JP4365509B2 - Data management apparatus, data management method, and recording medium recording data management procedure - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to management of data recorded on a recording medium, and more particularly to a data management apparatus and data management method for recording / reproducing temporally continuous data such as video data.
[0002]
[Prior art]
In a system (for example, a video server system or a non-linear editing system) that digitizes temporally continuous analog signals such as video and audio and records or reproduces the stream data on a recording medium such as a hard disk, It is required to record and reproduce data seamlessly, and various data management apparatuses have been proposed from this viewpoint.
[0003]
FIG. 29 is a block diagram of a data management apparatus disclosed in Japanese Patent Laid-Open No. 10-275058. A computer 8 constituting the data management apparatus includes a CPU 9, an I / O adapter 10, a memory 5, and a host bus. And a SCSI adapter 2 (k) as a host adapter connected to the PCI bus 1 (k: a natural number, 1 to 4 are used in this specification), and the SCSI adapter 2 ( k) A hard disk 4 (k) as a recording medium connected to each SCSI bus 3 (k) and a control unit 11 for controlling access to these hard disks 4 (k).
[0004]
“SCSI adapter 2 (1) and hard disk 4 (1)” to “SCSI adapter 2 (4) and hard disk 4 (4)”, that is, a group of SCSI adapters and hard disks having the same numbers in parentheses constitute a recording unit. These four recording units constitute one recording unit group.
[0005]
Each SCSI adapter 2 (k) enables bus master DMA transfer. That is, the controller 6 built in the SCSI adapter 2 (k) acquires the occupation right of the PCI bus 1, and arbitrary data on the memory 5 is built into the SCSI adapter 2 (k) via the PCI bus 1. It is possible to transfer to the buffer 7.
[0006]
Although the PCI bus is used as the host bus, other types of data transfer buses may be used as long as the transfer speed condition is satisfied. Further, although the SCSI adapter is used as the host adapter, other types of host adapters may be used as long as the bus master DMA transfer is possible.
[0007]
The I / O adapter 10 inputs high bit rate stream data in units of fixed size blocks based on an instruction from the control unit 11 and transfers the data to a predetermined area of the memory 5 or data already recorded in the predetermined area of the memory 5. Can be output as stream data.
[0008]
The control unit 11 includes a recording control unit 111, a file management unit 112, and a reproduction control unit 113, which will be described below, composed of a CPU and software that operates under the CPU. The file management information managed by the recording control unit 111 is stored in a recording means (for example, a hard disk) different from the hard disk 4 (k).
[0009]
In other words, the recording control unit 111 controls the CPU 9 so that recording of stream data continuously input to the memory 5 in units of fixed size blocks is always in time (that is, while maintaining the state where the memory 5 does not overflow). ) Control to sort and record each hard disk 4. The file management unit 112 manages the arrangement of data recorded on each hard disk 4. The reproduction control unit 113 performs control to sequentially read the data allocated and recorded on each hard disk 4 to reconstruct the stream data and output it to the I / O adapter 10.
[0010]
Hereinafter, the procedure of the recording process performed by the recording control unit 111 will be described in more detail.
[0011]
The input means (not shown) included in the recording control unit 111 divides the stream data input from the I / O adapter 10 into fixed size blocks (1), (2), (3), and the recording unit group. The fixed size blocks {circle around (1)} {2} {3}... Divided as described above so that a plurality of hard disks 4 (k) constituting one logical recording space correspond to each hard disk 4 (k). Distribute.
[0012]
At the same time, the writing means (not shown) included in the recording control unit 111 transfers the fixed size blocks (1), (2), (3), etc., distributed as described above to each hard disk 4 (k). The SCSI adapter 2 (k) is instructed, and the controller 6 (k) built in the SCSI adapter 2 (k) that has received this instruction stores the fixed size block (1) (2) (3). After storing in k), record in hard disk 4 (k). That is, as shown in FIG. 30, the fixed size block (1) is recorded on the hard disk 4 (1) by the controller 6 (1), and the fixed size block (2) is recorded on the hard disk 4 (2) by the controller 6 (2). The fixed size blocks (1), (2), (3),... Are serially recorded so that the fixed size block (3) is recorded on the hard disk 4 (3) by the controller 6 (3).
[0013]
According to the above procedure, the data transfer procedure from the stream data source to one buffer 7 (k) constituting the recording unit and the recording procedure from the buffer 7 (k) to the hard disk 4 (k) are alternated. However, since the above two procedures are apparently performed in units of a plurality of recording units, data can be recorded at high speed and seamlessly.
[0014]
Note that various information related to stream data recorded on the hard disk 4 (k) as described above is managed by the file management unit 112 as file management information (the contents of the file management information are left to the description of the present invention). .
[0015]
[Problems to be solved by the invention]
  When a file is normally managed by a computer or the like, a file system (for example, FAT (file allocation table)) supported by the operating system adopted by the computer is used. That restores data (hereinafter referred to as "recovery function")NaThis is executed using the above file system.
[0016]
  However, without using a file system supported by the operating system, in a data management device having its own file system as described above, it is natural that the data recovery function is destroyed.AlsoYou need to design your own.
[0017]
In the above file system such as FAT, the data to be recorded and the file management information are stored in the same recording medium, and the management information and the data to be recorded are recorded in different areas of the same storage medium. Therefore, the seek of the head of the hard disk frequently occurs. However, in order to record or reproduce stream data such as video data, it is required to be fast and seamless. Therefore, it is necessary to reduce the seek of the head on the hard disk as much as possible.
[0018]
  Considering this point, it is necessary to develop a unique file system that stores file management information in a storage medium different from the data to be recorded, and furthermore, a data recovery function corresponding to the file system.WithA file system needs to be developed.
[0019]
That is, when the file management information stored in another storage medium is destroyed for some reason, the stream data itself cannot be recorded or reproduced as a system even though it exists in the hard disk 4. . In this case, it is necessary to repair the file management information stored in the other storage medium. Also, if a recording interruption occurs due to some inconvenience during recording, various information relating to the stream data recorded until this interruption is not normally registered in the other storage medium as file management information. At this time, in order to prevent the stream data from becoming invalid, it is necessary to generate file management information on the other storage medium.
[0020]
If there is a defective sector during data recording, the head of the hard disk 4 performs retry (the number of times varies depending on the system), and if writing is still impossible, the recording process to the alternative sector is performed. Here, since the alternative sector exists at a position distant from the original sector, a head seek occurs. Therefore, when the number of bad sectors increases, the seek to the alternative sector increases, thereby increasing the access performance of the hard disk 4. As a result, there is a problem that an error (overflow error / underflow error) occurs when recording / reproducing stream data. This problem is a phenomenon in which the recording command / reproducing command itself to the hard disk 4 does not cause an error, and the time required for recording / reproducing becomes long, so that countermeasures are very difficult.
[0021]
  The present invention has been proposed on the basis of the above-mentioned conventional circumstances, and can record and reproduce data at high speed and seamlessly and can also be restored.TheIt is an object of the present invention to provide a data management apparatus, a data management method therefor, and a recording medium on which a procedure for realizing the method is recorded.
[0022]
[Means for Solving the Problems]
  The stream data management device of the present invention
  Divide stream data into fixed size data blocksFixed sizeDivided search information to identify data blocksFixed sizeGenerated for each of the data blocks,The fixed-size data block and the search information corresponding to the fixed-size data block are associated with the fixed-size data block so that each of the generated search information is associated with the corresponding fixed-size data block.First recording mediumInA recording control unit for recording;
  A file management unit for managing stream data recorded on the first recording medium based on file management information for specifying the stream data;
  When a recovery request for the file management information is issued,Corresponding to each fixed-size data block obtained by dividing stream data recorded on the first recording mediumA recovery unit that recovers the file management information based on the search information,
  The first recording medium includes a plurality of sectors for storing a fixed amount of data,
  The recording control unit is connected to the first recording medium.Each fixed-size data block obtained by dividing the recorded stream data isSectorN (N is an integer of 2 or more)And aboveFixed sizeData blockOf the N sectors included in each, the remainder of the last sectorAnd aboveFixed size dataRecord search information corresponding to blockAnd
  The fixed size of the data block is larger than when (N-1) sectors are used, and smaller than when N sectors are used.It is characterized by that.
[0023]
  Also,The stream data management apparatus of the present invention,
  UpThe search information is at least one of serial number, date information, authentication information, and creator information.
[0029]
DETAILED DESCRIPTION OF THE INVENTION
  Embodiments of the present invention will be described below in detail with reference to the drawings.
As shown in FIG. 1, after dividing the stream data into fixed size blocks, the recording control unit 111 that records the fixed size blocks on the first recording medium 4, and the stream data recorded on the first recording medium The data management apparatus includes a file management unit 112 that manages file data based on file management information that identifies the stream data.
Here, the recording control unit 111 generates search information related to the fixed size block, and then records the search information added to the fixed size block. When the recovery unit 114 requests to recover the file management information, The file management information is restored based on the search information.
If the search information is recorded in the remainder area of the last sector, it is not necessary to prepare a special sector, and the recovery process can be performed with a small capacity. However, the search information may be recorded not in the remaining area but in the sector next to the last sector.
The recording control unit 111 divides the file management information into predetermined size units to generate a plurality of divided storage information, and then adds the divided storage information to the fixed size block for recording. When there is a request for restoring the file management information, the file management information is restored based on the divided storage information.
In the above, it is preferable to record the file management information on a recording medium different from the first recording medium in order to maintain high speed recording and reproduction. The second recording medium is a hard disk, A flash memory or the like can be selected. Here, the file management information stored in the second storage medium is divided into predetermined size units to generate a plurality of pieces of divided storage information, and then the divided storage information is added to the fixed size block for the first recording. It is recorded on the medium.
Further, the recording control unit 111 generates access history information related to access processing of a fixed size block, for example, a writing time history, and then adds the access history information to the fixed size block and records it, as shown in FIG. The area detector 115 detects an area on the recording medium 4 where the access history information is equal to or greater than a predetermined value as a defective area. Further, the reproduction control unit 113 generates history information related to the reproduction time of the fixed size block, and then adds and records history information related to the reproduction time (reading time) to the fixed size block, and records history information about the reproduction time.
An area on the recording medium 4 in which is equal to or greater than a predetermined value is detected as a defective area.
Instead of the write time history and the read time history, the number of bad sectors in the fixed size block may be used. Also, instead of deciding whether a fixed-size block is good or bad at a specific threshold level as described above, multiple levels are generated according to the value indicated by the access history in the bad sector detection procedure. Alternatively, the level may be selected according to the application to be used.
(First embodiment)
  FIG. 1 is a schematic functional block diagram of a computer to which the present invention is applied. The configuration of the computer will be described below only with respect to differences from the prior art.
[Recording Format] FIG. 3 is a diagram showing a recording format in the hard disk 4 as the first storage means. Various information that produces the effect of the present invention is placed between the fixed size block and the next fixed size block. Additional recorded (stored).
[0030]
That is, the stream information described later is stored in the head sector (latest stream information storage sector 200) of each hard disk 4 (k) every time stream data is recorded. Here, video data representing a series of continuous scenes is referred to as stream data, and one file is composed of at least one stream data.
[0031]
In addition, search information, which will be described later, is stored in the remaining area of the last sector (search information storage sector 201) in which fixed-size blocks are recorded. If this search information is larger than the remainder area of the last sector, the sector next to the last sector may be used as the search information storage sector 201.
[0032]
Further, divided storage information to be described later is stored in a sector (divided storage information storage sector 202) next to the search information storage sector 201, and a sector (write time history information storage sector 203 next to the divided storage information storage sector 202). ) Stores write time history information described later, and further stores read time history information described later in the next sector (read time history information storage sector 204).
[0033]
Note that the information stored in the write time history information storage sector 203 and the read time history information storage sector 204 shown in FIG. 3 is information related to the time history of writing and reading fixed-size blocks.
[Recording procedure]
First, a procedure for recording stream data on the hard disk 4 will be described.
[0034]
When it is necessary to record the stream data on the hard disk 4 such as when a user who has seen a monitor displaying an input video to the I / O adapter 10 gives a recording instruction using a mouse or the like, the recording is performed under the control of the CPU 9. The input means and the writing means included in the control unit 111 wake up.
[0035]
That is, the recording control unit 111 includes an input unit 17, a writing unit 18, and a recording buffer manager 16 as shown in FIG. 4, and this recording buffer manager 16 stores a stream data in units of fixed size blocks. 14 and a recording management queue 15 (k) (corresponding to each hard disk 4 (k)) that holds the address of an area where fixed-size block data is stored in the recording buffer 14. The recording buffer 14 and the recording management queue 15 (k) are both formed in the memory 5.
[0036]
Hereinafter, a procedure of processing (hereinafter referred to as “input processing”) performed by the input unit 17 that wakes up as described above will be described with reference to a flowchart shown in FIG.
[0037]
First, the input means 17 generates search information in the memory 5 (FIG. 13, step S1). This search information is various information related to fixed-size blocks (hereinafter referred to as “blocks”), and is information generated in units of blocks constituting stream data. In the present invention, after the search information structure 231 shown in FIG. 7 is generated in the memory 5, the search information is generated by setting the following information in the search information structure 231.
[0038]
That is, an authentication character string (search information authentication character string) indicating search information is set in the “authentication character string area”, and the first block of stream data to be recorded is set in the “serial number area”. Set the serial number assigned in order from the first to the last block. In the “creation time area”, the current time is acquired and set from a clock element (RTC 60 shown in FIG. 1) that holds the current time with a predetermined accuracy (eg, 0.001 second accuracy). In the “area”, creator information is acquired from the file management unit 112 and set. Further, in the “last block flag area”, information (flag) indicating that the block is the last block is set.
[0039]
The current time set in the “creation time area” (corresponding to the start time of the recording) and the creator information set in the “creator area” (the user information instructing the recording) are It is the same value from the first block to the last block of stream data to be recorded.
[0040]
Next, the input unit 17 requires the divided storage information structure 241 shown in FIG. 8 in the memory 5 to store the divided storage information which is information obtained by dividing the file management information into a predetermined number m (hereinafter referred to as “divided storage information”). A number is generated (FIG. 13, step S2).
[0041]
The file management information is file management information currently managed by the file management unit 112, and is stored in the second recording means (for example, hard disk). Therefore, as a matter of course, this file management information does not include information on the stream data currently being written. The required number is a number necessary for dividing and saving all file management information, and corresponds to the total number m of file management information here. That is, the file management information can be classified into four types of structures shown in FIG. 6 (common information structure 221, volume information structure 222, stream information structure 223, and free space information structure 224).
[0042]
The common information structure 221 is one structure for each target file, and the volume information structure 222 is a structure for storing information related to the logical space that stores the target file. Normally, one structure for each file is the stream information structure 223. It is a structure for storing information on stream data constituting one file, and is required for the number of streams constituting one file. Furthermore, the free space information structure 224 is a logical space for storing the target file. It is a structure that stores information on empty files, and is assigned to each continuous series of empty spaces. The total number m means the total number of the structures.
[0043]
The size of these four types of structures does not exceed the size of the data storage area 241d (see FIG. 8) of the divided storage information structure 241. Therefore, if m divided storage information structures 241 (hereinafter referred to as “divided storage information structures 241 (1) to (m)”) are generated as described above, all file management information can be divided and stored.
[0044]
Here, the input unit 17 generates the divided storage information structures 241 (1) to (m) as described above, while sequentially obtaining the four types of structures from the file management unit 112, and obtains these structures. After copying sequentially to the “data storage area” of the divided storage information structures 241 (1) to 241 (m), the following information is further set in the divided storage information structures 241 (1) to (m).
[0045]
That is, an authentication character string (divided storage information authentication character string) indicating that it is divided storage information is set in the “authentication character string area”, and a format type (version version) of the file management information is set in the “format type area”. Information) is obtained from the file management unit 112 and set. In the “total number of sectors area”, the number of sectors necessary for storing all the divided storage information structures 241, that is, the total number m of the divided storage information structures 241 is set. set. In the “sector serial number area”, a number indicating the order in which the divided storage information structures 241 are generated, that is, any number from 1 to m is set, and in the “structure type area”, the divided storage information structure 241 is stored. An authentication character string that indicates which of the above four types of structures is set as the object to be divided and stored in the information structure 241 is set. Further, the number of divisions of information to be divided and stored (described later) is set in the “structure division number area”, and the information to be divided and saved is divided in the “structure division serial number area”. Serial numbers (to be described later) assigned in the order in which they are assigned are set, and an effective size (effectively existing size) of information to be divided and stored is set in the “effective data area size area”.
[0046]
As described above, since the size of the four types of structures is smaller than the size of the normal data storage area 241d, 1 is set in the “structure division number area” and the “structure division serial number area”. become. Since each divided storage information structure 241 (m) is stored in one sector (divided storage information storage sector 202), the size of the divided storage information structure 241 is set to 512 bytes which is a sector size (described later). The write time history information structure 251 and the read time history information structure 252 are set to 512 bytes for the same reason).
[0047]
Next, the input unit 17 inquires the recording buffer manager 16 about an area for storing the stream data to be input from this, and the recording buffer manager 16 that has received this inquiry returns a B pointer 212 described later to the input unit 17 ( FIG. 13, step S3).
[0048]
By the way, writing / reading in the hard disk is performed in units of sectors. Therefore, when data of one block is written to the hard disk, even if the data size of one block is specified as the write size of the write command, an area having a size N times the sector size is always used. Insignificant data is generated in the last sector. For example, when writing 513 bytes of data to the hard disk, since the sector size of the hard disk is 512 bytes, an area of 2 sectors (1024 bytes area) is actually used. In this case, the second sector The valid data is only the first byte, and the remaining 511 bytes are meaningless data.
[0049]
Therefore, in the present invention, as shown in FIG. 11, the recording buffer manager 16 assigns an area having a size of “(number of sectors necessary for writing a block N × sector size) + (sector size × 3)” to the recording buffer 14. To secure. When block size data is stored from the B pointer 212 which is the head position of the third sector of the N + 3 sectors, a surplus area is generated in the second sector from the end of the N + 3 sectors. Become. The search information structure 231 is stored in the remaining area (head position C pointer 213) of this sector.
[0050]
In FIG. 11, the first two sectors (A pointer) and the last one sector (D pointer) of the area secured in the memory 5 will be described later.
[0051]
In this way, since the search information structure 231 is written in the second remaining area from the end of the N + 3 sectors, generation of meaningless data can be minimized. Further, the point that the storage format of the hard disk 4 (FIG. 3 (a)) and the storage format of the memory block (FIG. 3 (b), FIG. 11) are not the same will be described later.
[0052]
Next, the input unit 17 designates the B pointer 212 acquired as described above to instruct the I / O adapter 10 to store the stream data, and the I / O adapter 10 that has received this instruction receives the B pointer. The first block of stream data to be recorded is stored in the area 212 (FIG. 13, step S4). During this storing process, the input means 17 is at rest.
[0053]
When the above storage processing is completed, the input means 17 stores the search information structure 231 of serial number 1 in the area indicated by the C pointer 213 (see FIG. 11), and then stores the divided storage information structure 241 (1) as the D pointer. It is stored in the area indicated by 214 (see FIG. 11) (FIG. 13, steps S5 → S6).
[0054]
Next, the input means 17 notifies the recording buffer manager 16 that the storage of the divided storage information structure 241 has been completed, passes the A pointer 211 (see FIG. 11) to the recording management queue 15 (1), and then writes it. The means 18 is woken up (FIG. 13, steps S7 → S8).
[0055]
The above procedure is repeated for each block (FIG. 13, step S9: No). When the storage of the last block of stream data is completed (FIG. 13, step S9: Yes), the input process is terminated.
[0056]
Since the above description is the procedure for inputting the first block, the divided save information structure 241 (1) is stored in the D pointer 214 (FIG. 13, step S7). The divided storage information structure 241 (m) is stored so that the divided storage information structure 241 (2) is stored in the procedure of inputting the third block and the divided storage information structure 241 (3) is stored in the procedure of inputting the third block. Are stored again from the divided storage information structure 241 (1) (the divided storage information is stored cyclically).
[0057]
Next, a procedure of processing (hereinafter referred to as “writing processing”) performed by the writing unit 18 that wakes up as described above will be described with reference to a flowchart shown in FIG.
[0058]
First, the writing means 18 obtains the head address of the free area in each hard disk 4 (k) from the file management unit 112, and temporarily enters a sleep state (FIG. 14, steps S11 → S12). The file management unit 112 specifies the head address of the free area based on the contents of the free area information structure 224 (see FIG. 6).
[0059]
Next, when the writing means 18 is woken up by the input means 17 as described above (see FIG. 13, step S8), the writing means 18 inquires of the recording buffer manager 16 whether there is data to be written, and receives this inquiry. The recording buffer manager 16 returns the A pointer 211 (hereinafter referred to as “write address”) stored in the recording management queue 15 (k) as described above to the writing means 18 (FIG. 14, steps S13 (k) → S14 (k). → S15 (k)). However, when there is no data to be written to the inquiry-target hard disk 4 (k) (FIG. 14, step S13 (k): No), or when the inquiry-target hard disk 4 (k) is being recorded ( In FIG. 14, step S14 (k): Yes), the writing means 18 makes an inquiry similar to the above for the next hard disk 4 (k).
[0060]
Next, the writing means 18 issues a write command to the hard disk 4 (k) specified by the write address (FIG. 14, step S16 (k)).
[0061]
Hereinafter, the procedure for issuing the write command (FIG. 14, step S16 (k)) will be described in more detail with reference to FIG.
[0062]
First, the writing means 18 determines whether or not the write command is the first write command (FIG. 15, step S16-1 (k)). Here, the first write command means a write command issued first in the recording process.
[0063]
Here, if the write command is the first write command, the writing means 18 obtains the current time from the RTC 60, saves the current time in the memory 5 as the write command issue time, and immediately thereafter, the A pointer 211. A write command is issued by designating a size (N + 1 sector size) obtained by subtracting two sectors from the size of the memory block from the address obtained by adding two sectors to the address (B pointer) (FIG. 15, step S16). -1 (k) → S16-5 (k) → S16-6 (k)).
[0064]
The two sectors are added or subtracted because the A pointer 211 stores the access time history information (write time history information structure 251 and read time history information structure 252) shown in FIG. This is because it is stored in the area shown.
[0065]
On the other hand, if the write command is a second or subsequent write command, the writing means 18 first stores access time history information (the generation procedure will be described later) in the area indicated by the A pointer 211, and then After the write command issue time is stored in the memory 5 in the same procedure as described above, the write command is issued immediately after designating the address of the A pointer 211 and the size of the memory block (size of N + 3 sectors) (see FIG. 15, Step S16-1 (k) → S16-2 (k) → S16-3 (k) → S16-4 (k)).
[0066]
At this point, the memory block stores ““ access time history time information ”,“ block ”,“ search information ”, and“ divided storage information ”from the beginning. The “access time history information” is not information related to the “block” written to the next B pointer, but information related to the block generated when the previous write command was issued. For example, if the blocks are A, B, C... And the access time history information corresponding to these blocks is A ′, B ′, C ′..., <Write A> → <Write A ′> → Rather than <B write> → <B 'write> → <C write> ..., <A write>-> <A' + B write>-> <B '+ C write> ... It is a procedure (<> represents a writing range of 1). As described above, according to the procedure of writing the access time history information after adding it to the next block (writing A ′ together with B), it is possible to prevent an increase in the number of write commands issued.
[0067]
The writing means 18 that has issued the write command as described above is in a sleep state while each controller 6 (k) is writing to the hard disk 4 (k), and wakes up when writing to any of the controllers is completed. Thus, the access time history information is generated (FIG. 14, steps S17 → S18 → S19).
[0068]
Hereinafter, the procedure in which the writing unit 18 wakes up when the writing of the first block of data to the hard disk 4 (1) is completed generates access time history information (FIG. 14, step S19) will be described with reference to the flowchart shown in FIG. Will be described in detail.
[0069]
First, the writing means 18 obtains the current time from the RTC 60, and after issuing the write command by subtracting the write command issue time for the hard disk 4 (1) stored in the memory 5 as described above from the current time. The time until writing is completed (hereinafter referred to as “writing time”) is calculated (FIG. 16, steps S19-1 → S19-2).
[0070]
Next, the writing means 18 secures an area for two sectors in the memory 5, and then has already been written to the hard disk 4 (1) (in the case of overwriting) access time history information storage sectors 203 and 204 (FIG. 3 ( a)) is read, and it is confirmed whether or not the access time history information authentication character string is stored in the information thus read (FIG. 16, steps S19-3 → S19-4).
[0071]
Here, when the writing means 18 confirms the authentication character string, the writing means 18 regards the information read as described above as access time history information, and a predetermined position in the sample time storage area of the writing time history information structure 251. The write time is stored in (initial value: “write time sample 1” storage area position) (FIG. 16, Step S19-4: Yes → S19-6). The writing means 18 that has stored the writing time adds 1 to the history start pointer (identifies the predetermined position) in preparation for the next writing time storage. When the sample time storage area becomes full, the old sample time is overwritten in order. That is, normally, data in the same area is read and written several times, each time a write time is recorded, a history of the write time in the area is recorded, and it is used later for detecting a defective sector.
[0072]
On the other hand, when the authentication character string cannot be confirmed, the writing means 18 writes the data in the state where the data is not yet written in the hard disk 4 (1) (new writing and still access time history information structure). First, a write time history information structure 251 and a read time history information structure 252 are newly generated in the memory 5. Thereafter, the writing time is stored in the same procedure as when the authentication character string is confirmed (FIG. 16, Step S19-4: No → S19-5 → S19-6).
[0073]
The access time history information used in the write command issuing step (FIG. 15, step S16-2 (k)) is the access time history information generated here. That is, the access time history information storing the writing time as described above is stored in the area indicated by the A pointer 211 by the writing means 18 and then accessed by the controller 6 (k) together with the next block size data. It is written in the time history information storage sector 203/204. The storage format of the memory 5 in FIG. 11 is shown in FIG. 3 (b) and the writing unit is shown in FIG. 3 (c) so that it can be compared with the storage format in the hard disk 4.
[0074]
Note that the above-described reading of the access time history information storage sectors 203 and 204 (FIG. 16, step S19-3) is performed during the stream data recording process. . However, since the magnetic head immediately before the reading is stopped at the position where the divided storage information storage sector 202 was written, the access time history information storage sectors 203 and 204 that are the next sector of the divided storage information storage sector 202 are read. As a result, no seek load occurs on the magnetic head.
[0075]
The writing means 18 having added the writing time to the writing time history information structure 251 as described above notifies the recording buffer manager 16 that the writing to the hard disk 4 (1) is completed, and also notifies the hard disk 4 (1). A value indicating the completion of writing is stored in a predetermined position of the corresponding recording management queue 15 (1) (FIG. 14, step S20).
[0076]
Here, the writing means 18 (FIG. 14, step S18) in the suspend state wakes up not only in the above case (when any of the hard disks 4 (k) has completed writing) but also in the input means 17 When a wake-up command is issued (step S8 in FIG. 13), the wake-up command is also issued. In this case, however, the writing means 18 inquires of the recording buffer manager 16 whether or not it exists in the memory 5 to be written (FIG. 14, steps S18 → S13 (k)).
[0077]
As described above, in the present invention, when there is an area to be written to the next hard disk (for example, 4 (2)) at the time when the write command to the certain hard disk (for example, 4 (1)) is completed, This is a procedure for issuing a write command to the hard disk 4 (2). In this way, a write command is issued in parallel to the hard disks 4 (1) to (4). As a result, the idle time of the hard disk 4 (k) and the idle time of the host bus 1 It is possible to continue recording stream data with as much as possible.
[0078]
If there is no write address in each recording management queue 15 (k) and the input of the stream data has been completed (step S21: Yes in FIG. 14), the writing means 18 sends each hard disk 4 (k ) After writing the access time history information of the last written block (last block) to each, the search information generated as described above is passed to the file management unit 112, thereby completing the registration of the stream data. (FIG. 14, steps S22 → S23). The writing of the access time history information of the last block (step 22 in FIG. 14) is necessary in view of the difference between the storage format and the recording format.
[0079]
Here, in preparation for recovery processing (recovery procedure 2) described later, a stream information structure 223 (information indicating that the stream information is the latest stream information) is generated for the stream data, and the stream data structure 223 is converted to the latest stream. The information is written in the information storage sector 200 (FIG. 14, step S24). The stream information structure 223 may be generated by the writing unit 18 or the file management unit 112.
[0080]
As a result, the stream data is recorded on the hard disk (k) in the format shown in FIG.
[Reproduction procedure]
A procedure for reproducing the stream data recorded as described above will be described below.
[0081]
First, when it is necessary to reproduce the stream data recorded on the hard disk 4, such as when an application (not shown) instructs the controller 6 (k) of each SCSI adapter 2 (k) to reproduce, under the control of the CPU 9, The reading means included in the reproduction control unit 113 wakes up.
[0082]
That is, as shown in FIG. 5, the playback control unit 113 includes a reading unit 22, an output unit 23, and a playback buffer manager 19. The playback buffer manager 19 stores the stream data in units of blocks. The details of one memory block are shown in FIG. 12) and a read management queue 21 holding a predetermined address on the reproduction buffer 20 is managed. Both the reproduction buffer 20 and the reproduction management queue 21 are formed in the memory 5.
[0083]
Hereinafter, a procedure of processing (hereinafter referred to as “reading processing”) performed by the reading unit 22 woken up as described above will be described with reference to a flowchart shown in FIG.
[0084]
First, the reading unit 22 inquires of the file management unit 112 whether or not there is data to be read (to be reproduced), and the file management unit 112 that has received this inquiry stores the hard disk 4 on which the data to be read is recorded. The top address of the upper area is returned to the reading means 22 (FIG. 17, step S30). The file management unit 112 identifies the head address based on the stream information structure 223 (see FIG. 6) and the like.
[0085]
Next, the reading means 22 inquires the reproduction buffer manager 19 about an area on the memory 5 for reading out and temporarily storing data to be reproduced, and the reproduction buffer manager 19 that has received this inquiry receives the reproduction buffer 20 on the memory 5. The address (storage address) of the empty area is returned to the reading means 22 (FIG. 17, steps S31 → S32: Yes).
[0086]
Here, the reading means 22 first determines the order and area of the hard disk 4 (k) to be read based on the head address acquired as described above, and then acquires the current time from the RTC 60, and this current After the time is stored in the memory 5 as a read command issue time, a read command is issued immediately to the controller 6 (k) corresponding to the hard disk 4 (k) determined as described above (FIG. 17, step S33 → S34).
[0087]
The read size of the read command is the size of the memory block shown in FIG. Accordingly, the controller 6 (k) that has received the read command reads from the head sector of the block to the read time history information storage sector 204. During this reading, the reading means 22 is at rest (FIG. 17, step S35).
[0088]
Next, the reading means 22 acquires the current time from the RTC 60, and subtracts the read command issue time stored in the memory 5 as described above from this current time, so that the read command is issued and the read is completed. Time (hereinafter referred to as “reading time”) is calculated (FIG. 17, steps S36 → S37).
[0089]
Here, the reading means 22 stores the read time calculated as described above at the position indicated by the current history start pointer (initial value: the position of the “read time sample 1” storage area shown in FIG. 9) (FIG. 17). Step S38). The point that 1 is added to the history start pointer and the point that the old sample time is overwritten are the same as in the writing process.
[0090]
Next, the reading means 22 notifies the reproduction buffer manager 19 of the completion of reading, and sets the storage address of the reproduction buffer 20 from the top of the read management queue 21 (FIG. 17, step S39).
[0091]
Here, if the above-described procedure is continued, the reading unit 22 cannot obtain the storage address because there is no free space in the reproduction buffer 20. Therefore, when there is no free space in the reproduction buffer 20, the reading means 22 wakes up the output means 23 and then pauses (FIG. 17, steps S31 → S32: No → S41 → S42).
[0092]
Next, a procedure of processing (hereinafter referred to as “output processing”) performed by the output unit 23 will be described with reference to a flowchart shown in FIG.
[0093]
As described above, the output means 23 woken up by the reading means 22 inquires the reproduction buffer manager 19 about the area where the data to be output exists, and the reproduction buffer manager 19 that has received this inquiry has the data to be output. The address of the area (the address on the reproduction buffer 20 and hereinafter referred to as “output address”) is returned to the output means 23 (FIG. 18, steps S50 → S51).
[0094]
Next, the output means 23 instructs the I / O adapter 10 to output the data of the block stored at the output address acquired as described above, and the I / O adapter 10 that has received this instruction Are reconstructed into stream data and output to an application or the like (not shown) (step S52 in FIG. 18). As a result, the application or the like starts to reproduce the stream data.
[0095]
When the output is completed, the output unit 23 wakes up, and the output unit 23 wakes up in this way notifies the reproduction buffer manager 19 that the output of the stream data is completed, and reads and manages a value indicating the completion of the stream data output. The queue 21 is set (FIG. 18, steps S53 → S54).
[0096]
Here, when the output is completed, an empty area for one block is created in the reproduction buffer 20. Accordingly, the output means 23 wakes up the reading means 22 at the time point (FIG. 18, step S55), and the reading means 22 that wakes up in this way acquires the address of the vacant area (FIG. 17, steps S42 → S31). In this way, it is possible to always read from the empty area of the reproduction buffer 20.
[0097]
[Recovery procedure 1]
A procedure for restoring the file management information using the search information when the file management information is destroyed due to some inconvenience will be described below with reference to the flowcharts shown in FIGS.
[0098]
First, when the user inputs an instruction for recovery conditions described below, the recovery means 114 sets the status status for authenticating the status to [free space], sets the frame counter (counter for counting the number of blocks) to 0, and The empty area first frame and stream first frame are initialized to zero. This set value is stored in a predetermined area of the memory 5 (FIG. 19, steps S60 → S61).
[0099]
In the present invention, the date information and the creator information are stored in the search information structure 231. Therefore, as the restoration condition, for example, “restore streams after M / D in Y year” Enter the date and conditions related to the creator, such as “Restore only the created stream data”.
[0100]
Next, the recovery means 114 stores the first search information in the hard disk 4 (1).Structure231 is read and it is confirmed whether or not the search information authentication character string is stored in the information read in this way (FIG. 19, steps S62 → S63 → S64).
[0101]
The recovery means 114 that has confirmed that the authentication character string is stored checks whether or not the status is [free space], and if it is [free space], whether the serial number is 0 or not. That is, it is determined whether or not the first retrieval information storage structure 231 has been read. Here, at the stage where the first search information storage structure 231 is read out, the date information and creator information stored in the search information structure 231 satisfy the recovery conditions instructed and input as described above. After confirming this, the value of the frame counter is set to the value of the first frame of the frame area, and the status state is set to [stream area] (FIG. 19, step S64: Yes → FIG. 20, steps S65-0 → S65-1). → S65-2 → S65-3 → S65 → S65-41 → S65-42).
[0102]
Next, the restoration unit 114 counts up the frame counter and reads the next search information storage structure 231 (step S66 in FIG. 19).
After confirming that the serial number stored in the retrieval information structure 231 read in this way is serialized (previous serial number + 1), whether or not the end block flag is set in the retrieval information structure 231 (FIG. 19, steps S66 → S62 → S63 → S64 → FIG. 20, steps S65-5 → S65-6).
[0103]
When the end block flag is set, the current frame counter value is set as the stream area end frame value, and the first frame value and end frame value are notified to the file management unit 112 (FIG. 20, step). S65-71 → S65-72). As a result, the file management unit 112 can create a storm information structure 223 for the area (I) in FIG.
[0104]
At this time, since the free area start frame = 0 and the stream area start frame = 0, a negative answer is obtained at the stage of comparing the two, and the value of the free area start frame is the current frame counter value + 1 (= α The status is set to [free space] (step S65-73 No.fwdarw.S65-76.fwdarw.S65-77).
[0105]
By the way, different data is overwritten on the hard disk several times. Therefore, some unnecessary data may be written as in the area (II) in FIG. Here, it is assumed that the silus number of the search information at the head of the unnecessary data is not 0 and the end block flag is not set.
[0106]
In this case, a procedure for acquiring search information continuously from the region (I) to the region (II) will be considered. As described above, at the stage after the storm information structure 223 of the region (I) is created, the status is set to [free space] (S65-76). Next, the search information of the next block is repeatedly acquired through the frame counter count-up stage (S66). Further, when the search information of the first block of the stream area in the area (III) in FIG. 21 (a) is obtained, the value of the frame counter (= β) at this time is the value of the first frame of the stream area. And the status is set to [stream area] (steps S65-41 and S65-42).
[0107]
Next, the routine on the right side of FIG. 20 is repeated, and when the end block flag is set, the stream information structure 223 is created as described above (step S65-62). At this time, since the empty area head frame area = α and the stream area head frame = β, the empty area end frame is set as β−1, the empty area start frame α, and the empty area end frame β−. 1 is notified to the file management unit 112 to create a free area structure 224 for the area (II) (step S65-73 Yes → S65-74 → S65-75).
[0108]
As shown in FIG. 21 (b), even if there is an area (IIb) where the serial number of the stream starts with 0 in the area (II), the status becomes [stream area] in this area (IIb). 20, step S65-42), and thereafter, the routine of S65-6 No-> S66 in FIG. 20 is repeated. Further, the stream area head frame is again updated (= β) with the head frame of area (III) (steps S65-5 → S65-8 → S65-1 to S65-42), and then the stream of area (III) After the information structure 223 is created, the area (II) empty area structure 224 is created.
[0109]
The above procedure is executed for all sectors, and the stream information structure 223 for all stream data satisfying the recovery conditions can be generated in the hard disks 4 (1) to (4).
[0110]
When the status for all sectors is completed and the status is “empty area”, the empty area head frame set in step S65-76 is set with the frame counter value −1 as the value of the empty area end frame. (Or the initial setting of 0) is notified to the file management unit 112. Based on this, the file management unit 112 creates a free space structure 224 (steps S67 to S69).
[0111]
Next, the recovery unit 114 generates a free space information structure 224 based on the stream information structure 223 generated as described above. That is, the area specified by the stream information structure 223 is an area (stream area) where stream data is recorded. Therefore, if an area other than the stream area is an empty area, all the empty area information structures 224 are stored. Can be generated.
[0112]
As described above, according to the present embodiment, since the stream data is recorded by adding the search information including date information and creator information, the file management information is set based on the date and the creator. It can be recovered.
[0113]
Furthermore, according to this recovery method, the latest stream data can also be recovered unlike the case of the recovery procedure 2 described below.
[Recovery procedure 2]
Hereinafter, the procedure for recovering the destroyed file management information using the divided storage information will be described with reference to the flowcharts shown in FIGS.
[0114]
First, when the user inputs an instruction to start the recovery procedure, the recovery unit 114 initializes the status number (sets 0), stores the status number in a predetermined area of the memory 5, and then determines the read start sector. (FIG. 22, steps S70 → S71).
[0115]
The read start sector is the head sector of the divided storage information storage sector 202 (total number m) that stores the divided storage information to be restored, and the “sector of the stream information structure 223 stored in the latest stream information storage sector 200 is“ This is the sector specified by the “first sector address”.
[0116]
Next, the recovery unit 114 reads the read start sector determined as described above, and confirms whether or not the divided storage information authentication character string is stored in the information thus read (FIG. 22, steps S72 → S73). → S74).
[0117]
The recovery unit 114 that has confirmed that the authentication character string is stored has the latest version of the file management format type information stored in the divided storage information structure 241 (information currently managed by the file management unit 112 and After confirming that the information is the same version), it is confirmed that the sector serial number stored in the divided storage information structure 241 is 1 (step S74 in FIG. 22: Yes → step S75-1 in FIG. 23). → S75-2).
[0118]
The recovery means 114 that has confirmed that the sector serial number is 1 calculates “sector size × total number of sectors” based on the information set in the “total number of sectors area” of the divided storage information structure 241. The area of the size obtained by this calculation is secured in the memory 5 (FIG. 23, step S75-2: Yes → S75-3). Hereinafter, the memory area secured in this way is represented as memory areas (1) to (m).
[0119]
Next, the restoration means 114 copies the first divided storage information structure 241 to the memory area (1) secured as described above, and then sets 1 to the status number (FIG. 23, steps S75-4 → S75- Five).
[0120]
Further, after the recovery unit 114 reads the divided storage information structure 241 from the next divided storage information storage sector 202 and confirms the file management format type and the sector serial number for the divided storage information structure 241 thus read, The divided storage information structure 241 is copied to the memory area (2) (steps S72 → S73 → S74 → FIG. 23 steps S75-6 → S75-7 → S75-8 in FIG. 22).
[0121]
The above procedure is repeated (step S72 → S73 → S74 → FIG. 23 step S75-6 → S75-7 → S75-8 → S75-9: No → step S72 in FIG. 22), “sector serial number = total number of sectors”. (Step S75-9: Yes in FIG. 23), the process is terminated.
[0122]
As a result, all the divided storage information structures 241 are stored in the memory areas (1) to (m), and the data (files) stored in the data storage areas of these divided storage information structures 241 are stored. All file management information can be recovered by passing the management information to any of the four types of structures that are management information).
[0123]
Here, the description has been made on the assumption that all of the four types of structures that are file management information are smaller in size than the data storage area of the divided storage information structure 241. When the size is larger than the area, the data stored in the plurality of data storage areas are combined to restore one structure. It is clear from the configuration of the divided storage information structure 241 (“structure division number” “structure division serial number”) that such a connection is possible.
[0124]
In addition, since the file management information is generated by the file management unit 112 after a certain file is stored in the recording medium, the latest stream information includes the file of the stream immediately before storing the latest stream information. Management information is divided and stored. Therefore, according to this recovery method, the latest stream data cannot be recovered.
[0125]
In addition, although the two recovery procedures have been described separately here, it goes without saying that these recovery procedures may be used in combination.
(Second Embodiment)
In this embodiment, in order to prevent deterioration in recording performance and reproduction performance of stream data, a defective area detection procedure described below is periodically performed.
[0126]
[Bad area detection procedure]
When it is necessary to start the defective area detection process, for example, when the user inputs an instruction to detect a defective area, the defective area detection unit 115 wakes up under the control of the CPU 9, and the defective area detection unit thus wakes up. 115 first reads out the first access time history information storage sectors 203 and 204 in the hard disk 4 (1), and confirms whether or not the access time history information authentication character string is stored in the information thus read ( FIG. 24, steps S80 → S81 → S82).
[0127]
The defective area detection means 115 which has confirmed that the authentication character string is stored extracts the latest sample times from the access time history information structure by a predetermined number (for example, 5), and extracts each of the extracted times. The magnitude relationship between the sample time and the maximum access time (set value) is examined (FIG. 24, steps S82 → S83).
[0128]
The maximum access time is determined based on the block size and hard disk performance. For example, when the block size is 120 KB and the hard disk has high performance (for AV use), the maximum access time is preferably set to a value of about 0.03 seconds.
[0129]
Here, when all the sample times are smaller than the maximum access time, the defective area detection means 115 reads the next access time history information storage sectors 203 and 204 (FIG. 24, step S84: No → S80 → S81).
[0130]
On the other hand, if any of the above sample times is larger than the maximum access time, the defective area detecting unit 115 determines that the block is a defective block, and generates a defective block structure 261 (see FIG. 10) in the memory 5. After transferring the bad block structure 261 to the file management unit 112, the next access time history information storage sector is read (FIG. 24, step S84: Yes → S85 → S80 → S81). The file management unit 112 that has received the bad block structure 261 registers the bad block structure 261 as file management information.
[0131]
When the above procedure is completed for all access time history information storage sectors 203 and 204 in the hard disks 4 (1) to (4) (FIG. 24, step S80: Yes), the defective area detection process is terminated.
[0132]
As the access time history information, either read time history information or write time history information may be used. Moreover, you may use both simultaneously.
[Recording procedure]
In the recording process in the present embodiment, stream data is recorded avoiding the defective blocks detected as described above, and the procedure will be described below with reference to the flowchart shown in FIG.
[0133]
The writing means 18 that has acquired the free space start address in each hard disk 4 (k) in the same procedure as in the first embodiment designates this start address and determines whether or not a defective block exists. The file management unit 112 that has received the inquiry returns information on the defective block (defective block structure 261) to the writing means 18 if there is a defective block behind the head address (FIG. 25, FIG. 25). Step S90 → S91).
[0134]
The procedure until the writing means 18 obtains the write address from the recording buffer manager 16 (FIG. 25, steps S92 → S93 → S94 → S95) is the same as that in the first embodiment, and the description thereof will be omitted. The write command issuance procedure (FIG. 25, S96) will be described in detail below with reference to the flowchart shown in FIG.
[0135]
First, the writing means 18 determines whether or not the write command is the first write command, and when it is determined that the write command is the first write command (when the access time history information is not written), Whether or not there is a defective block in a predetermined area of the hard disk 4 (k) to be written is determined based on the entire defective block structure 261 acquired as described above (FIG. 26, step S96-1: Yes). → S96-8).
[0136]
Next, when it is determined that there is no defective block, the writing means 18 issues a write command in the same procedure as in the first embodiment (FIG. 26, step S96-8: Yes → S96-9). → S96-10). If it is determined that a defective block exists, a write command is issued for the defective block portion by designating an address obtained by adding one memory block to the write address on the hard disk (FIG. 26, step S96-). 8: No → S96-11 → S96-9 → S96-10).
[0137]
On the other hand, when it is determined that the write command is a write command for the second time or later (when the access time history information is written), the writing means 18 first determines that the bad block does not exist, It is determined based on all defective block structures 261 whether or not there is a defective block in a predetermined area of the hard disk 4 (k) to be written (FIG. 26, step S96-1: No → S96-2).
[0138]
Next, when it is determined that there is no defective block, the writing means 18 issues a write command in the same procedure as in the first embodiment (FIG. 26, step S96-2: No → S96-3). → S96-4 → S96-5). If it is determined that there is a bad block, access time history information is temporarily stored in the memory 5 as delayed write information (to be described later) in order to prevent a decrease in recording performance, and then 1 memory is stored at the write address on the hard disk. A write command is issued by specifying an address obtained by adding blocks. (FIG. 26, step S96-2: Yes → S96-6 → S96-7 → S96-4 → S96-5).
[0139]
Here, as shown in FIG. 28 (a), block data (including the search information structure 231 and the divided storage information structure 241) Bd₂Block data Bd at the beginning of₁Access time history information Ah₁In the state in which is written along with the block data Bd₁And its access time history information Ah₁Has continuity. However, if a write command is issued by specifying the address added as described above, the access time history information Ah₁And block data Bd₁Are present across the defective block Bn, and physical continuity cannot be maintained.
[0140]
Therefore, if the above continuity is to be maintained, first, the block data Bd₁Access time history information Ah₁Block data Bd₁After writing to the back of the block data Bd₂Is written after the bad block Bn, but in this procedure, the number of write instructions increases, and the recording performance of the data stream decreases. Therefore, in the present invention, as shown in FIG.₁Is temporarily stored in the memory 5 as delayed write information (FIG. 26, S96-6), and the delayed write information temporarily stored in this way is stored in the corresponding block data Bd after the data stream input is completed.₁This is the procedure for writing after (FIG. 25, S103). In this way, the block data Bd₂Recording performance does not deteriorate because a single write command is sufficient to write.
[0141]
The subsequent procedure is the same as that of the first embodiment except that the step of writing the delayed write information (FIG. 25, step S103) is added, and the description thereof will be omitted.
[0142]
As described above, according to the recording process in the present embodiment, it is possible to record stream data while avoiding defective blocks by acquiring information on defective blocks in advance.
[Reproduction procedure]
Hereinafter, the procedure for reproducing the stream data recorded while avoiding the defective block as described above will be described with reference to the flowchart shown in FIG.
[0143]
The reading unit 22 that has acquired the read head address in the same procedure as in the first embodiment inquires the file management unit 112 whether or not there is a bad block by designating this read head address, and makes this inquiry. The received file management unit 112 returns information on the defective block (defective block structure) to the reading unit 22 if there is a defective block behind the read head address (FIG. 27, steps S110 → S111).
[0144]
The procedure until the reading means 22 obtains the storage address from the reproduction buffer manager 19 (FIG. 27, step S113: Yes) is the same as that in the first embodiment, so that the description thereof will be omitted. Explain the procedure.
[0145]
First, the reading unit 22 determines whether or not there is a defective block in a predetermined area of the hard disk 4 (k) to be read based on the entire defective block structure 261 acquired as described above (FIG. 27). Step S116).
[0146]
Next, when it is determined that there is no defective block, the reading unit 22 issues a read command in the same procedure as in the first embodiment (FIG. 27, step S116: Yes → S118 → S119), If it is determined that a bad block exists, a read command is issued by designating an address obtained by adding one memory block to the write address acquired as described above (FIG. 27, step S116: No → S117 → S118 → S119).
[0147]
Subsequent procedures are the same as those in the first embodiment, and a description thereof will be omitted.
[0148]
As described above, according to the reproduction procedure in the present embodiment, it is possible to reproduce the stream data recorded while avoiding the defective blocks as described above.
[0149]
In the above, access time history information (read time history information and write time history information) is used as a criterion for determining access performance of the hard disk. However, for example, the number of bad sectors included in one block may be used. In addition, although the reference access time is one type (0.03 seconds in the above example), a plurality of standards are set, and each standard is set to an application that requires particularly high-speed access at a certain speed. It may be used for an application that requires access, or an application that does not require a high-speed access. In this case, at the stage of the defective block structure generation process of FIG. 24 (step S85), levels corresponding to the plurality of criteria are written in the defective block structure, and the stage of defective block information income shown in FIG. In S91), a block having a predetermined level or less may be determined as a bad block according to the access speed required by the application to be used.
[0150]
In addition, the file management unit 112 manages file management information stored in a second recording medium. The second recording medium is normally a hard disk, but is not limited to a hard disk, and a non-volatile storage medium having a small capacity such as a flash memory may be used.
[0151]
【The invention's effect】
According to the restoration procedure 1, since search information is additionally recorded in the remaining area of the last sector in which the block is recorded, it is not necessary to separately prepare an area on the hard disk for recording the search information. Further, it is possible to restore only desired file management information by specifying a restoration condition. In this case, the file management information of the latest stream data stored can be recovered.
[0152]
According to the restoration procedure 2, the divided storage information is additionally recorded in parallel with the stream data recording process, so that the user does not need to consciously perform the backup process. Further, since it is only necessary to read out the sector in which the divided storage information is stored, and the reading process is terminated when the necessary information is confirmed, the file management information can be restored at a higher speed than in the restoration procedure 1.
[0153]
According to the second embodiment, the history information of the time required for writing the block, the history information of the time required for reading, or the access history information such as the number of bad sectors in the block is added to the block. Therefore, the performance of writing and reading performance for each block with respect to a predetermined address of the hard disk can be confirmed, and as a result, a defective area of the hard disk can be detected. Further, since the stream data is recorded while avoiding the defective area detected in this way, an overflow error that occurs when the recording speed becomes slower than the data input speed can be prevented. Furthermore, since the record of access time history information for the block immediately before the bad block is temporarily stored as delay record information and recorded at the end of the recording of a series of stream data, the stream data is not reduced. Can be recorded.
[0154]
If the defect level of a certain block is set based on the access history information, the defect level can be properly used according to the request of the application. As described above, the data management apparatus of the present invention does not manage files via a file system supported by the operating system in order to realize high-speed and seamless data recording / reproduction, but independently manages the files. After adopting the configuration, it has a recovery function and a defective area detection function.
[Brief description of the drawings]
FIG. 1 is a schematic functional block diagram of a data management apparatus according to a first embodiment.
FIG. 2 is a schematic functional block diagram of a data management apparatus according to a second embodiment.
FIG. 3 is an explanatory diagram of a recording format.
FIG. 4 is an explanatory diagram of a recording control unit.
FIG. 5 is an explanatory diagram of a reproduction control unit.
FIG. 6 is a configuration example of file management information.
FIG. 7 is a configuration example of search information.
FIG. 8 is a configuration example of divided storage information.
FIG. 9 is a configuration example of access time history information.
FIG. 10 is a configuration example of defective area information.
FIG. 11 is a diagram showing details of one memory block in a recording buffer.
FIG. 12 is a diagram showing details of one memory block in a reproduction buffer.
FIG. 13 is a flowchart illustrating an input processing procedure.
FIG. 14 is a flowchart illustrating a write processing procedure according to the first embodiment.
FIG. 15 is a flowchart showing a write command issue procedure in the first embodiment;
FIG. 16 is a flowchart showing a procedure for generating access time history information.
FIG. 17 is a flowchart illustrating a read processing procedure according to the first embodiment.
FIG. 18 is a flowchart illustrating an output processing procedure.
FIG. 19 is a flowchart showing a recovery processing procedure based on search information.
FIG. 20 is a flowchart showing a part of a restoration processing procedure based on search information.
FIG. 21 is a diagram illustrating a recovery processing state based on search information.
FIG. 22 is a flowchart showing a restoration processing procedure based on divided storage information.
FIG. 23 is a flowchart showing a part of a restoration processing procedure based on divided storage information.
FIG. 24 is a flowchart showing a defective area detection processing procedure;
FIG. 25 is a flowchart illustrating a write processing procedure according to the second embodiment.
FIG. 26 is a flowchart showing a write command issue procedure in the second embodiment;
FIG. 27 is a flowchart illustrating a read processing procedure according to the second embodiment.
FIG. 28 is an explanatory diagram illustrating a recording state according to the second embodiment.
FIG. 29 is a schematic functional block diagram of a conventional data management apparatus.
FIG. 30 is a conceptual diagram of recording processing.
[Explanation of symbols]
1 PCI bus
2 SCSI adapter
3 SCSI bus
4 Hard disk
5 memory
6 Controller
7 Buffer
8 Computer
9 CPU
10 I / O adapter
11 Control unit
14 Recording buffer
15 Record management queue
16 Recording buffer manager
17 Input means
18 Writing means
19 Playback buffer manager
20 Playback buffer
21 Read management queue
22 Reading means
23 Output means
111 Recording control unit
112 File Management Department
113 Playback control unit
114 Recovery measures
115 Defective area detecting means
200 Latest stream information storage sector
201 Search information storage sector
202 Division save information storage sector
203 Write time history information storage sector
204 Read time history information storage sector
231 Search information structure
241 Division save information structure
251 Write time history information structure
252 Read time history information structure
261 Bad block structure

Claims (10)

The stream data is divided into fixed-size data blocks, search information for identifying the divided fixed-size data blocks is generated corresponding to each of the divided fixed-size data blocks, and each of the generated search information A recording control unit that records the fixed-size data block and search information corresponding to the fixed-size data block on the first recording medium , so as to associate with the corresponding fixed-size data block ,
A file management unit for managing stream data recorded on the first recording medium based on file management information for specifying the stream data;
A recovery unit that recovers the file management information based on search information corresponding to each of the fixed-size data blocks obtained by dividing the stream data recorded on the first recording medium when the file management information recovery request is issued And
The first recording medium includes a plurality of sectors for storing a fixed amount of data,
Each of the fixed size data blocks obtained by dividing the stream data recorded on the first recording medium includes N sectors (N is an integer of 2 or more) , and the fixed size data block Search information corresponding to the fixed-size data block is recorded in the remaining area of the last sector among the N sectors included in each of the
A stream data management apparatus characterized in that the fixed size of the data block is larger than when (N-1) sectors are used and smaller than when N sectors are used . 2. The stream data management apparatus according to claim 1, wherein the file management information is recorded on a second recording medium different from the first recording medium. Above Symbol Search Information serial number, date information, authentication information, stream data management apparatus according to claim 1, wherein at least is one of the creator information. The first recording medium includes a plurality of sectors for storing a fixed amount of data,
The recording control unit stream data management apparatus according to claim 3, further recording the retrieval information to continue Ku first sector to the last sector included in the corresponding data block. The stream data is divided into fixed-size data blocks, search information for identifying the divided fixed-size data blocks is generated corresponding to each of the divided fixed-size data blocks, and each of the generated search information A recording control step for recording the fixed-size data block and the search information corresponding to the fixed-size data block on the first recording medium so as to associate with the corresponding fixed-size data block ;
A file management step for managing stream data recorded on the first recording medium based on file management information for identifying the stream data;
A recovery step for recovering the file management information based on search information corresponding to each of the fixed-size data blocks obtained by dividing the stream data recorded on the first recording medium when the file management information recovery request is issued And
The first recording medium includes a plurality of sectors for storing a fixed amount of data,
In the recording control step, each fixed-size data block obtained by dividing the stream data recorded on the first recording medium includes N sectors (N is an integer of 2 or more) , and the fixed-size data Search information corresponding to the fixed-size data block is recorded in the remaining area of the last sector among the N sectors included in each block ,
A stream data management method, wherein the fixed size of the data block is larger than when (N-1) sectors are used and smaller than when N sectors are used . 6. The stream data management method according to claim 5 , wherein the file management information is recorded on a second recording medium different from the first recording medium. 6. The stream data management method according to claim 5, wherein the search information is at least one of serial number, date information, authentication information, and creator information. The first recording medium includes a plurality of sectors for storing a fixed amount of data,
The recording control step corresponding stream data management method of claim 7, further recording the retrieval information to continue Ku first sector to the last sector in the data block. The stream data is divided into fixed-size data blocks, search information for identifying the divided fixed-size data blocks is generated corresponding to each of the divided fixed-size data blocks, and each of the generated search information A recording control step for recording the fixed-size data block and the search information corresponding to the fixed-size data block on the first recording medium so as to associate with the corresponding fixed-size data block ;
A file management step for managing stream data recorded on the first recording medium based on file management information for identifying the stream data;
A recovery step for recovering the file management information based on search information corresponding to each of the fixed-size data blocks obtained by dividing the stream data recorded on the first recording medium when the file management information recovery request is issued And
The first recording medium includes a plurality of sectors for storing a fixed amount of data,
In the recording control step, each fixed-size data block obtained by dividing the stream data recorded on the first recording medium includes N sectors (N is an integer of 2 or more) , and the fixed-size data Search information corresponding to the fixed-size data block is recorded in the remaining area of the last sector among the N sectors included in each block ,
A program recording medium having a program recorded thereon, wherein the fixed size of the data block is larger than when (N-1) sectors are used and smaller than when N sectors are used. . The search information is a serial number, date information, authentication information, program recording medium according to claim 9, wherein at least is one of the creator information.

Images