JP2008158572A - Data storage device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a data storage device capable of storing data requiring real-time performance such as moving image data at a high speed with a smaller number of disk drives while improving data reliability.
A first operation mode is provided in which the same data is written to a plurality of HDDs during writing and data is read from any of the plurality of HDDs during reading. At the same time, the entire stream for which reading / writing is requested is divided into n so that the number of individual streams does not exceed the allowable number of streams per HDD, and each is distributed to a plurality of HDDs for processing. The operation mode is provided. Based on the number of determination results obtained by the stream number determination means, the storage operation is performed in either the first or second operation mode.
[Selection] Figure 1

Description

  The present invention relates to a data storage device capable of storing data that requires real-time performance such as moving image data at high speed while improving data reliability by duplicating data (or multiplexing of 3 or more). It is.

  In recent years, data storage devices that store moving image data, audio data, and the like in hard disk drives instead of conventional tape media are becoming widespread. Hard disk drives have excellent random accessibility. For this reason, it has a usability that is not available in conventional tape media, such as the ability to instantly play back desired content or to quickly find a free area for recording.

  In addition, in information devices such as personal computers (hereinafter referred to as PCs), many models are commercially available that are equipped with a tuner or codec (CODEC) LSI and have a function of storing moving image data in an internal / external hard disk drive. ing.

  Moving image data and the like are continuous data, and it is not sufficient that the data is simply stored on the storage medium without error, and the recording / reproducing speed on the time axis, that is, the data transfer speed exceeds a predetermined value. Need to be maintained.

  In particular, moving image data tends to have a larger capacity and a higher data transfer rate due to demands for high definition and a large screen, and in addition, it is required to simultaneously handle recording and reproduction of a plurality of moving image data. For this reason, high data transfer performance is desired today for hard disk drives that store data and the like.

  On the other hand, since the hard disk drive is a mechanical component, it is one of components having a relatively short life as a component constituting an electronic device such as a PC. The failure of the parts responsible for memory will not only mean that the device will not function, but will also cause the stored information to be lost. It becomes necessary.

  In order to improve reliability, data with multiple disk drives written to the host device is duplicated and the same data is written to multiple disk drives, and the data is duplicated (or 3 or more multiplexed) Storage devices are known.

  However, it is known that the operation speed of a data storage device in which data is multiplexed in this way is almost the same as or slower in some cases than that of a data storage device using one disk drive. .

  Therefore, a device for improving the operation speed of a data storage device in which data is multiplexed has been proposed. For example, the mirrored disk control device disclosed in Patent Document 1 has a mirroring configuration in which data is simultaneously written to two disk devices at the time of writing. In consideration of the head position of each disk at the time of reading, a read request is issued to one of the disk devices having a closer head position to improve the operation speed at the time of reading.

  Further, Patent Document 2 discloses a multiplexed disk control device in which the responsiveness of the disk device is improved by using the first disk device as a writing subject and the second disk device as a reading subject.

  Furthermore, Patent Document 3 discloses a data access device in which data storage is distributed to a plurality of hard disk devices to increase the speed, and the distributed hard disk devices are configured in a mirroring configuration to ensure reliability by redundancy. Is disclosed.

JP-A-5-165579 JP-A-6-67811 JP-A-7-302173

  However, although the invention disclosed in Patent Document 1 shows an effect of speed improvement for reading, no improvement is made for writing.

  On the other hand, the invention disclosed in Patent Document 2 can improve the operation speed by distributing the write request and the read request to different disks and processing them.

  However, the effect is limited to cases where write requests and read requests are moderately mixed. For example, when only write requests are concentrated, almost no effect can be expected.

  In the invention disclosed in Patent Document 3, the disadvantages of the inventions disclosed in Patent Documents 1 and 2 are eliminated by combining distributed storage and multiplexed storage (duplication of storage). However, since this configuration requires at least four disk drives, it is extremely disadvantageous for reducing the cost and size of the apparatus.

  SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to realize a data storage device that can store data that requires real-time performance, such as moving image data, at a high speed while improving the reliability of the data with a smaller number of disk drives. There is.

According to a first aspect of the data storage device of the present invention, a plurality (n) of storage media, a stream number determination means for determining the number of streams requested to be read / written, and the plurality of storage media at the time of writing The first operation mode for reading data from one of the plurality of storage media at the time of reading and reading the same data, and the entire stream requested to be read / written A second operation mode in which the number of streams is divided into n so as not to exceed the allowable number of streams, and each is distributed to the plurality of storage media for processing, and the determination result obtained by the stream number determination means is Based on this, the storage operation control means for performing the storage operation in either the first operation mode or the second operation mode, and data is multiplexed (n-multiplexed) on the plurality of storage media. When the determination result obtained by the data multiplexing management storage means storing the number of streams and the number-of-streams determination means is less than a predetermined value, based on the storage result of the data multiplexing management storage means And data multiplexing processing means for multiplexing (n-duplicating) and writing data to the plurality of storage media.
According to a second aspect of the data storage device of the present invention, a plurality (n) of storage media, a total data transfer rate determining means for determining a total data transfer rate of the entire stream for which reading / writing is requested, Writes the same data to the plurality of storage media and reads the data from any of the plurality of storage media at the time of reading, and the entire stream that is requested to be read / written as individual data A second operation mode in which the total of transfer rates is divided into n so that the total data transfer rate cannot exceed the permissible data transfer rate per storage medium, and each is distributed to the plurality of storage media for processing. Based on the determination result obtained by the total data transfer rate determination means, the storage operation control for performing the storage operation in either the first operation mode or the second operation mode. A determination result obtained by the means, a data multiplexing management storage means for storing whether or not the data is multiplexed (n multiplexed) and written on the plurality of storage media, and the total data transfer rate determination means And a data multiplexing processing means for multiplexing and writing data to the plurality of storage media based on the storage result of the data multiplexing management storage means when the value is less than or equal to a predetermined value. It is characterized by.

  In the present invention, as long as the number of streams to be handled is small, the same data is duplicated (or n-duplicated) and stored in a plurality of HDDs. However, when the number of streams handled increases and the allowable data transfer rate of the HDD becomes insufficient, the duplex storage (or n-duplex storage) is canceled and a plurality of HDDs are operated individually. Then, different stream processing is allocated to each to increase the allowable data transfer rate of the entire data storage device. Then, the data duplication processing (or n-duplication processing) is executed when the stream handled by the data storage device is a low load of “0 to a few”.

  Therefore, according to the present invention, data reliability can be improved by duplicating (or n-duplicating) data. In addition, according to the present invention, a data storage device capable of storing data requiring real-time performance such as moving image data at high speed can be realized with a configuration using a minimum of two disk drives.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments to which the invention is applied will be described in detail with reference to the accompanying drawings.

(First embodiment)
FIG. 1 is a block diagram of a data storage device according to the first embodiment of the present invention. Reference numeral 10 in FIG. 1 denotes a CPU that controls the entire apparatus. On the bus (system bus) of the CPU 10, a ROM 11, a RAM 12, a NIC 13, an external device connection interface 14, a decoder 15, a disk controller 16, and a graphic / sound controller 17 are connected. These constitute the data storage device 30 as a whole.

  Reference numeral 13 denotes a network controller (NIC) for connecting the data storage device 30 to the LAN. Reference numeral 14 denotes a controller of an external device connection interface (such as USB or IEEE 1394) that connects the data storage device 30 to 31 video devices (such as a digital video camera).

  Reference numeral 16 denotes a disk controller that controls the operations of the 18 and 19 hard disk drives (HDD). In the data storage device 30, the moving image / audio data transferred from the network camera or the image distribution server via the LAN or the moving image / audio data output from the video device 31 connected via the external device connection interface 14 is stored in the HDD 18, 19 That is, video / audio data is recorded / stored in the HDDs 18 and 19.

  The moving image / audio data stored in the HDDs 18 and 19 is usually compressed (encoded) in a known format such as MPEG or Motion JPEG.

  Reference numeral 15 denotes a decoder circuit for decompressing the compressed data. At the time of data reproduction, the moving image / audio data expanded by the decoder circuit 15 is output to the monitor device 32 connected to the outside via the graphic / sound controller 17.

  Reference numeral 11 denotes a ROM which stores an operation program for the CPU 10 and various table data.

  A RAM 12 is used as an operation memory of the CPU 10 as a RAM, and is also used as a buffer memory for data transfer performed between each block on the system bus.

  Reference numeral 20 denotes an operation panel connected to the input / output port of the CPU 10, which is composed of a button switch, an LCD, and the like.

  The user can issue a desired operation instruction to the data storage device 30 by operating the operation panel 20.

  With the above configuration, the data storage device 30 stores the moving image / audio data transferred via the LAN and the moving image / audio data output from the video device 31 in the HDDs 18, 19, and stores the stored data in the monitor device 32. Functions as a recording / playback device for output.

  Further, the data storage device 30 also has a function as a network server, and can distribute the moving image / audio data stored in the HDDs 18 and 19 to an information device such as a PC connected on the LAN.

  Next, with reference to FIG. 2, a logical configuration (area division and the like) of the HDDs 18 and 19 will be described.

  FIG. 2 is a diagram showing a format state of the HDDs 18 and 19. The storage areas of the HDDs 18 and 19 are roughly divided into two, and a system area is allocated on the outer peripheral side on the storage medium, and a data area is allocated on the inner peripheral side. The FAT entry in the system area is an allocation table of clusters constituting a file, and stores chain information (next cluster number) of a plurality of clusters constituting one file.

  In addition to chain information, the FAT value defines an unused cluster, a defective cluster, and an end of file (EOF).

  The directory entry stores a file name (including path information) and the first cluster number of the file.

  The storage data can be handled in units of “files” by the FAT and the directory entry. The redundant management area will be described later.

The data area is divided into 0 to n clusters in order from the outer periphery side. When the size of the data area is 72 Gbytes (72 Gbytes × 1024 × 1024 = 75,497,472 Kbytes) and the size of the cluster is 128 Kbytes, 7549472/128 = 589824
Thus, there are 589 and 824 clusters in the entire data area.

  As is apparent from FIG. 2, the HDD 18 and the HDD 19 have exactly the same configuration, and the data storage device in this embodiment duplicates (duplicates) the same data in the HDDs 18 and 19 in both the system area and the data area. Remember.

  The duplication management table in the system area is for managing whether or not individual clusters are duplicated and recorded in the HDDs 18 and 19.

The duplex management table has a size corresponding to the number of clusters × 2 bits (589824 × 2 bits = 1179648 bits = 144 Kbytes in this embodiment), and stores the following status information for each cluster.
00b: Unused cluster 01b: Data stored only in HDD 18 Cluster 10b: Data stored only in HDD 19 Cluster 11b: Cluster in which data is stored redundantly in HDDs 18 and 19

  Next, a detailed configuration of the disk controller 16 will be described with reference to FIG. Reference numeral 40 in FIG. 3 denotes an MPU that controls the entire disk controller 16.

  In the MPU 40, a CPU, a ROM, a RAM, an I / O port, and the like are integrated on one chip.

  On the bus (local bus) of the MPU 40, a bus bridge 41, a status register 42, a command queuing memory 43, a data buffer memory 44, a duplex management table memory 45, and ATA interfaces 46 and 47, which will be described later, are connected. These constitute the disk controller 16 as a whole.

  The disk controller 16 is connected to the system bus described with reference to FIG.

  As a result, data can be exchanged between each block connected on the system bus and each block connected on the local bus.

  Further, inter-processor communication can be performed between the CPU 10 and the MPU 40 via the bus bridge 41.

  The status register 42 is a register circuit for storing the operation status of the disk controller 16 (progress status of various operations, etc.).

  The set value of the status register 42 is set every moment by the MPU 40 according to the operation status. The status register 42 can also be referred to from the CPU 10 in FIG. 1, whereby the CPU 10 can know completion of various operations instructed to the disk controller 16.

  The status register 42 is provided with a read / write accessible area from both the CPU 10 and the MPU 40, and this area is used for inter-processor communication between the CPU 10 and the MPU 40.

  The command queuing memory 43 is a memory for queuing transaction commands issued to the disk controller 16.

  FIG. 4 shows a transaction command format. The transaction command is composed of a header part (2 bytes), a DMA start address (4 bytes), a cluster number, or an LBA (2 bytes) for 10 bytes.

  The first 1 bit of the header part is a bit for distinguishing between a transaction for reading and writing a FAT and a directory entry and a transaction for reading and writing a stream of moving image / audio data. The next 1 bit is a bit for distinguishing between reading and writing.

  The next 6 bits indicate the stream number. Data for one cluster (128 Kbytes) can be transferred in one transaction, but normal video / audio data is much larger than this, so multiple transactions are spent for transferring operation / audio data. It is.

  Therefore, the stream number is used to distinguish which stream data transfer belongs to which transaction. The stream number does not need to be an absolutely unique number, and it is only necessary to distinguish a plurality of streams in which transactions are executed simultaneously (in parallel in a time division manner). It is sufficient to use 63) repeatedly.

  Transaction commands can be issued one after another while the command queuing memory 43 is empty.

  The disk controller 16 reads the queued transaction commands from the command queuing memory 43 and sequentially executes corresponding processing operations.

  Therefore, there is a time lag between issuing the command and executing the processing operation, and there is no way to know when the processing operation of the issued command is completed.

  The last 8 bits are for dealing with this, and are a field indicating a command number. When the MPU 10 finishes executing the processing operation of the command to which a certain number is attached, the MPU 10 sets the value “1” only to the corresponding numbered register among the command number registers existing in the status register 42. The CPU 10 can know the completion of the processing operation of the issued command by polling the command number register.

  The command number does not have to be an absolutely unique number. It is only necessary that there is no duplication among a plurality of commands queued in the command queuing memory 43. Therefore, 8 bits (0 to 255) are repeated. Use is sufficient.

  Data for the transaction is stored in the RAM 12. That is, at the time of writing, a transaction command is issued after data to be written to the HDDs 18 and 19 is stored in the RAM 12. At the time of reading, after issuing a transaction command, the processing operation is executed and the data read from the HDDs 18 and 19 is stored in the RAM 12.

  Data transfer is performed by DMA transfer between “RAM 12 and disk controller 16”, and the DMA transfer start address at this time is designated by 4 bytes following the header portion.

  The last 4 bytes are a field for designating the number of the cluster to be read / written. Since reading and writing of FAT and directory entries are performed in units of sectors, not in clusters, in transactions for FAT and directory entries, LBA (logical address) is designated for this part instead of cluster numbers.

  Returning to FIG. 3, the data buffer memory 44 is a buffer memory that temporarily stores read data and write data from the HDDs 18 and 19.

  The DMA transfer between “RAM 12 and disk controller 16” is actually performed between “RAM 12 and data buffer memory 44”. The ATA interfaces 46 and 47 are interface circuits for physically connecting the HDDs 18 and 19.

  The duplex management table memory 45 is a memory for caching the entire contents of the duplex management table described with reference to FIG. The contents stored in the duplex management table are copied to the duplex management table memory 45 when the power is turned on (or after the HDD is initialized). Thereafter, when updating the duplex management table, the duplex management table memory 45 is operated instead of directly rewriting the duplex management table on the HDDs 18 and 19.

  When the operation of the apparatus is finished (that is, when the power is turned off), the storage contents of the duplex management table memory 45 are copied to the duplex management table on the HDDs 18 and 19.

  For convenience of explanation, in FIG. 4, the command queuing memory 43, the data buffer memory 44, and the duplex management table memory 45 are independent blocks. However, it is of course possible to provide only a single system memory and share the memory for each application.

  The data storage device in the present embodiment basically stores the same data in the HDDs 18 and 19 in a duplicated manner (mirroring). However, when the number of video / audio data streams to be handled increases and the allowable data transfer rate of the HDD becomes insufficient, the duplex storage (mirroring storage) is canceled and the HDDs 18 and 19 are operated individually. Then, a different stream process is assigned to each, and the allowable data transfer rate of the entire data storage device is doubled.

  Then, when the stream handled by the data storage device is a low load of “0 to a small number”, data duplication processing (copy processing between HDDs 18 and 19) is executed.

  The operation of the data storage device 30 will be described in detail using the flowcharts of FIGS. FIG. 5 is an operation flow when the HDDs 18 and 19 are initialized. This operation flow is executed by the program operation of the CPU 10.

  After the physical formatting is completed, first, in step 100, the sizes of the system area and the data area are calculated from the total capacity of the HDDs 18 and 19 and the size of the cluster (128 Kbytes). When the capacities of the HDDs 18 and 19 are different, the calculation is performed based on the smaller one.

  Next, in step 101, a duplex management table is created in the system area. At this time, “unused” is set for all clusters.

  Next, in step 102, a FAT (file allocation table) is created in the system area. At this time, the FAT value for each cluster is set as “unused cluster”.

  Next, in step 103, a root directory is created in the directory entry.

  6 to 9 are operation flows when data is read from and written to the HDDs 18 and 19. The operation flow shown in FIGS. 6 and 7 is executed by the program operation of the CPU 10.

  When a situation in which the data storage device 30 should access the HDDs 18 and 19 occurs, first, in steps 200 and 201, the disk controller 16 is inquired whether a transaction request is accepted, and the acceptance is permitted. Wait until. The inquiry is executed by communication between the CPU 10 and the MPU 40 processor.

  Next, in step 202, it is determined whether writing or reading. In the case of writing, the FAT and directory entry read commands are issued multiple times in steps 203 to 206 to read the FAT and directory entries. Next, a free cluster is searched from the FAT and directory entry read in step 207 to determine a write cluster.

  In steps 208 to 211, the FAT and directory entry write commands are issued as many times as necessary to update the contents of the FAT and directory entries.

  At this time, the file name is used when the data input is via the LAN and the file name can be known, and the name is used when the video equipment 31 designates the file name.

  When the file name is not designated from the outside, the file name is appropriately combined with alphanumeric characters.

  Further, when the user designates a file name or path information using the operation panel 20 or the like, it follows that.

  When the update writing of the FAT and the directory entry is completed, the disk controller 16 is inquired whether or not a transaction request is accepted in steps 212 and 213, and waits until the acceptance is permitted. If acceptance is permitted, a write command for one cluster is issued in step 214.

  The above operation is repeatedly executed from step 212 for all clusters to be written (step 215).

  When the command issuance is completed, the process of the command issued in step 216 is waited for to be completed, and a series of processes for writing is terminated.

  Returning to step 202, if it is read, the FAT and directory entry read commands are issued a plurality of times in steps 217 to 219 of FIG. 7 to read the FAT and directory entries.

  Next, the cluster in which the data of the file to be read is stored from the FAT and directory entry read in step 220 is searched and determined.

  Next, in steps 221, 222, the disk controller 16 is inquired whether a transaction request is accepted, and waits until the acceptance is permitted.

  If acceptance is permitted, a read command for one cluster is issued in step 223. The above operation is repeatedly executed from step 221 for all clusters to be read (step 224).

  When the command issuance is completed, the processing for the command issued in step 225 is waited for completion, and a series of processing for reading is terminated.

  As for the tasks from step 200 to step 226, when a plurality of streams are handled at the same time, a plurality of tasks are activated simultaneously (as time division parallel processing) within an allowable range (up to six in this embodiment). Executed. However, in the case of a write transaction, only the processing related to the FAT and directory entry operations (that is, the processing in steps 203 to 211) must be executed exclusively to prevent double updating of the FAT and directory entries.

  The operation flows shown in FIGS. 8 and 9 are executed by the program operation of the MPU 40. First, with reference to FIG. 8, the operation process for accepting or not accepting a transaction request will be described.

  In step 300, it is determined whether or not there is an inquiry about whether or not the transaction request can be accepted from the CPU 10.

  If there is no inquiry, the process ends without doing anything. If there is an inquiry, it is determined in step 301 whether or not the command queuing memory 43 is in the “full” state.

  If it is in the “full” state, the transaction request is rejected in step 302. If it is not in the “full” state, the transaction request is permitted in step 303.

Next, read / write operation processing to the HDDs 18 and 19 will be described with reference to FIG.
In step 400, one command is acquired from the command queuing memory 43, and in step 401, it is determined whether the command is a FAT and directory entry transaction command or a stream transaction command.

  If the transaction command is a FAT / directory entry transaction, writing / reading of the FAT / directory entry is executed in step 402. Specifically, an operation of writing data to the instructed LBA or reading data from the instructed LBA is executed.

  The FAT and directory entry are always written in duplicate in the HDDs 18 and 19.

  If it is a stream transaction command, the number of streams currently being processed is calculated in steps 403 and 404. If the number of streams being processed is three or less, it is determined in step 405 whether it is a write transaction command or a read transaction command.

  If it is a write transaction command, in step 406, the same data is duplicated and written to the designated cluster of both HDDs 18 and 19. In step 407, the corresponding cluster portion of the duplex management table memory 45 is updated to “11b”. In step 408, the register is updated. If it is a read transaction command, the cluster data designated in step 409 is read (in this case, data can be read from either arbitrary drive).

  Return to step 404. If the number of streams being processed exceeds three, it is determined in step 410 whether it is a write transaction command, a read transaction command, or a read transaction command, and whether the data is duplicated data. To do. Whether or not it is duplicated can be easily known by searching the duplicated management table memory 45.

  If it is a read transaction command for data that has not been duplicated, in step 411, the data of the designated cluster is read from the drive in which the data exists. It can be immediately known from which value the data exists in the duplex management table memory 45 (whether it is “10b” or “01b”). Then, returning to step 410, if it is a duplicated data read transaction command, in step 412, the cluster data designated by either drive is read.

  If the transaction command is a write transaction command, the cluster data instructed to either drive is written in step 413, and the corresponding cluster portion of the duplex management table memory 45 is set to "10b" or "01b" in step 414. Update.

  When the processing in steps 408, 409, 411, 412, and 414 is completed, the command number register is updated in step 408 to notify the CPU 10 that the processing operation of the command has been completed. The above operation is repeated from step 400.

  Now consider which drive should be selected in steps 412, 413. Here, it is desired to divide the six streams into three, and let the HDDs 18 and 19 be in charge of each. Accordingly, writing is given priority to a drive storing non-duplicated data, and reading is given priority to a drive with a small number of streams, and a drive to be used is selected.

  Even if such a device is used, the six streams cannot always be processed completely. For example, there is naturally a case where reading of a stream that needs to read non-duplicated data cannot be handled beyond three.

Therefore, the specification of the data storage device 30 is that the number of streams that can be handled at the same time is 6 writing or less (recording) reading (reproducing) 6 or less writing (recording) 3 or less, and reading (reproducing) 3 or less. .

  FIG. 10 is an operation flow of data duplication processing. The operation flow shown in FIG. 10 is a process that is executed by the program operation of the MPU 40 and is always operated in the background.

  First, in steps 500 and 501, it is determined whether or not there is a margin in the performance of the HDDs 18 and 19.

  Specifically, the number of streams currently being processed is calculated in step 500, and it is determined in step 501 whether the number is 2 or less. If there is no margin in performance, the process ends.

  If there is a margin in performance, the duplex management table memory is searched in steps 502 and 503 to determine whether there is a cluster in which data is duplicated and not stored. If there is no non-redundant cluster, the process is terminated as it is.

  If there is a non-redundant cluster, in step 504, the cluster data is copied to the cluster with the same number on the other HDD. If the value of the duplex management table memory 45 is “01b”, copying is performed from the HDD 18 to the HDD 19, and if “10b”, copying is performed from the HDD 19 to the HDD 18.

  In step 505, the corresponding cluster portion of the duplex management table memory 45 is updated to “11b”. The above processing is repeatedly executed from step 500.

  Although the operation flow for deleting a stream file is not particularly shown, the FAT value of the cluster storing the data of the stream file may be updated to “unused cluster”. Further, the corresponding cluster portion of the duplex management table memory 45 may be updated to “00b”. That is, it is only necessary to update the FAT and duplex management table, not rewrite the stored data of each cluster itself.

As described above, in this embodiment, a) when the number of streams handled simultaneously is equal to or less than a predetermined value (three or less), the same data is duplicated (mirrored) and stored in the HDDs 18 and 19.
b) When the number of simultaneously handled streams exceeds a predetermined value (more than 3), the duplex storage is canceled and different streams are stored (recorded / reproduced) in the HDDs 18 and 19.
c) When the number of streams handled at the same time is equal to or less than a predetermined value (two or less), duplex processing of data that is not duplexed and stored is performed in the background.

  As described above, there are fewer data storage devices (two in this embodiment) than the conventional technology that store data that requires real-time performance such as moving image data at high speed while improving data reliability. It can be realized with a disk drive.

  In the present embodiment, it is assumed that the effective data transfer rate of the HDDs 18 and 19 is about 150 Mbps, the data transfer rate of each stream is about 50 Mbps as a moving image of SXGA size, Motion JPEG (30 fps, 1/20 compression). Based on this premise, the number of streams that can be handled simultaneously by one HDD has been described as three (six for the entire data storage device). However, this is merely an example, and it is needless to say that it can be appropriately changed depending on the characteristics of the HDD to be used, the data encoding format of the stream to be handled, the image size, the frame rate, and the like.

  6 and 7 show an operation flow for accessing the FAT area and the directory entry area on the HDD every time the stream is read and written. However, the contents of the FAT and directory entry may be cached in the cache area on the RAM 12, and thereafter, the FAT and directory entry may be accessed in the cache area on the RAM 12.

  As a result, the access operation to the FAT area and the directory entry area on the HDD is omitted, and as a result, the stream storage operation can be executed at a higher speed.

  Note that steps S403 and S404 in FIG. 9 and steps S500 and S501 in FIG. 10 are processes that are one example of processing of the stream number determination means of the present invention. Steps S406 and S409 are processing as an example of the first operation mode of the present invention. Steps S411 to 413 are processing as an example of the second operation mode of the present invention. The control process between the process from step S404 to S405 and the subsequent process and the process from step S404 to S410 and subsequent process is a process that is one control example of the storage operation control means of the present invention. The duplex management table memory 45 has a configuration as an application example of the data multiplexing management storage means of the present invention. Step S504 in FIG. 10 is a process as an example of the process of the data multiplexing processing means of the present invention.

(Second Embodiment)
In the above embodiment, it is assumed that a plurality of streams having the same data transfer rate are stored. However, this type of data storage device can handle multiple streams of various data transfer rates (different compression rates, different encoding formats, different image sizes, different frame rates, etc.) simultaneously. Conceivable.

  In such a case, it is not convenient to focus only on the number of streams handled. This embodiment has coped with this point.

  FIG. 11 shows a transaction command format in this embodiment. As is clear from FIG. 11, the header part is extended to 3 bytes. The third byte represents a data transfer rate coefficient.

  That is, the coefficient of the data transfer rate required by the stream is set here in units of 1 Mbps. Thereby, the MPU 40 can know the data transfer rate required for each stream in units of “Mbps”.

  The CPU 10 may issue transaction commands for a plurality of streams within an allowable range, but in this embodiment, the total data transfer rate required by each stream is limited to a range that does not exceed 250 Mbps.

  FIG. 12 shows an operation flow of read / write operation processing to the HDDs 18 and 19 in this embodiment. This operation flow corresponds to the operation flow described in FIG. 9 in the first embodiment. Steps 400 to 402 and 405 to 414 in FIG. 12 are the same as those in FIG.

  In step 403 ′, the total data transfer rate of the stream currently being processed is calculated. If the total data transfer rate is 100 Mbps or less, the processing from step 405 is executed. If the total data transfer rate exceeds 100 Mbps, the processing from step 409 is executed.

  In steps 412, 413, the drive selection is distributed so that the total data transfer rate of the streams handled by the individual drives is approximately ½ of the total data transfer rate.

  With this process, the total data transfer rate for each HDD can be reduced to approximately 100 to 150 Mbps or less.

  By the way, when the number of streams handled simultaneously in the HDD increases, the seek time of the head and the rotation waiting time of the disk platter increase, and the effective data transfer rate decreases.

  Therefore, a good result can be obtained by limiting the number of streams rather than simply summing up the data transfer rates of individual streams. Alternatively, a good result can be obtained by multiplying the total data transfer rate by the correction coefficient in accordance with the number of streams and calculating a larger data transfer rate.

  FIG. 13 shows an operation flow of data duplication processing in this embodiment. This operation flow corresponds to the operation flow described in FIG. 10 in the first embodiment. Steps 502 to 505 in the figure are the same as those in FIG.

  In step 500 ′, the total data transfer rate of the stream currently being processed is calculated. If it is determined in step 501 ′ that the total data transfer rate exceeds 50 Mbps and there is no performance margin, the processing ends. If it is 50 Mbps or less and there is margin in performance, processing from step 502 is executed. .

  As described above, in the present embodiment, based on the total data transfer rate of simultaneously handled streams, the same data is stored in the HDDs 18 and 19 in a duplicated manner (mirroring), or the redundant storage is canceled and the HDDs 18 and 19 are deleted. To decide whether to store (record / play) different streams.

  Therefore, the present invention can be satisfactorily implemented even in a data storage device that handles a plurality of streams of various data transfer rates simultaneously.

  Note that steps S403 ′ and S404 ′ in FIG. 12, and steps S500 ′ and S501 ′ in FIG. Steps S406 and S409 are processing as an example of the first operation mode of the present invention. Steps S411 to 413 are processing as an example of the second operation mode of the present invention. The control process between the process from step S404 to S405 and the subsequent process and the process from step S404 to S410 and subsequent process is a process that is one control example of the storage operation control means of the present invention. The duplex management table memory 45 has a configuration as an application example of the data multiplexing management storage means of the present invention. Step S504 in FIG. 13 is a process as an example of the process of the data multiplexing processing means of the present invention.

  Another object of the present invention is to supply a storage medium storing software program codes for realizing the functions of the above-described embodiments to a system or apparatus, and a computer such as the system reads and executes the program codes from the storage medium. Is also achieved.

  In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing the program code constitute the present invention.

  As a storage medium for supplying the program code, for example, a flexible disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

  In addition, the case where the functions of the above-described embodiment are realized by performing part or all of the actual processing by an OS or the like running on the computer based on the instruction of the program code read by the computer. It is.

  Further, after the program code read from the storage medium is written in a memory provided in a function expansion unit connected to the computer, the CPU or the like performs actual processing based on the instruction of the program code, and the above-described processing is performed. The case where the functions of the embodiment are realized is also included.

It is a block diagram which shows the structure of the data storage device which concerns on embodiment of this invention. It is a figure for demonstrating the logical structure of HDD. It is a figure which shows the detailed structure of a disk controller. It is a figure for demonstrating the format of a transaction command. It is a flowchart which shows operation | movement of a data storage device. It is a flowchart which shows operation | movement of a data storage device. It is a flowchart which shows operation | movement of a data storage device. It is a flowchart which shows operation | movement of a data storage device. It is a flowchart which shows operation | movement of a data storage device. It is a flowchart which shows operation | movement of a data storage device. It is a figure for demonstrating the command format in the 2nd Embodiment of this invention. It is a flowchart which shows the operation | movement in the 2nd Embodiment of this invention. It is a flowchart which shows the operation | movement in the 2nd Embodiment of this invention.

Explanation of symbols

10 CPU
11 ROM
12 RAM
13 NIC (network controller)
14 External Device Connection Interface 15 Decoder Circuit 16 Disk Controller 17 Graphic / Sound Controller 18, 19 HDD (Hard Disk Drive)
20 Operation Panel 30 Data Storage Device 31 Video Equipment 32 Monitor Device

Claims (2)

A plurality (n) of storage media;
Stream number determination means for determining the number of streams for which reading and writing are requested;
When writing, the same data is written to the plurality of storage media, and when reading, the first operation mode for reading data from any of the plurality of storage media and the entire stream for which reading / writing is requested The number of streams is divided into n so that the number of streams does not exceed the allowable number of streams per storage medium, and each is distributed to the plurality of storage media for processing. Based on the determination result obtained by the determination means, storage operation control means for performing a storage operation in either the first operation mode or the second operation mode;
Data multiplexing management storage means for storing whether or not data is multiplexed (n multiplexed) and written to the plurality of storage media;
When the determination result obtained by the number-of-streams determination unit is equal to or less than a predetermined value, data is multiplexed (n-multiplexed) on the plurality of storage media based on the storage result of the data multiplexing management storage unit. And a data multiplexing processing means for writing the data. A plurality (n) of storage media;
Total data transfer rate determination means for determining the total data transfer rate of the entire stream for which reading and writing are requested;
When writing, the same data is written to the plurality of storage media, and when reading, the first operation mode for reading data from any of the plurality of storage media and the entire stream for which reading / writing is requested A second operation mode in which the total of the data transfer rates is divided into n so that the total data transfer rate cannot exceed the allowable data transfer rate per storage medium, and each is distributed to the plurality of storage media for processing. Storage operation control means for performing a storage operation in either the first operation mode or the second operation mode based on the determination result obtained by the total data transfer rate determination means;
Data multiplexing management storage means for storing whether data is multiplexed (n-multiplexed) and written to the plurality of storage media;
When the determination result obtained by the total data transfer rate determination means is less than or equal to a predetermined value, data is multiplexed (n-duplexed) on the plurality of storage media based on the storage result of the data multiplexing management storage means And a data multiplexing processing means for writing.

Images