JP4131953B2 - File control unit - Google Patents

Description

  The present invention relates to a file control apparatus such as a disk control apparatus that performs asynchronous transfer in which a transfer path between a channel transfer and a direct access storage device (DASD), which is a host device, operates independently in time.

  FIG. 7 is an explanatory diagram of a conventional example, FIG. 7A is a description of a CKD format, and FIG. 7B is a description of a method for emulating a CKD track on an FBA track.

  Conventionally, a method for storing data in a direct access storage device (DASD), which is an external storage device such as a magnetic disk device or an optical disk device, is a CKD format that is a variable length method, and a host device (OS: operating system) is used. The format for writing data to DASD is also the CKD format.

  7A, a track m having a CKD format disc includes an index (Index), a home address HA, a record (Record) 0, a record 1, a record 2,..., A record k.

  The index is a track origin marker. The home address HA indicates a track state and the like, and is provided with a flag, a track address (Track Address: cylinder number and head number), and an error correction code (Error Correction Code: ECC). This flag (Flag) indicates whether or not the track is defective. For example, the flag (Flag) is composed of 2 bits. If the target track is normal, it is “00”. It is what. The error correction code ECC is for data check.

  Record 0 is a record that cannot be used by the user, and records the address of the replacement track when a bad track occurs in the count section. Records 1 to k are records that can be used by the user, and each has three fields: a count part C, a key part K (may be omitted), and a data part D. The count part C contains the lengths of the key part K and the data part D that follow it. The key part K contains a key which is an indicator for searching the data part D. The data part D is a part for recording data.

  However, in recent years, the RAID (RAID) technology has been used to improve reliability by adding redundant information by changing the actual storage medium from DASD in the CKD format, which is a variable length method, to DASD in the FBA format, which is a fixed length method. The tendency to do is increasing.

  In this case, since the format issued by the host device (OS) is different from the actual DASD format, the file control device (hereinafter referred to as “FCU”) emulates (converts) the CKD format on the FBA format DASD. (Ckd On Fba = COF) is spreading.When this COF technology is used, a track in the CKD format is emulated on an actual DASD in the FBA format as shown in FIG.

  7B, the upper diagram shows one track of the FBA format, and in this example, there are block n, block n + 1, block n + 2, block n + 3, block n + 4, block n + 5... Block n + 84. One block can contain 0 to a maximum of 8 records in the CKD format.

  For example, when one track in the CKD format uses the block n to the block n + 15 (16 blocks) in the FBA format, the upper diagram in FIG. 7B looks as shown in the lower diagram. That is, in one FBA format track, the emulated CKD format tracks are track m (FBA blocks n to n + 15), track m + 1 (FBA blocks n + 16 to n + 31), track m + 2 (FBA blocks n + 32 to n + 47), track m + 3 (FBA). Block n + 48 to n + 63) and track m + 4 (FBA block n + 64 to n + 79), which is about 5 tracks.

  Conventionally, in the write operation, in order to speed up the processing, at the time of a write request from a host device, the data to be written is not written to the DASD, but a non-volatile memory (FCU) in the file control unit (FCU). A method of writing data only to NVS = Non Volatile Storage) and using the FCU idle time to write back to DASD (high-speed writing mechanism to DASD) is considered.

  Further, as a method of writing back data from NVS to DASD, which is an improvement of this method, a method of writing back a plurality of continuous tracks to DASD in one event (multitrack destage) is considered.

  However, in a file subsystem using the COF technology, destaging of continuous tracks cannot always expect an improvement in performance. This is because two consecutive tracks in the CKD format become adjacent blocks in the FBA format, and in the FCU based on the one-track processing in the CKD format, after the track is written back, the post-processing is performed. This is because the leading portion of the following track passes and generates a waiting for one rotation.

  Referring to FIG. 7, for example, when the conventional CKD format DASD is used, the time from the destaging of the track m to the arrival of the head of the DASD at the home address HA of the track m + 1 is the record of the track m. This is the time from when the processing of k is completed until the head passes the index and is further positioned at the home address HA portion of the track m + 1.

  However, when the COF technique is used, only the time corresponding to the inter-block gap G is given from the end of the process of the track m to the start of the process of the track m + 1. Logically, the FCU cannot complete the track end processing such as error check and the subsequent track start processing only by this time. For this reason, the head waits for the disk to make one revolution and to be positioned at the beginning of the track m + 1 (block n + 16 in the FBA format) again.

  In general, a response to a read command (Read Command) that is a read command from a host device is read out from the cache memory if the target data is stored in the cache memory, and the target data is stored in the cache memory. Otherwise, access to the actual DASD occurred, and the time for positioning the head to the DASD and waiting for rotation was spent.

  The above-described conventional devices have the following problems.

  (1) The problem that the multi-track destaging method, which is a method for eliminating DASD rotation waiting by continuously destaging, causes rotation waiting when COF technology is used, and fails to perform its function. was there.

  (2) Response to general read commands from the host device consumes time to perform hash conversion to obtain the storage location and read operation from the cache memory even if the target data is stored in the cache memory In addition, the probability of occurrence of an access error due to a hardware failure or the like has increased.

  The present invention solves the above-mentioned conventional problems and does not process the logical CKD destage track to DASD continuously, but ensures sufficient time for post-processing and pre-processing of the track, and wasteful rotation. It can respond to an instruction to read the home address at a high speed (Read HA instruction) regardless of whether the target data is stored in the cache memory without waiting, and whether the target data is stored in the cache memory. The purpose is to reduce the probability of occurrence.

  The present invention is configured as follows to solve the above problems.

  FIG. 1 is a diagram illustrating the principle of the present invention. In FIG. 1, 1 is a host device, 2 is a file control unit (FCU), 3 is a direct access storage device (DASD), 20 is a memory unit, and 21 is a channel adapter (CA). ), 22 is a device adapter (DA), 23 is a cache management unit (CFE), 24 is a cache memory, 25 is a nonvolatile memory (NVS), and 27 is a defective replacement information table.

In the FCU 2 having the cache memory 24 for controlling the DASD 3, the storage unit of the FCU 2 has a storage unit for the defective replacement information table 27 and a cache management unit 23 provided in the FCU 2, and the cache management unit 23 is turned on. At the time of resetting or at the time of resetting, the bad track areas of all the subordinate DASDs 3 are read out, and it is determined whether or not a defective alternate link has been created by the flag stored in the home address of each track, and the result is stored in the storage unit When the read instruction for the home address including the flag indicating the use state of whether the track is defective and the home address including the track address is received from the higher-level device 1 as the failure change information table 27, the FCU 2 From the seek command to specify each track issued to The track address of the rack is obtained, and based on the track address, whether or not the defective track information can be read from the home address of the DASD 3 is determined to determine whether or not the defective replacement information table 27 is valid, and the received home address Flag information for the read command is determined, and response information for the read command for the home address is created based on the track address and the flag information . Therefore, it is possible to respond to a command for reading the home address at high speed.

  Also, when the DASD 3 is a fixed-length array disk and the FCU 2 emulates a variable-length system with the array disk, the track flaw information that is a response to the read command of the home address including additional information is standardized. Based on the flaw information, the flag information, and the track address, response information for a home address read command including additional information is created. For this reason, a response to the read instruction of the home address including the additional information when the defect replacement information table 27 is valid can be performed at high speed without accessing the cache memory or the like, and the error occurrence probability can be lowered. it can.

  The present invention has the following effects.

  (1) When the power is turned on or reset, the FCU reads out the defective track areas of all the subordinate DASDs 3 and determines whether a defective replacement link has been created based on the flag stored in the home address portion of each track. The result is stored in the storage unit as a defective replacement information table, and when a home address read command is received, a response is received with invalid defective replacement information table information to determine whether the defective replacement information table is valid. Can be prevented.

  (2) When the FCU determines that the defective replacement information table is invalid, the information is transferred from the cache memory if the target home address exists on the cache memory, and is read from the DASD 3 if it does not exist. When the replacement information table is valid, the failure replacement information table 27 is referred to by referring to the failure replacement link information in the failure replacement information table and determining flag information which is a part of the response to the read instruction of the received home address. When invalid, flag information can be returned, and when valid, flag information can be created at high speed.

  (3) To create complete response information for the home address read command from the confirmed flag information and the track address that is part of the response to the home address read command using the already received seek address. In addition, when the defective replacement information table is valid, a response to the read instruction of the home address can be performed at high speed.

  (4) When the FCU emulates a variable length method with an array disk, the track flaw information that is a part of the response to the read command of the home address including the additional information is fixed to a standard value, and the flaw information In order to create response information for the home address read command including complete additional information from the flag information and the track address, the response to the home address read command including the additional information when the defective replacement information table is valid is cached. It can be performed at high speed without accessing the memory or the like, and the error occurrence probability can be reduced.

  2 and 3 show the first embodiment of the present invention, and FIGS. 4 to 6 show the second embodiment of the present invention. Embodiments of the present invention will be described below with reference to the drawings.

[Description of the first embodiment]
1) Description of File Control Unit (FCU) FIG. 2 is an explanatory diagram of the file control unit. The file control unit (FCU) 2 reads / writes data to / from the DASD 3 according to an instruction from the host unit. The host apparatus is connected by a channel adapter (CA) 21, and the DASD adapter 31 of DASD 3 such as a disk array is connected by a device adapter (DA) 22. Further, the FCU 2 is provided with a service adapter (SA) 26 connected to the hard desk (HD) 4.

  In the FCU 2, a channel adapter 21, a device adapter 22, a cache management unit 23, a cache memory 24, and a nonvolatile memory (NVS) 25 are provided.

  A plurality of channel adapters 21 are provided, which mainly control the interface with the channel of the host device, and the device adapter 22 controls the interface with the DASD 3. The cache management unit 23 manages tracks and the like registered in the cache memory 24.

  The cache memory 24 is a buffer storage device that holds a copy of instructions and data on the DASD 3 that are likely to be used later. The nonvolatile memory 25 is backed up by a battery, and data is not lost even when the power is turned off.

  The service adapter 26 performs initial setting of the file control device, management of a history of error information of the file device, and the like based on information from the hard disk 4. DASD3 is a storage device of FBA format (fixed length system) such as a disk array using RAID technology. The DASD adapter 31 serves as an interface with the FCU 2.

2): Explanation of the operation of the file control device When the FCU 2 receives an instruction to write data to the DASD 3 from the host device, the channel adapter 21 converts the CKD format from the host device into the FBA format, and the cache memory 24 and the nonvolatile memory The high-speed writing process for simultaneously writing to 25 is performed.

  When the cache management unit 23 detects that there is a track generated by the high-speed writing mechanism in the nonvolatile memory 25, the cache management unit 23 searches whether there is a continuous track among all the tracks. Here, the cache management unit 23 calculates a track interval at which continuous destaging is possible, and requests the device adapter 22 to destage using the destage start track, end track and destage track interval as information.

  Upon receiving the destage request, the device adapter 22 issues a selection command for positioning the head to the DASD 3 with respect to the head track based on the information.

  After the DASD 3 selection is completed, the device adapter 22 determines a track to be destaged based on the information received from the cache management unit 23.

  The first destaged track is the first track notified from the cache management unit 23, and the next destaged track is a track obtained by adding the destage track interval to the first track. Further, the next destage target track is a track obtained by adding the destage track interval to the current destage track.

  The device adapter 22 repeats this operation until the end track address notified from the cache management unit 23 is exceeded.

  Note that the destage track interval is an interval at which the destage target track is skipped by an amount that allows a sufficient time for track post-processing and pre-processing to be secured. In the conventional multi-track destage, wasteful waiting for rotation corresponding to the number of destaged tracks has occurred, but by using this embodiment, useless waiting for rotation does not occur.

3): Description of Destage Operation by Specific Example FIG. 3 is an explanatory diagram of a destage target track.

  The cache management unit 23 detects that there is a destage target track on the nonvolatile memory 25. This time, it is assumed that the target DASD 3 has 15 heads per cylinder, and this destage target track is as shown in FIG. For the target DASD 3, if the post-processing of the track and the pre-processing can be completed in the rotation time of 3 tracks, the continuous destage track interval is calculated to be 4.

  Here, the tracks to be destaged this time are the tracks marked with a circle in the first column in FIG. 3 and the tracks marked with a ● in the second column.

  (1): The cache management unit 23 searches the destage track, and as the first time, the first track (0) = cylinder value “0005”, the head value “0000”, and the end track (+20) = cylinder value “0006” “Destage is issued to the device adapter 22 with the head value“ 0005 ”and the continuous destage track interval = 4.

  (2): The device adapter 22 that has received the destage request from the cache management unit 23 performs DASD3 selection for the cylinder value “0005” and the head value “0000” that are the track (0).

  (3): After the selection of DASD3 is completed, the device adapter 22 starts from the first track (0) in the first column in FIG. 3 and tracks (+4) (+8) (+12) with a circle in order. After destaging (+16) and completing the destage of the end track (+20), the completion of continuous destage processing is reported to the cache management unit 23.

  (4): The cache management unit 23, as the second time, starts track (+11) = cylinder value “0005” / head value “000B”, end track (+23) = cylinder value “0006”, head value “0008”, Destage is issued to the device adapter 22 with continuous destage track interval = 4.

  (5): The device adapter 22 that has received the destage request from the cache management unit 23 selects DASD3 for the cylinder value “0005” and the head value “000B” that are the track (+11).

  (6): After the DASD3 selection is completed, the device adapter 22 destages the tracks (+15) (+19) marked with ● in order from the first track (+11) in the first column in FIG. Then, after the destage of the end track (+23) is completed, the completion of the continuous destage processing is reported to the cache management unit 23.

  In the above embodiment, the continuous destage track interval is set to 4, but the cache management unit 23 determines the type of the destage target DASD 3 and calculates the continuous destage track interval based on the performance of the target DASD 3. Normally, track pre-processing and post-processing can be performed while one track is skipped. In this case, the continuous destage track interval is 2.

  Also, by storing the number of continuous destage tracks in advance in the hard disk 4 or the like, the number of continuous destage tracks can be defined in advance by the initial setting of the FCU. Thereby, the cache management unit 23 extracts continuous tracks that can be destaged from the predetermined number of continuous destage tracks.

  Then, when extracting the destageable continuous tracks, the cache management unit 23 postpones destage if the destageable continuous tracks are less than the number of predetermined continuous destage tracks, When the number of continuous destage tracks is larger than the predetermined number, the device adapter 22 is requested to destage the number of the predetermined continuous destage tracks.

  Furthermore, after finishing the continuous destage processing, the device adapter 22 compares the number of tracks destaged by the continuous destage with the predefined number of continuous destage tracks, and if the comparison value is different, Provide a mechanism for notifying the cache management unit 23 that a logical contradiction has been detected, and a mechanism for notifying the cache management unit 23 that destaging has been completed normally if the comparison values are equal. Provide.

[Description of the second embodiment]
1): Explanation of File Control Unit (FCU) FIG. 4 is an explanatory diagram of the file control unit in the second embodiment. The file control unit (FCU) 2 reads / writes data to / from the DASD 3 according to instructions from the host unit. It is connected to the host device by a channel adapter (CA) 21 and connected by a DASD adapter 31 of DASD 3 and a device adapter (DA). Further, the FCU 2 is provided with a service adapter (SA) 26 connected to the hard desk (HD) 4.

  In the FCU 2, a channel adapter 21, a device adapter 22, a cache management unit 23, a cache memory 24, and a nonvolatile memory (NVS) 25 are provided.

  A plurality of channel adapters 21 are provided, which mainly control the interface with the channel of the host device, and the device adapter 22 controls the interface with the DASD 3. The cache management unit 23 manages tracks and the like registered in the cache memory 24, and the storage unit is provided with a place for storing the defective replacement information table 27.

  The cache memory 24 is a buffer storage device that holds a copy of instructions and data on the DASD 3 that are likely to be used later. The nonvolatile memory 25 is backed up by a battery, and data is not lost even when the power is turned off.

  The service adapter 26 performs initial setting of the file control device, management of a history of error information of the file device, and the like based on information from the hard disk 4. The DASD 3 is a storage device including a plurality of devices such as a disk device and an optical disk device. The DASD adapter 31 serves as an interface with the FCU 2.

2): Explanation of the defective replacement information table FIG. 5 is an explanatory diagram of the defective replacement information table, FIG. 5 (a) is a description of the defective replacement information table, and FIG. 5 (b) is a description of one set of defective replacement information. It is.

  In FIG. 5A, in this example, the failure replacement information per device can be stored up to the maximum number of 45 tracks as set numbers 0 to 44 at the left end. One set of the defective replacement information includes storage units for 4 bytes on each of the defective side and the replacement side. In addition, the defect replacement information table 27 is provided with storage units for the maximum number of devices that can be connected (in this example, 256 devices).

  In FIG. 5B, one set of defective replacement information is divided into a defective side and a replacement side. A storage unit for cylinder values, head values, and information is provided on the defective side, and a storage unit for cylinder values, head values, and device values (machine numbers) is provided on the replacement side. For example, if “40” is stored, the storage unit of this information indicates that the replacement side is valid, and “20” indicates that the defective side is valid. This indicates that both the defective side and the replacement side are effective, that is, the link has been established.

3): Description of Read Read HA Command of Home Address Including Additional Information In this embodiment, the failure replacement information table 27 in the file control unit (FCU) 2 stores the failure replacement information of all the subordinate DASDs 3. Is stored. This defective replacement information is created and updated at the time of processing a home address write (Write HA) or a home address write including additional information (Diag Write HA).

  Data to be transferred by a home address read command (Diag Read HA Command) including additional information includes a flag, a track address, a flaw information, a cell value, and a physical track address.

  In this embodiment, since the defective change information of the track is stored in the FCU 2, the FCU 2 can create the flag information of the home address HA part. Since the track address and the physical track address are the physical position of the track, they should match the seek value, which can be created in the FCU 2.

  The flaw information is a standard value with no flaws when there is an array disk using RAID technology, and there is no problem. The cell value indicates the distance from the index, and is a fixed value when the track is not damaged.

  The data transferred by the home address read command (Read HA Command) is a flag and a track address, and is part of the information required by the home address read command (Diag Read HA Command) including additional information. After all, it can be created in FCU2.

  In this way, by storing the defect replacement information in the FCU 2, the Read HA Command and the Dia Read HA Command are always accessed without accessing the actual DASD 3 or without reading data from the cache memory. Target information can be sent out.

4): Explanation of operation of file control unit (FCU) Hereinafter, the operation of FCU 2 will be explained.

  (1): The FCU 2 (cache management unit 23) reads out the defective track areas of all subordinate DASDs 3 when the FCU 2 is turned on or reset, and uses the flags stored in the home address HA section of each track. Then, it is determined whether or not a defective replacement link has been created, and the result is stored as a defective replacement information table 27 in a memory (storage unit) in the FCU 2.

  (2): In order to issue a read command for DASD3 access, a seek command that identifies the target track must be issued in advance. For this reason, the FCU 2 can recognize the target track address from the seek command issued from the channel of the host device prior to the read command.

  (3): When the FCU 2 receives a read instruction (Read HA Command) of the subordinate DASD 3 home address from the channel of the host device, the FCU 2 determines whether or not the defect replacement information table 27 is valid.

  (4): If, as a result of this determination, the failure alternation information table 27 is invalid (for example, failure to read defective track information from the home address HA portion of DASD 3), the FCU 2 sets flag information of the target home address HA. Therefore, as usual, if there is data on the cache memory 24, the data is read from the cache memory 24, and if not, it is read from the DASD 3 and a normal read operation is performed.

  (5): If, as a result of the determination, the defective replacement information table 27 is valid, the defective replacement link information in the defective replacement information table 27 is referred to to determine whether a defective replacement link has been created for the target track. (Determining whether the track is a defective track, a replacement track, or a normal track).

  (6): If a defective alternate link has been created, a bit corresponding to the flag area of the home address HA section is set. Also, the cylinder head value, which is the track address of the home address HA part, always matches the physical cylinder head value, so by setting the value already specified by the seek instruction, the cache memory is not used. Can respond with an appropriate value.

  (7): The data that the FCU 2 has to transfer with the read instruction of the home address including the additional information (Diag Read HA Command) is the data to be transferred with the read instruction of the home address (Read HA Command). This is a track flaw information and cell value added. However, in the FCU 2 using the array disk (FCU using the COF technology) 2, the flaw information can always be specified as a standard value, so that the information can be created inside the FCU 2. For this reason, the FCU 2 can transfer an appropriate value also to the Diag Read HA Command.

  FIG. 6 is a process flowchart for the Read HA Command. Hereinafter, a description will be given according to the processes S1 to S5 of FIG.

  S1: The FCU 2 determines whether or not the defective replacement information table 27 is valid when the Read HA Command of the subordinate DASD 3 is received from the host device. If the defective replacement information table 27 is valid, the defective replacement information is determined. With reference to the defective replacement link information in the table 27, it is determined whether or not the target track is a defective replacement track. If it is determined that the target track is a defective replacement track, the process proceeds to step S2. If the target track is not a defective replacement track, the process proceeds to step S5 (in the case where the defective replacement information table 27 is invalid, the cache memory 24 or the DASD 3). Read, perform normal read operation).

  S2: The FCU 2 determines whether the target track is a defective track. If it is determined that the target track is a defective track, the process proceeds to step S3. If the target track is not a defective track, the process proceeds to step S4.

  S3: The FCU 2 sets the defective track information in the flag area of the home address HA portion of the target track, and proceeds to processing S5.

  S4: The FCU 2 sets the replacement track information in the flag area of the home address HA portion of the target track, and proceeds to processing S5.

  S5: The FCU 2 notifies the high-order apparatus of the defective replacement information set in the flag area and the cylinder head value that is the track address, and ends this processing.

It is a principle explanatory view of the present invention. It is explanatory drawing of the file control apparatus in 1st Example. It is explanatory drawing of the destage object track | truck in 1st Example. It is explanatory drawing of the file control apparatus in 2nd Example. It is explanatory drawing of the defect replacement information table in 2nd Example. It is a process flowchart with respect to Read HA Command in 2nd Example. It is explanatory drawing of a prior art example.

Explanation of symbols

1 Host device 2 File control unit (FCU)
3 Direct Access Storage Device (DASD)
20 Memory unit 21 Channel adapter (CA)
22 Device adapter (DA)
23 Cache Management Department (CFE)
24 cache memory 25 non-volatile memory (NVS)
27 Defect change information table

Claims (2)

In a file control device having a cache memory for controlling a direct access storage device,
A storage unit for a defective replacement information table in a storage unit of the file control device;
A cache management unit provided in the file control device;
The cache management unit
When power is turned on or reset, the bad track area of all the direct access storage devices under it is read, and it is determined whether a bad replacement link is created by the flag stored in the home address of each track. The result is stored in the storage unit as a defective replacement information table in the storage unit,
When a read command for a home address including a flag indicating whether the track is defective or a home address including a track address is received from the host device, a seek command specifying each track issued from the host device to the file control device Obtain the track address of each track, check whether bad track information can be read from the home address of the direct access storage device based on the track address, determine whether the defective replacement information table is valid, and receive the track address Home to confirm the flag information for the address read instruction, on the basis of the track address and the flag information, file control apparatus, wherein the benzalkonium create a response information in response to the read command of the home address. When the direct access storage device is a fixed length array disk and the file control device emulates a variable length method with the array disk,
The scratch information of the track, which is a response to the read command of the home address including additional information, is fixed to the standard value,
2. The file control apparatus according to claim 1, wherein response information for a read command of a home address including additional information is created from the flaw information, the flag information, and the track address.

Images