are transferred while guaranteeing the order, the remote FCU 10B on the secondary center 1B side recovers the recovery flag and the update data from the data block DBm transferred by the track T1, as shown in FIG. (See {circle around (1)}), the update data paired with the update data of the data block DBm is written from the data block DB (m + 1) transferred by the subsequent track T2 (see {circle around (2)}). Thereafter, a pointer for managing the data block DB (m + 1) is written to the index of the data block DB1 (see (3)).
[0035]
Here, if the recovery flag of the data block DBm is set to ON, the data block DB (m + 1) is written even if a system down occurs between the operations (2) and (3) shown in FIG. You can proceed to the state. That is, the contents of the disk can be restored by setting a pointer from the data block DB1 to the data block DB (m + 1).
[0036]
On the other hand, when the transfer order of the two tracks T1 and T2 is reversed due to the occurrence of a transfer error or the like, the remote FCU 10B on the sub-center 1B side first starts with the track T2, as shown in FIG. After writing the update data paired with the update data of the data block DBm from the transferred data block DB (m + 1) (see (2)), the recovery flag is then transferred from the data block DBm transferred by the track T1. And update data is written (see (1)). Thereafter, a pointer for managing the data block DB (m + 1) is written to the index of the data block DB1 (see (3)).
[0037]
At this time, if a system down occurs between the operations (2) and (1) shown in FIG. 25, the recovery flag included in the data block m transferred on the track T1 at that time is not yet set. Since it is arrival, the recovery flag is treated as OFF. That is, it is determined that no recovery process is necessary in the remote FCU 10B, and nothing is updated in the index to the data block DB (m + 1). Therefore, it is determined that the area of the data block DB (m + 1) is an area where no data exists, and the data block DB (m + 1) area is overwritten by another data in the subsequent data processing, resulting in data destruction. There is a fear.
[0038]
In order to prevent such data destruction, an order guarantee function is required.
The present invention has been devised in view of such problems. When the order guarantee mode is designated, even if a transfer error occurs, the data (track) is correctly corrected while ensuring the order. It is an object of the present invention to provide a remote transfer method by a magnetic disk control device that can transfer data from a center side to a sub-center side.
[0039]
[Means for Solving the Problems]
In order to achieve the above object, a remote transfer method using a magnetic disk control device according to the present invention (Claim 1) includes a main center and a sub center having a magnetic disk device and a magnetic disk control device for controlling the magnetic disk device. In a remote system in which the primary center side magnetic disk control device and the secondary center side magnetic disk control device are connected by a plurality of communication lines, the primary center side magnetic disk control device and the secondary center side are connected through the plurality of communication lines. A method for transferring data to a side magnetic disk control device, wherein a plurality of tracks are continuously transferred from a primary center side magnetic disk control device to a secondary center side magnetic disk control device while ensuring the order. When the order guarantee mode is specified by the host on the positive center side, the magnetic disk on the positive center side Control device,A plurality of tracks transferred by the same input / output request from the primary center host are assigned the same identification number and a serial number corresponding to the transmission order from the primary center host, and the identification number and serial number are The given plurality of tracks are simultaneously transferred to the sub-center side magnetic disk control device via a plurality of communication lines, and the sub-center side magnetic disk control device transfers the tracks transferred from the main center side and written on the sub-center side. There is an order management table that holds the identification number assigned and the serial number LSN assigned to the track that has the identification number and was written last on the secondary center side. When a new track is transferred, the serial number SN assigned to the track and the order management table are retained. The serial number LSN corresponding to the identification number assigned to the track is compared. If SN ≠ LSN + 1, the serial number LSN in the order management table is updated by writing another track, and SN = LSN + 1 Until SN, the track is kept waiting without giving the write permission to the track. On the other hand, when SN = LSN + 1, the write permission is given to the track, and the sub-center side magnetic disk device of the track If the serial number LSN of the track is registered / updated as the serial number LSN of the order management table when writing to the track is completed, and a transfer error occurs due to a line failure during transfer of a certain track, Side magnetic disk control device has an unused communication line among other communication lines. If there is an unused communication line, use the unused communication line to transfer the track in which a transfer error occurred, while there is no unused communication line. Determine whether the serial number assigned to the track currently being transferred contains a serial number larger than the serial number assigned to the track that caused the transfer error. If the corresponding serial number exists, the transfer of the track having the corresponding serial number is interrupted. , Use the communication line that was used to transfer the track, and transfer the track that caused the transfer error.It is characterized by that.
[0042]
At this time, the order management table may be managed by a resource module that manages / controls the resources in the secondary center side magnetic disk control device.2The track may be provided on a cache memory for temporary writing before writing to the secondary center side magnetic disk device.3).
Also,A remote transfer method by a magnetic disk control device according to the present invention (Claim 4) comprises a main center and a sub-center having a magnetic disk device and a magnetic disk control device for controlling the magnetic disk device, and controls the magnetic disk on the main center side. In a remote system in which the system and the secondary center side magnetic disk control unit are connected by a plurality of communication lines, data is transferred from the primary center side magnetic disk control unit to the secondary center side magnetic disk control unit through the multiple communication lines. The transfer method is a method for transferring a plurality of tracks continuously from the primary center side magnetic disk controller to the secondary center side magnetic disk controller while guaranteeing the order. When specified by the center side host, the primary center side magnetic disk control device A plurality of tracks transferred in response to the same input / output request from the host are assigned the same identification number and a serial number corresponding to the order of transmission from the primary center host, and the identification number and the serial number are given. The plurality of tracks are simultaneously transferred to the secondary center side magnetic disk controller via a plurality of communication lines,If a transfer error occurs due to a line failure during transfer of a track, the primary center side magnetic disk controller checks whether there are other unused communication lines among the multiple communication lines, and When there is a communication line, the unused track is used to transfer the track in which a transfer error has occurred. When there is no unused link, the track is being transferred. It is determined whether there is a serial number that is larger than the serial number assigned to the track in which the transfer error occurred. If there is no corresponding serial number, the communication line in use While waiting for the transfer of the track in which the transfer error occurred until it is released, if the corresponding serial number exists, the track with the corresponding serial number is present. Interrupt the transfer, row transfer of the track caused transmission errors by using a communication line that was being used for the transfer of the trackIt is characterized by that.
[0043]
The remote transfer method of the present invention described above (claims)1~3), In the secondary center side magnetic disk controller, the serial number SN of the track transferred from the primary center side is the serial number LSN (tracks belonging to the same I / O as that track) held in the order management table. It is determined whether or not the last written serial number) is equal to a value obtained by adding 1. For example, when the SN of the track is LSN + 2 (SN ≠ LSN + 1), the track has been transferred before the track assigned LSN + 1 as the serial number, and the predetermined order is lost. It can be determined that
[0044]
Therefore, if SN ≠ LSN + 1, the track is made to wait until the serial number LSN in the order management table is updated to SN = LSN + 1 by writing another track. That is, even if an error occurs during track transfer and the transfer order of the tracks to the sub-center side is reversed and the predetermined order is lost, the track order can be adjusted on the sub-center side.
[0045]
At this time, a transfer error occurs due to a line failure during transfer of a certain track, there is no unused communication line for transferring the track in which the transfer error occurred, and the serial number of the track currently being transferred Consider a case in which three conditions such that there is a serial number larger than the serial number of a track in which a transfer error has occurred overlapped. In this case, the track to be written after that track is transferred to the sub-center side first, and is waiting for the transfer of the track in which the transfer error has occurred. Since it does not exist, a transfer deadlock that locks a track in which a transfer error has occurred in a state where it cannot be transferred occurs.
[0046]
Therefore, if the above three conditions overlap, the primary center-side magnetic disk controller interrupts the transfer of the transfer deadlocked track and uses the communication line that was in use for the transfer of that track. To transfer a track with a transfer error to avoid a transfer deadlock.1, 4).
[0047]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
The remote system to which the remote transfer method by the magnetic disk control device as the embodiment (first to fourth embodiments) of the present invention is applied is the same as that shown in FIG. Since the magnetic disk control apparatuses, that is, the remote FCUs 10A and 10B in the first to fourth embodiments) are the same as those shown in FIG. 18, detailed descriptions of these systems and the remote FCUs 10A and 10B are omitted. In the figure, the same reference numerals as those already described indicate the same or similar parts, and the description thereof is omitted.
[0048]
Also, in the remote system of this embodiment described below (see FIGS. 2, 4, 8, 11, and 14), there is a gap between the remote FCU 10A on the primary center 1A side and the remote FCU 10B on the secondary center 1B side. They are connected by two communication lines 6-1 and 6-2. That is, two RAs 14 (hereinafter referred to as RA01 and RA02) are mounted on the remote FCU 10A, and two RAs 14 (hereinafter referred to as RA03 and RA04, respectively) are mounted on the remote FCU 10B. RA01 and RA03 are connected by a communication line 6-1, and RA02 and RA04 are connected by a communication line 6-2.
[0049]
[A] Description of the first embodiment
First, the remote transfer method according to the first embodiment of the present invention will be described with reference to FIGS. 1 and 2 are a sequence diagram and a block diagram for explaining the remote transfer method of the first embodiment, respectively. FIG. 3 shows the remote transfer method of the first embodiment as a time axis, and FIG. 3 is a flowchart for explaining in correspondence with (1) to (8) in FIG. In FIG. 2, the same reference numerals as those described above indicate the same or substantially the same parts, and the description thereof will be omitted.
[0050]
In order to realize the remote transfer method according to the first embodiment of the present invention, in the remote system of the first embodiment, when the order guarantee mode is designated, the remote FCU 10A (CA11) on the primary center 10A side is set to 1 from the host 2A. When a transfer request for one track is received, only a notification that the track has been received (Channel End) is reported to the host 2A, and "Device End" is not returned. Is transferred to the remote FCU 10B, and “Device End” is sent back to the host 2A as a transfer completion notification when the transfer is completed.
[0051]
Further, when the order guarantee mode is designated, the host 2A on the primary center 1A side makes a track transfer request to the remote FCU 10A one by one, receives a transfer completion notification (Device End) from the remote FCU 10A, and then A track transfer request is issued to the remote FCU 10A.
Note that the characteristic functions / configurations of the first embodiment described above are all easily realized by slightly modifying the firmware (program) for realizing the remote transfer that has been used conventionally. The
[0052]
Next, the procedure of the remote transfer method according to the first embodiment in a state where no transfer error occurs will be described with reference to FIGS.
As shown in FIGS. 1 to 3, when the order guarantee mode is designated by the host 2A on the primary center 1A side, the host 2A performs the following operation on the cache memory 16a secured on the SS 16 of the remote FCU 10A. Only one track T1 is written (see (1)). At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
[0053]
When the remote FCU 10A receives the track T1, it reports "Channel End" to the host 2A (see (2)), reads the track T1 written in the cache memory 16a by RA01, and transmits the communication line 6 -1 (RA01 / RA03 path) is used to transfer to the remote FCU 10B on the secondary center 1B side (see (3)). When the transfer of the track T1 to the sub-center 1B is completed, the remote FCU 10A reports a transfer completion notification (Device End) to the host 2A (see (4)).
[0054]
Receiving the transfer completion notification from the remote FCU 10A, the host 2A writes only one second track T2 following the track T1 to the cache memory 16a of the remote FCU 10A (see (5)).
After that, when the remote FCU 10A receives the track T2 as described above, it reports "Channel End" to the host 2A (see (6)), and the track T2 written in the cache memory 16a by RA01 is recorded. Read and transfer to the remote FCU 10B on the secondary center 1B side using the communication line 6-1 (RA01 / RA03 path) (see (7)). When the transfer of the track T1 to the sub-center 1B is completed, the remote FCU 10A reports a transfer completion notification (Device End) to the host 2A (see (8)).
[0055]
Accordingly, the tracks T1 and T2 are sent one by one from the host 2A, and the track T2 is not sent until the transfer of the track T1 is completed. Therefore, the sub-center 1B side keeps the order of the track T1. , T2 are written to the magnetic disk device 4B (cache memory 16a).
On the other hand, the procedure of the remote transfer method according to the first embodiment when a transfer error occurs due to a line failure or the like during the transfer of the track T1 will be described with reference to FIGS. 4 is a block diagram for explaining a remote transfer method (when a transfer error occurs) as a first embodiment of the present invention, and FIG. 5 is a remote transfer method (first embodiment of the present invention). FIG. 6 is a sequence diagram for explaining (when a transfer error occurs). FIG. 6 shows a remote transfer method (when a transfer error occurs) according to the first embodiment of the present invention. It is a flowchart for demonstrating corresponding to-(9). In FIG. 4, the same reference numerals as those already described indicate the same or substantially the same parts, and the description thereof is omitted.
[0056]
As shown in FIGS. 4 to 6, when the order guarantee mode is designated by the host 2A on the primary center 1A side, the host 2A makes the cache memory 16a secured on the SS 16 of the remote FCU 10A with respect to the cache memory 16a. Only one track T1 is written (see (1)). At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
[0057]
When the remote FCU 10A receives the track T1, it reports "Channel End" to the host 2A (see (2)), reads the track T1 written in the cache memory 16a by RA01, and transmits the communication line 6 -1 (RA01 / RA03 path) to the remote FCU 10B on the secondary center 1B side, but if the transfer fails due to a line error (line failure), the remote FCU 10A uses the alternative path It is determined to transfer the track T1 (see (3)).
[0058]
At this time, since the other communication line 6-2 (RA02 / RA04 path) is unused, the remote FCU 10A uses the communication line 6-2 to transfer the track T1 to the remote FCU 10B on the sub-center 1B side. Transfer (see (4)).
When the transfer of the track T1 to the sub-center 1B is completed through the communication line 6-2, the remote FCU 10A reports a transfer completion notification (Device End) to the host 2A (see (5)).
[0059]
Receiving the transfer completion notification from the remote FCU 10A, the host 2A writes only one second track T2 following the track T1 to the cache memory 16a of the remote FCU 10A (see (6)).
Thereafter, when the remote FCU 10A receives the track T2, it reports "Channel End" to the host 2A (see (7)), reads the track T2 written in the cache memory 16a by RA01, and causes a line failure. The communication line 6-2 (RA02 / RA04 path) is used without using the communication line 6-1 in which the error occurred, and is transferred to the remote FCU 10B on the secondary center 1B side (see (8)). When the transfer of the track T1 to the sub-center 1B is completed, the remote FCU 10A reports a transfer completion notification (Device End) to the host 2A (see (9)).
[0060]
As described above, according to the remote transfer method as the first embodiment of the present invention, in the state where the order guarantee mode is designated by the host 2A, the host 2A makes track transfer requests one by one. When a transfer request for one track is made to the remote FCU 10A, a transfer request for the next track is not issued until a transfer completion notification for that track is received from the remote FCU 10A. Each is transferred to the sub-center 1B side.
[0061]
That is, as described with reference to FIGS. 4 to 6, even if an error occurs during track transfer, the communication line 6-2 other than the communication line 6-1 in which the communication failure has occurred is used.UseSince the start of transfer of the next track can be suppressed until the transfer of the track is completed, the order of the track can be reliably guaranteed.
[B] Description of the second embodiment
Next, a remote transfer method according to the second embodiment of the present invention will be described with reference to FIGS. 7 is a diagram showing an example of an order management table used in the remote transfer method of the second embodiment, FIG. 8 is a block diagram for explaining the remote transfer method of the second embodiment, and FIG. FIG. 10A to FIG. 10D are flowcharts for explaining the remote transfer method of the second embodiment in correspondence with the time axis and (1) to (10) in FIG. It is a figure which shows the registration state of the order management table at the time of performing the remote transfer method of embodiment. In FIG. 8, the same reference numerals as those already described indicate the same or substantially the same parts, and the description thereof is omitted.
[0062]
In order to realize the remote transfer method according to the second embodiment of the present invention, the host 1A and the remote FCUs 10A and 10B in the remote system of the second embodiment have the following functions.
In the second embodiment, on the main center 1A side, unlike the first embodiment, the host 2A sequentially sends a plurality of tracks to the remote FCU 10A in a predetermined order as in the conventional case. A plurality of communication lines 6-1 and 6-2 (only two are shown in the present embodiment) are simultaneously used and transferred to the sub-center 1B side.
[0063]
However, in the second embodiment, when the order guarantee mode is designated, the remote FCU 10A gives the following information (a1) and (a2) to the track transferred through the communication line 6-1 or 6-2.
(A1) Track number (serial number SN): Information indicating the number of tracks related to the I / O of the track belonging to the same I / O from the host 2A. This track number matches the order in which tracks are sent from the host 2A to the remote FCU 10A.
[0064]
(A2) ID value (identification number): The same ID value is assigned to tracks belonging to the same I / O from the host 2A, and the I / O to which the track belongs can be identified by the ID value. It has become. This ID value is incremented every time a new I / O (Start I / O) is received from the host 2A by the remote FCU 10A.
[0065]
In the remote FCU 10B on the secondary center 1B side in the second embodiment, the RM 12 (TS 13) is provided with an order management table (Table) 20 that holds the ID value and the track number (SN) described above. Yes.
The order management table 20 includes an ID value assigned to a track transferred from the primary center 1A side and written on the secondary center 1B side, and a track having the ID value and last written on the secondary center 1B side. The track number (serial number) LSN assigned to is stored in correspondence with each other. That is, as the track number in the order management table 20, the track number of the last written track is written as LSN in the cache memory 16a secured on the SS 16 of the remote FCU 10B with respect to the ID value.
[0066]
A specific example of values held in the order management table 20 is shown in FIG. In the specific example shown in FIG. 7, the track number “0001”, that is, only writing of the first track is completed for the track having the ID value “0001”, and the track is recorded for the track having the ID value “0002”. The writing of the number “0005”, that is, the fifth track is completed, and the track having the ID value “0003” is completed until the writing of the track number “0007”, that is, the seventh track.
[0067]
When the RA 14 in the remote FCU 10B on the secondary center 1B side receives a track from the RA 14 on the primary center 1A side, the RA 14 notifies the RM 12 of the ID value and the track number SN of the track and makes a write request to the cache memory 16a. have.
In response to a request from the RA 14, the RM 12 in the remote FCU 10B on the secondary center 1B side performs the following operation according to two conditions (b1) and (b2).
[0068]
Condition (b1): If the track number notified from the RA 14 is “1”, the ID value and the track number are written in the unused area or the oldest area in the order management table 20. However, the ID value notified from the RA 14 is written as it is as the ID value, but “0” is first written as the track number. Thereafter, the RM 12 notifies the RA 14 of permission to write the track to the cache memory 16a. The track number “0” in the order management table 20 is rewritten to “1” when the writing of the track to the cache memory 16a is completed.
[0069]
Condition (b2): If the track number notified from the RA 14 is other than “1”, it is searched whether or not the same ID value as that track is stored in the order management table 20. When the same ID value is already held, the track number SN notified from the RA 14 is compared with the track number LSN already held in the order management table 20, and whether SN = LSN + 1 is determined. to decide. If SN = LSN + 1, the RM 12 notifies the RA 14 of permission to write to the cache memory 16a of the track. If SN ≠ LSN + 1, the RM 12 notifies the RA 14 to the cache memory 16a of the track. Write permission is not given. Even when the same track ID value is not stored in the order management table 20, the RM 12 does not give the RA 14 permission to write to the cache memory 16a of the track.
[0070]
Further, when the RA 14 receives a write permission notification as a response from the RM 12, the RA 14 writes the track into the cache memory 16a secured on the SS 16, whereas when the RA 14 receives a non-permission notification, the RA 14 writes the track. After waiting for a while, the RM 12 operates again to make an inquiry.
Further, after writing the track in the cache memory 16a, the RA 14 notifies the RM 12 of the ID value and the track number SN of the track again, and requests to register in the order management table 20. The received RM 12 searches for the same ID value from the order management table 20, and operates to write the track number SN notified from the RA 14 as the LSN in the track number area linked to the ID value.
[0071]
Note that any of the characteristic functions / configurations of the second embodiment described above can be easily realized by slightly modifying the firmware (program) for realizing remote transfer that has been conventionally used. The
Next, the procedure of the remote transfer method according to the second embodiment will be described with reference to FIGS.
[0072]
As shown in FIG. 8 and FIG. 9, when the order guarantee mode is designated by the host 2A on the primary center 1A side, the host 2A first accesses the cache memory 16a secured on the SS 16 of the remote FCU 10A. Thus, the tracks T1 and T2 belonging to the same I / O are sequentially written [see (1) and (6)]. At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
[0073]
The remote FCU 10A then reads the track T1 previously written in the cache memory 16a by RA01 and transfers it to the remote FCU 10B on the secondary center 1B side using the communication line 6-1 (RA01 / RA03 path). At this time, an ID value = 0001 and a track number = 0001 are added to the track T1 [see (2)].
[0074]
Following this, the remote FCU 10A next transfers the track T2 written in the cache memory 16a. Since the communication line 6-1 (RA01 / RA03 path) is in use, the track T2 is transferred by RA02. Read and transfer to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path). At this time, an ID value = 0001 and a track number = 0002 are added to the track T2 [see (7)].
[0075]
In the remote FCU 10B on the secondary center 1B side, the RA03 first receives the track T1 transferred from the primary center 1A side, and this RA03 also notifies the RM12 of the ID value = 0001 and the track number = 0001 of the track T1 at the same time. RM12 is inquired whether or not the track T1 can be written. At this time, the RM 12 operates according to the condition (b1) described above, and notifies the RA03 of the write permission of the track T1 [see (3)]. By the operation of the RM 12, the order management table 20 changes from the state shown in FIG. 10A to the state shown in FIG.
[0076]
Upon receipt of the write permission notification from the RM 12, the RA 03 writes the track T1 in the cache memory 16a secured on the SS 16 [see (4)], and then again with respect to the RM 12, the ID value = 0001 and the track number = In response to this request, the RM 12 registers a track number in the order management table 20 [see (5)]. Thereby, the order management table 20 changes from the state shown in FIG. 10B to the state shown in FIG.
[0077]
In the remote FCU 10B on the secondary center 1B side, when the RA04 receives the track T2 from the primary center 1A side during the writing process of the track T1 (see (3) and (4)), the RA04 The T2 ID value = 0001 and the track number = 0002 are also notified to the RM 12 at the same time, and the RM 12 is inquired whether or not the track T2 can be written. At this time, the order management table 20 is in the state shown in FIG. 10B, and the RM 12 operates according to the condition (b2) described above. That is, SN ≠ LSN + 1 (where SN = 2, LSN = 0, and 2 ≠ 0 + 1), and the track T1 is not written in the cache memory 16a. (See (8) in FIG. 8, (8) -1 in FIG. 9)).
[0078]
The RA 04 that has received the write refusal notification from the RM 12 repeatedly inquires the RM 12 until the RM 12 gives permission to write the track T2 (see (8) in FIG. 8, (8) -2 in FIG. 9).
In the meantime, the writing of the track T1 is completed by RA03 [see (4)], and when the registration to the ordering management table 20 is completed [see (5)], the ordering management table 20 is shown in FIG. Since it is updated as shown and SN = LSN + 1 (where SN = 2, LSN = 1, 2 = 1 + 1), the RM 12 notifies the RA04 of the write permission of the track T2 [(8) of FIG. FIG. 9 (8) -3].
[0079]
The RA 04 that has received the write permission notification from the RM 12 writes the track T2 in the cache memory 16a secured on the SS 16 [see (9)], and then again with respect to the RM 12, the ID value = 0001 and the track number = In response to this request, the RM 12 registers a track number in the order management table 20 [see (10)]. Thereby, the order management table 20 changes from the state shown in FIG. 10C to the state shown in FIG.
[0080]
Thus, according to the remote transfer method as the second embodiment of the present invention, when the order guarantee mode is designated by the host 2A, the remote FCU 10B on the secondary center 1B side transfers from the primary center 1A side. The track number (serial number) SN of the track that has been recorded is the serial number LSN (the track number of the last written track among the tracks belonging to the same I / O as the track) held in the order management table 20. It is determined whether or not it is equal to the value obtained by adding 1. For example, when the track number SN of the track is LSN + 2 (SN ≠ LSN + 1), the track has been transferred before the track to which LSN + 1 is assigned as the track number. It can be determined that it is in a state of being damaged.
[0081]
Therefore, if SN ≠ LSN + 1, the track is made to wait until the serial number LSN of the order management table 20 is updated to SN = LSN + 1 by writing another track. In other words, even if an error occurs during track transfer and the order of tracks transferred to the sub center 1B is reversed and the predetermined order is lost, the order of the tracks is adjusted on the sub center 1B side. Since this can be done, the order of the tracks can be reliably guaranteed.
[0082]
In FIG. 8, only two communication lines 6-1 and 6-2 are illustrated for the sake of explanation. However, if three or more communication lines are provided, the communication line 6-1 is transferred during the transfer of the track T1. When a transfer error occurs due to a communication failure, the track T1 is transferred to the sub-center 1B side through a third communication line (not shown). At this time, the track T1 arrives at the sub-center 1B after the track T2. However, as described above, the remote FCU 10B waits for the track T2 to be written until the track T1 is written. Even if the track arrives after the track T2, writing is performed in the order of the tracks T1 and T2, and the order is not lost.
[0083]
Further, in the remote transfer method according to the second embodiment of the present invention, when the order guarantee mode is designated, two or more communication lines 6-1 and 6-2 can be used simultaneously unlike the first embodiment. The track transfer efficiency can be significantly improved as compared to the first embodiment.
[C] Description of the third embodiment
Next, a remote transfer method according to a third embodiment of the present invention will be described with reference to FIGS. FIG. 11 is a block diagram for explaining a remote transfer method as a third embodiment of the present invention, and FIG. 12 shows a remote transfer method as a third embodiment of the present invention in the time axis and ( It is a flowchart for demonstrating corresponding to 1)-(10). In FIG. 11, the same reference numerals as those described above indicate the same or substantially the same parts, and the description thereof is omitted.
[0084]
In order to realize the remote transfer method according to the third embodiment of the present invention, the host 1A and the remote FCUs 10A and 10B in the remote system of the third embodiment also have the following functions as in the second embodiment. Yes.
That is, also in the third embodiment, on the primary center 1A side, as in the second embodiment, the host 2A sequentially sends a plurality of tracks to the remote FCU 10A in a predetermined order. In the embodiment, only two communication lines 6-1 and 6-2 are used at the same time and transferred to the sub-center 1B side.
[0085]
Also in the third embodiment, when the order guarantee mode is designated, the remote FCU 10A transfers the track number (a1) and the ID value (a2) to the tracks transferred through the communication line 6-1 or 6-2. ).
However, in the remote FCU 10B on the sub-center 1B side in the third embodiment, the order management table 20 that holds the ID value and the track number (SN) described above is provided on the cache memory 16a secured on the SS 16. It has been. The major difference in configuration between the second embodiment and the third embodiment is that the order management table 20 is provided in the cache memory 16a instead of the RM 12. Since the order management table 20 is as described above in the second embodiment, the description thereof is omitted.
[0086]
In the second embodiment, the RM 12 determines whether or not track writing is possible in response to a request from the RA 14, whereas in the third embodiment, the order management table 20 includes a cache memory. The determination is made by the RA 14 by reading the order management table 20 on the cache memory 16a.
[0087]
That is, when the RA 14 in the remote FCU 10B of the third embodiment receives a track from the RA 14 on the primary center 1A side, the RA 14 reads the order management table 20 from the cache memory 16a, and receives the information in the order management table 20 and the received track. Are compared with the track number SN, and the following operations are performed under two conditions (c1) and (c2).
[0088]
Condition (c1): If the track number notified from the RA 14 on the primary center 1A side is “1”, the ID value and the track number are written in the unused area or the oldest area in the order management table 20. However, the ID value notified from the RA 14 is written as it is as the ID value, but “0” is first written as the track number. Thereafter, the RA 14 writes the track into the cache memory 16a. The track number “0” in the order management table 20 is rewritten to “1” when the writing of the track to the cache memory 16a is completed.
[0089]
Condition (c2): If the track number notified from the RA 14 on the primary center 1A side is other than “1”, a search is made as to whether the same ID value as that track is stored in the order management table 20 . When the same ID value is already held, the track number SN is compared with the track number LSN already held in the order management table 20, and it is determined whether or not SN = LSN + 1. If SN = LSN + 1, the RA 14 writes the track to the cache memory 16a. If SN ≠ LSN + 1, the RA 14 does not write the track to the cache memory 16a and waits for a while to write the track. After that, the operation is performed again to read the order management table 20 from the cache memory 16a. Even if the same track ID value is not held in the order management table 20, the RA 14 does not write the track to the cache memory 16a and waits for a while to write the track. After that, the sequence management table 20 is read again from the cache memory 16a.
[0090]
When the RA 14 of the remote FCU 10B determines that the track can be written as described above, the track is written to the cache memory 16a. After the writing is completed, the ID value same as the track ID value is obtained from the order management table 20. And the track number SN of the track is written as the LSN in the track number area linked to the ID value.
[0091]
Note that any of the characteristic functions / configurations of the third embodiment described above can be easily realized by slightly modifying the firmware (program) for realizing remote transfer that has been used conventionally. The
Next, the procedure of the remote transfer method according to the third embodiment will be described with reference to FIGS. At that time, reference is also made to FIGS. 10A to 10D used in the description of the second embodiment.
[0092]
As shown in FIGS. 11 and 12, when the order guarantee mode is designated by the host 2A on the primary center 1A side, the host 2A first stores the cache memory 16a secured on the SS 16 of the remote FCU 10A. Thus, the tracks T1 and T2 belonging to the same I / O are sequentially written [see (1) and (6)]. At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
[0093]
The remote FCU 10A then reads the track T1 previously written in the cache memory 16a by RA01 and transfers it to the remote FCU 10B on the secondary center 1B side using the communication line 6-1 (RA01 / RA03 path). At this time, an ID value = 0001 and a track number = 0001 are added to the track T1 [see (2)].
[0094]
Following this, the remote FCU 10A next transfers the track T2 written in the cache memory 16a, but since the communication line 6-1 (RA01 / RA03 path) is in use, the track T2 is transferred by RA02. Read and transfer to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path). At this time, an ID value = 0001 and a track number = 0002 are added to the track T2 [see (7)].
[0095]
In the remote FCU 10B on the secondary center 1B side, when the RA03 receives the track T1 transferred from the primary center 1A side, the RA03 reads the order management table 20 from the cache memory 16a and writes the track T1. Judging by all means. At this time, the RA 14 operates according to the condition (c1) described above, and determines that the writing of the track T1 is possible [see (3)]. By the operation of RA14, the order management table 20 is changed from the state shown in FIG. 10A to the state shown in FIG.
[0096]
Then, RA03 writes the track T1 to the cache memory 16a secured on the SS 16 [see (4)], and then updates the track number area corresponding to the track T1 of the order management table 20 on the cache memory 16a. The track number SN of the track T1 is reflected as LSN [see (5)]. Thereby, the order management table 20 changes from the state shown in FIG. 10B to the state shown in FIG.
[0097]
In the remote FCU 10B on the secondary center 1B side, the RA04 receives the track T2 from the primary center 1A side during the writing process of the track T1 (see (3) and (4)). The order management table 20 is read from the memory 16a to determine whether or not the track T2 is to be written. At this time, the order management table 20 is in the state shown in FIG. 10B, and RA04 operates according to the condition (c2) described above. That is, SN ≠ LSN + 1 (where SN = 2, LSN = 0, and 2 ≠ 0 + 1), and the track T1 is not written on the cache memory 16a. Therefore, the RA04 writes the track T2. Judgment is impossible (see (8) in FIG. 11 and (8) -1 in FIG. 12).
[0098]
The RA 04 that determines that the writing of the track T2 is impossible reads the repetitive order management table 20 from the cache memory 16a until the track number of the order management table 20 becomes “0001” ((8) in FIG. 11). FIG. 12 (8) -2].
In the meantime, the writing of the track T1 is completed by RA03 [see (4)], and when the registration to the ordering management table 20 is completed [see (5)], the ordering management table 20 is shown in FIG. Since it is updated as shown, and SN = LSN + 1 (where SN = 2, LSN = 1, and 2 = 1 + 1), RA04 determines that writing of track T2 is possible [(8 in FIG. 11 ), See (8) -3 in FIG.
[0099]
The RA 04 that has determined that the writing of the track T2 is possible writes the track T2 in the cache memory 16a secured on the SS 16 [see (9)], and then the track of the order management table 20 on the cache memory 16a. The track number area corresponding to T2 is updated to reflect the track number SN of the track T2 as LSN [see (10)]. Thereby, the order management table 20 changes from the state shown in FIG. 10C to the state shown in FIG.
[0100]
As described above, the remote transfer method according to the third embodiment of the present invention can provide the same effects as those of the second embodiment described above.
[D] Description of the fourth embodiment
By the way, when remote transfer is performed according to the second and third embodiments described above, a failure occurs in one line in a two-line configuration as shown in FIGS. 8 and 11, or there are three or more communication lines. However, if an unused communication line does not exist when a line failure occurs, the following transfer deadlock may occur.
[0101]
When a track is transferred, a transfer error occurs due to a line failure, there is no unused communication line to transfer the track in which the transfer error occurred, and in the track number of the track currently being transferred, It is assumed that a condition in which three conditions exist such that there is a track larger than the track number of the track in which the transfer error has occurred is generated.
[0102]
At this time, the track to be written after that track is transferred to the sub-center 1B first, and is waiting for the transfer errored track to be transferred. Therefore, there is a transfer deadlock that locks a track in which a transfer error has occurred in a state where the track cannot be transferred.
Therefore, in the remote transfer method according to the fourth embodiment of the present invention, when the above three conditions overlap, the remote FCU 10A on the primary center 1A side interrupts the transfer of the track in the transfer deadlock state, and A communication line that has been used for track transfer is used to transfer a track in which a transfer error has occurred, thereby avoiding a transfer deadlock.
[0103]
Such a remote transfer method as the fourth embodiment will be described with reference to FIGS. FIG. 13 is a flowchart for explaining the remote transfer method according to the fourth embodiment of the present invention. FIG. 14 shows the remote transfer method (when a transfer error occurs due to a line failure) according to the fourth embodiment of the present invention. FIG. 15 is a block diagram for illustrating the remote transfer method (when a transfer error occurs due to a line failure) as the fourth embodiment of the present invention, corresponding to the time axis and (1) to (7) in FIG. It is a flowchart for making it explain.
[0104]
The remote transfer method according to the fourth embodiment can be applied to both the remote transfer method according to the second embodiment and the third embodiment, but the fourth embodiment is applied to the remote transfer method according to the third embodiment. The case will be described.
The remote FCU 10A on the primary center 1A side in the fourth embodiment is obtained by adding the functions shown in the flowchart (steps S1 to S5) in FIG. 13 to the remote FCU 10A in the third embodiment.
[0105]
That is, as shown in FIG. 13, when a transfer error occurs due to a line failure during transfer of a certain track, the remote FCU 10A determines whether there is another unused transfer path (communication line) among a plurality of communication lines. If there is an unused transfer path (YES route from step S1), the unused transfer path is used to transfer a track in which a transfer error has occurred (step S2). .
[0106]
On the other hand, when there is no unused transfer path (NO route from step S1), the remote FCU 10A sends the track number assigned to the track currently being transferred and the track number assigned to the track in which the transfer error has occurred. To determine whether there is a number larger than the track number of the track in which the transfer error occurred among the track numbers being transferred (step S3).
[0107]
If the corresponding serial number does not exist (NO route from step S3), the remote FCU 10A waits for the transfer of the track in which the transfer error has occurred until one of the transfer paths in use is released (step S4). If the corresponding serial number exists (YES route from step S3), the transfer of the track having the corresponding serial number is interrupted, and a transfer error is generated using the transfer path that was being used for the transfer of the track. The generated track is transferred (step S5).
[0108]
Next, a remote transfer method (when a transfer error occurs due to a line failure) according to the fourth embodiment, to which the function described above with reference to FIG. 13 is added, will be described with reference to FIGS.
As shown in FIGS. 14 and 15, when the order guarantee mode is designated by the host 2A on the primary center 1A side, the host 2A first stores the cache memory 16a secured on the SS 16 of the remote FCU 10A. Thus, the tracks T1 and T2 belonging to the same I / O are sequentially written (see (1) and (3)). At this time, it is assumed that neither of the two communication lines 6-1 and 6-2 is used.
[0109]
The remote FCU 10A then reads the track T1 previously written in the cache memory 16a by RA01 and transfers it to the remote FCU 10B on the secondary center 1B side using the communication line 6-1 (RA01 / RA03 path). At this time, an ID value = 0001 and a track number = 0001 are added to the track T1 (see (2) in FIG. 14 and (2) -1 in FIG. 15).
[0110]
Following this, the remote FCU 10A next transfers the track T2 written in the cache memory 16a. Since the communication line 6-1 (RA01 / RA03 path) is in use, the track T2 is transferred by RA02. Read and transfer to the remote FCU 10B on the secondary center 1B side using the communication line 6-2 (RA02 / RA04 path). At this time, an ID value = 0001 and a track number = 0002 are added to the track T2 (see (4) in FIG. 14 and (4) -1 in FIG. 15).
[0111]
However, when a transfer error occurs due to a line failure on the communication line 6-1 during the transfer of the track T1 and the transfer of the track T1 fails, the remote FCU 10A searches for an unused path as an alternative transfer path. Since the other communication line 6-2 (RA02 / RA04 path) is also in use, it is determined that there is no unused path (see (2) in FIG. 14 and (2) -2 in FIG. 15).
[0112]
In the meantime, in the remote FCU 10B on the secondary center 1B side, the RA 04 receives the track T2 from the primary center 1A side, and this RA 04 reads the order management table 20 from the cache memory 16a and writes the track T2. Since the track T1 is not written on the cache memory 16a, the RA04 determines that the track T2 cannot be written (see (4) in FIG. 14 and (4) -2 in FIG. 15). The RA 04 that determines that the writing of the track T2 is impossible reads the repetitive order management table 20 from the cache memory 16a until the track number in the order management table 20 becomes “0001” ((4) in FIG. ▼, see (4) -3 in FIG. 15).
[0113]
In parallel with this, the remote FCU 10A on the primary center 1A side has the track number “1” of the path that failed to be transferred and the track number “of the path being transferred (communication line 6-2; RA02 / RA04 path)” Since it is 2 ″, an interruption instruction is output to interrupt the transfer through the RA02 / RA04 path (see (2) in FIG. 14 and (2) -3 in FIG. 15).
[0114]
As a result, RA02 issues an interruption instruction to RA04 (see (5)), and RA04 interrupts the process for track T2 in response to the interruption instruction from RA02 (see (5) ').
Thereafter, the remote FCU 10A uses the communication line 6-2 (RA02 / RA04 path) to the remote FCU 10B on the secondary center 10B side to transfer the track T1 that has failed to be transferred (see (6)). Since the operation of the remote FCU 10B at this time is the same as that of the third embodiment, a description thereof will be omitted.
[0115]
Then, the remote FCU 10A uses the communication line 6-2 (RA02 / RA04 path) to the remote FCU 10B on the secondary center 10B side to transfer the track T1 that failed to be transferred (see (7)). Since the operation of the remote FCU 10B at this time is the same as that of the third embodiment, the description thereof is omitted.
As described above, according to the remote transfer method according to the fourth embodiment of the present invention, even when a predetermined condition overlaps and a transfer deadlock occurs, the remote FCU 10A on the primary center 1A side tracks the transfer deadlock state. The transfer deadlock is reliably avoided and the track is transferred reliably by interrupting the transfer of the track and transferring the track that caused the transfer error using the communication line that was used for the transfer of the track. can do.
[0116]
[E] Other
The present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the spirit of the present invention.
For example, the case where 10A and 10B are connected by two communication lines 6 is described, but the present invention is not limited to this, and the connection is made by three or more communication lines 6. May be.
[0117]
【The invention's effect】
As described in detail above, the remote transfer method by the magnetic disk control apparatus of the present invention (claim 1)~ 3)Even if an error occurs during track transfer and the order of tracks transferred to the sub-center is reversed and the predetermined order is lost, the track order can be adjusted on the sub-center side. It is possible to guarantee the order of
[0119]
Further, according to the remote transfer method (claims 1 and 4) by the magnetic disk control device of the present invention,When a transfer deadlock occurs due to overlapping of predetermined conditions, the transfer of the track in the transfer deadlock state is interrupted by the main center side magnetic disk controller, and the communication line used for the transfer of the track is used. By transferring the track that caused the transfer error, transfer deadlock can be reliably avoided and the track can be transferred reliably.The
[Brief description of the drawings]
FIG. 1 is a sequence diagram for explaining a remote transfer method according to a first embodiment of the present invention;
FIG. 2 is a block diagram for explaining a remote transfer method according to the first embodiment of the present invention;
FIG. 3 is a flowchart for explaining a remote transfer method according to the first embodiment of the present invention in correspondence with a time axis and (1) to (8) in FIGS. 1 and 2;
FIG. 4 is a block diagram for explaining a remote transfer method (when a transfer error occurs) as a first embodiment of the present invention;
FIG. 5 is a sequence diagram for explaining a remote transfer method (when a transfer error occurs) as a first embodiment of the invention.
6 is a flowchart for explaining a remote transfer method (when a transfer error occurs) according to the first embodiment of the present invention in correspondence with a time axis and (1) to (9) in FIGS. 4 and 5. FIG. It is.
FIG. 7 is a diagram showing an example of an order management table used in the remote transfer method according to the second embodiment of the present invention.
FIG. 8 is a block diagram for explaining a remote transfer method as a second embodiment of the present invention;
FIG. 9 is a flowchart for explaining the remote transfer method according to the second embodiment of the present invention in correspondence with the time axis and (1) to (10) in FIG. 8;
FIGS. 10A to 10D are diagrams illustrating registration states of an order management table when a remote transfer method according to a second embodiment of the present invention is executed.
FIG. 11 is a block diagram for explaining a remote transfer method as a third embodiment of the present invention;
12 is a flowchart for explaining a remote transfer method according to a third embodiment of the present invention in correspondence with a time axis and (1) to (10) in FIG.
FIG. 13 is a flowchart for explaining a remote transfer method as a fourth embodiment of the invention;
FIG. 14 is a block diagram for explaining a remote transfer method (when a transfer error occurs due to a line failure) as a fourth embodiment of the present invention;
FIG. 15 is a flowchart for explaining a remote transfer method (when a transfer error occurs due to a line failure) as a fourth embodiment of the present invention in association with the time axis and (1) to (7) in FIG. 14; It is.
FIG. 16 is a block diagram showing a conventional configuration of a system having a primary center for performing original business and a backup sub-center.
FIG. 17 is a block diagram showing an example of a general configuration of a remote system in which remote FCUs in a main center and a sub-center are connected by a communication line.
FIG. 18 is a block diagram showing a configuration of a general remote FCU.
FIG. 19 is a block diagram for explaining data writing / transfer in which order is maintained.
20 is a flowchart for explaining data writing / transfer in which order is maintained in association with the time axis and {circle around (1)} to {circle around (4)} in FIG.
FIG. 21 is a block diagram for explaining data writing / transfer in which order is not maintained.
22 is a flowchart for explaining data writing / transfer in which order is not maintained in correspondence with a time axis and (1) to (5) in FIG. 21. FIG.
FIGS. 23A and 23B are diagrams showing specific examples of two tracks to be transferred while guaranteeing the order; FIG.
FIG. 24 is a diagram for explaining the operation of the sub center-side remote FCU when the two tracks shown in FIGS. 23 (a) and 23 (b) are transferred while their order is guaranteed.
FIG. 25 is a diagram for explaining the operation of the sub center-side remote FCU when the transfer order of the two tracks shown in FIGS. 23 (a) and 23 (b) is reversed.
[Explanation of symbols]
1A Main center (business system)
1B Sub-center (backup system)
2A Main center host (CPU)
2B Secondary center host (CPU)
4A Main center side magnetic disk unit (volume)
4B Secondary center side magnetic disk unit (volume)
6 Communication line
10A Main center side remote FCU (Main center side magnetic disk control device)
10B Sub-center side remote FCU (sub-center side magnetic disk controller)
11 CA (Channel Adapter)
12 RM (Resource Manager)
13 TS (Table Storage)
14 RA (Remote Adapter)
15 CFE (Cache Function Engine)
16 SS (Shared Storage)
16a cache memory
17 DA (Device Adapter)
18 C-BUS (Control-BUS)
20 Order management table (Table)

Claims (4)

A primary center and a secondary center having a magnetic disk device and a magnetic disk control device for controlling the magnetic disk device are provided, and a plurality of spaces are provided between the primary center side magnetic disk control device and the secondary center side magnetic disk control device. In a remote system connected by a communication line, a remote transfer method for transferring data from the primary center side magnetic disk control device to the secondary center side magnetic disk control device through the plurality of communication lines,
A sequence center guarantee mode for continuously transferring a plurality of tracks from the primary center side magnetic disk controller to the secondary center side magnetic disk controller while guaranteeing the order is designated by the primary center host. If
The primary center side magnetic disk control device has the same identification number and a serial number corresponding to the order of transmission from the primary center host for a plurality of tracks transferred by the same input / output request from the primary center host. And simultaneously transferring the plurality of tracks to which the identification number and serial number are assigned to the secondary center side magnetic disk controller via the plurality of communication lines,
The secondary center side magnetic disk control device comprises:
An identification number assigned to a track transferred from the primary center and written on the secondary center side, and a serial number LSN assigned to the track having the identification number and last written on the secondary center side; With an order management table that holds
When a new track is transferred from the main center side, the serial number SN assigned to the track and the identification number assigned to the track held in the order management table Compare with serial number LSN,
When SN ≠ LSN + 1, the writing of the track is not permitted to the track until the serial number LSN of the order management table is updated to SN = LSN + 1 by writing another track. While waiting
When SN = LSN + 1, writing permission is given to the track, and when the writing of the track to the sub-center side magnetic disk device is completed, the track is set as the serial number LSN of the order management table. the addition to register and update the serial number LSN of,
If a transfer error occurs due to a line failure when transferring a track,
The primary center side magnetic disk control device comprises:
Check whether there are other unused communication lines among the plurality of communication lines,
When there is an unused communication line, the unused communication line is used to transfer the track in which a transfer error has occurred,
If there is no unused communication line, does the serial number assigned to the track currently being transferred contain a serial number greater than the serial number assigned to the track that caused the transfer error? Determine whether
If the corresponding serial number does not exist, while waiting for the transfer of the track that caused the transfer error until the communication line in use is released,
If the corresponding serial number exists, the transfer of the track having the corresponding serial number is interrupted, and the transfer of the track in which a transfer error has occurred is transferred using the communication line that was used for the transfer of the track. and performing remote transmission method according to the magnetic disk control device. The sequence management table, characterized in that it is managed by the resource module for managing / controlling the resources of the sub center side magnetic disk control device, the remote transmission method according to the magnetic disk control device according to claim 1. The sequence management table, characterized in that are provided to temporarily cache memory for writing before writing the track to the sub OLT magnetic disk device, a magnetic disk control apparatus according to claim 1, wherein Remote transfer method by. A primary center and a secondary center having a magnetic disk device and a magnetic disk control device for controlling the magnetic disk device are provided, and a plurality of spaces are provided between the primary center side magnetic disk control device and the secondary center side magnetic disk control device. In a remote system connected by a communication line, a remote transfer method for transferring data from the primary center side magnetic disk control device to the secondary center side magnetic disk control device through the plurality of communication lines,
A sequence center guarantee mode for continuously transferring a plurality of tracks from the primary center side magnetic disk controller to the secondary center side magnetic disk controller while guaranteeing the order is designated by the primary center host. If
The primary center side magnetic disk control device has the same identification number and a serial number corresponding to the order of transmission from the primary center host for a plurality of tracks transferred in response to the same input / output request from the primary center host. And simultaneously transferring the plurality of tracks to which the identification number and serial number are assigned to the secondary center side magnetic disk controller via the plurality of communication lines,
If a transfer error occurs due to a line failure when transferring a track,
The primary center side magnetic disk control device comprises:
Check whether there are other unused communication lines among the plurality of communication lines,
When there is an unused communication line, the unused communication line is used to transfer the track in which a transfer error has occurred,
If there is no unused communication line, does the serial number assigned to the track currently being transferred contain a serial number greater than the serial number assigned to the track that caused the transfer error? Determine whether
If the corresponding serial number does not exist, while waiting for the transfer of the track that caused the transfer error until the communication line in use is released,
If the corresponding serial number exists, the transfer of the track having the corresponding serial number is interrupted, and the transfer of the track in which a transfer error has occurred is transferred using the communication line that was used for the transfer of the track. and performing remote transmission method according to magnetic disk controller.

Images