JPH0969027A - Recording system for multimedia - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain the recording system for multimedia which cancels a request to a disk device, generates data from the other constituent disk device, and carries on data transfer if a previously set response expected time is exceeded without waiting for a response from the disk device at the time of internal calibration of a constituent disk or error occurrence at the time of continuous transfer of a large amount of moving picture data, etc. SOLUTION: Once a controller 102 receives an access instruction from a user, a disk controller 120 refers to a time-out table 103 and acquires the previously set response expected time of the user. When data 100a is accessed and there is no response within the response expected time, the disk controller 120 reads data out of redundant data 100b in a data generating means 104 to enable a continuous read of the data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a recording system for multimedia, which is required to respond to an access request within a fixed time. In particular, the present invention relates to a multimedia recording system in which data is generated and transmitted from a recording device which is not requested to access when a response cannot be made within a fixed time. The present invention also relates to a multimedia recording system for recording a large amount of data. In particular, the present invention relates to a multimedia recording system in which a recording device can be easily added as the amount of data increases.

[0002]

[Prior art]

Conventional example 1. In the conventional recording system, when a large amount of data such as moving image data is continuously transferred, the data transfer is interrupted when reading the data accompanied by the movement of the internal head of the disk device. Then, a method of accumulating data in a buffer while the internal head is moving has been dealt with. FIG. 20 shows a diagram for explaining the above example. 20, a personal computer or workstation 300 includes a buffer 301 and a communication line 303 in a control device (not shown).
Connected through. The buffer 301 is provided to enable continuous data transfer of 3 MB / s to the personal computer or workstation 300 via the communication line 303. Then, the buffer 301 is
It is connected to the disk device 302 via a communication line 304. The disk device 302 is connected to the communication line 30.
Data transfer of up to 10 MB / s is possible via the No. The buffer 301 is a disk device 302.
In order to support the internal calibration (30 to 200 ms) of 3 MB / s × 200 ms, a data capacity of 600 KB is required. Furthermore, in order to handle the internal error retry (30 seconds to 45 seconds) of the disk device 302, a capacity of 90 MB is required from 3 MB / s × 30 seconds. Therefore, the buffer 301 cannot handle the error retry.

Conventional example 2. Further, in "Parallel data transfer method and device" of Japanese Patent Laid-Open No. 2-81123, a read failure is detected by a data delay, read failure data is generated from data other than the read failure, and a whole read failure is caused by a part of the read failure. It is characterized by avoiding data transfer delay or suspension.

FIG. 21 is a block diagram showing an example of the second conventional example. FIG. 22 is a block diagram of a conventional parity generator / data corrector in FIG. FIG.
3 is a circuit diagram of the correction control unit 14 of FIG. FIG.
FIG. 24 is a timing chart diagram for explaining the operation of the circuit diagram of FIG. 23.

In the second conventional example, a method for avoiding data delay due to synchronization deviation or data delay due to read failure or transfer stop is provided. The process of detecting the data delay and performing the data restoration in the above method and apparatus will be described below. The data read from the magnetic disk device 9 is converted into 8-bit width data by the hard disk controller 8. Then, it is stored in the data buffer 7. Data buffers 70, 71, 72, 73,
The 7P data is sent to the bus control unit 3 via the selector 5 if there is no read error. And DATARE
The data is grouped into a data width of 32 bits in synchronization with the ADY signal 13a and sent to the bus 2. Then, it is transmitted from the bus 2 to the host system. The magnetic disk device 9P for recording parity data includes magnetic disk devices 90, 91,
When a failure occurs in any of 92, 93, 9P, it is used to generate data of the failed magnetic disk device.

For example, when a failure occurs in the magnetic disk device 93 and the data cannot be read, the data is restored as follows. D3 = D0 XOR D1 XOR D2 XOR DP The above data restoration is performed by the parity generator / data corrector 6. The restored data is output from the parity generator / data corrector 6 and input to b of the selector 5. The selector corresponding to the magnetic disk device having the read failure by the error signal 13 selects the repair data and repairs the data. This is the end of the description of data restoration by detecting a read failure.

Next, the data restoration for the data delay due to the synchronization deviation will be described below. It is assumed that the DRVs 0 to 3 and P of the magnetic disk device 9 shown in FIG. 24 are synchronously executing the read operation continuously with the sectors 0, 1 ,. EC in DRV3 when reading sector 1
When the C error occurs, the data corresponding to the sector of the data buffer 73 is not transmitted from the data buffer, and the timing becomes 99 as shown in FIG. LRESET at timing 99a
Since 22a is output, the error locker 22 has been cleared, and the timer 21 has also been cleared in synchronization with REQ0-3 and P of sector 0. After this, REQ signal 150
Is first output, passes through the OR circuit 23 of FIG. 23, and starts the timer 21. After that, REQ1, REQ
2, REQP is also output before the TimeouT signal is output from the timer 21, but REQ3 is not output because a read error occurs. Therefore, the AND circuit 24
Since the D condition cannot be taken, the timer 21 sets REQREADY
Count UP is continued without being cleared by the signal 24a. At the timing 99b, the timer 21 sets the TimeouT
Is detected and a TimeouT signal 21a is output. Due to this output, the error lock unit 22 receives this Timeou
The inverted signal of REQ0-3 and P at the time of T is latched and output. The output of this latch is REQ, as shown in FIG.
It is ORed with 0 to 3 and P to become a signal 15a. By this OR, the signal 15a corresponding to REQ3 becomes true, and all the REQ0 to P3 signals become true. This allows AN
In the D circuit 24, the condition of AND is taken, and REQREADY
The signal 24a is true. As a result, the timer 21 is cleared and the TimeouT signal 21a becomes false. The start signal 23a of the timer 21, that is, the output of the OR circuit 23 is
The output of the error lock unit 22 fixes the true value, and the count UP is not restarted until it becomes false.

In this state, the AND condition of the AND circuit 24 is irrespective of the signal of REQ3.
2, it is established when the output of the REQP signal is complete, DA
The TAREADY signal 24a is set to true. At this time, if the error detector 20 does not detect an error, the AND condition of the AND circuit 25 is satisfied, and the DATAREADY signal 1
3a is true. This DATAREADY signal 13a
Is input to the bus control unit 3 in FIG. 21 and serves as a synchronization signal for data transfer. Output of error lock device 22 ERROR0
~ 3, P are the error signals 160, 161, 16 of FIG.
2, 163, 16P, and only the selector corresponding to the magnetic disk device in which the error has occurred is operated to select the port a of the selector 17. Output restoration data 1 of the parity generator 18 of FIG. 22 output under this input condition
2 becomes the value of the above equation, which is exactly the repair data of DRV3. Further, the output ERROR of the error lock device 22
0 to 3 and P are output as the error signal 13 of FIG. 21, and the repair data 12 is selected. In case of the error, E
Since RROR3 is true, the selector corresponding to the data bus 113 selects the b input and sends the repair data to the bus control unit 3.

Conventional example 3. Further, as a conventional method of recording image data such as video, a method of recording the data on a plurality of disks in a distributed manner is general. Figure 25 shows the RAID
It is a figure which shows the data recording state in level 3. FIG.
FIG. 6 is a diagram showing a data recording state when one is added to the disk configuration in FIG. For example, if the disk array is configured with RAID level 3 using four disk devices for data and one for parity, one data is striped and recorded in each disk device as shown in FIG. To be done. Here, it is considered that one data disk device (# 5) is added and reconfigured into a RAID level 3 disk array consisting of six devices. Striping after reconstruction is as shown in FIG. Figure 25
From FIG. 26, in order to reconfigure as shown in FIG. 26, almost all the data in the disk devices # 0 to # 3 must be rearranged and the parity must be recreated, as can be seen by comparing the figures.

[0010]

In the above-mentioned conventional example 1, in order to cope with the internal calibration of the disk device,
There is a problem that a large buffer capacity is required. Further, there is a problem that it is impossible to cope with the internal error retry of the disk device.

Further, the "parallel data transfer method and device" in the above-mentioned conventional example 2 is a means for detecting the occurrence of the synchronization deviation or the read failure due to the delay of the parallel data and starting the generation of the failure data. For this reason, as shown in FIG. 24, data transfer to the host is interrupted until the delay is detected (40 to 50 ms). Further, the data buffer 7 in FIG. 21 is for correcting the rotation operation of the DRVs 3, P and does not function as a data transfer buffer to the host. Therefore, there is a problem that continuous data transfer to the host is not compensated.

In addition, in the data recording method as in the above-mentioned conventional example 3, the data reorganization is required due to the expansion of the disk device, and the data reorganization process is very complicated and takes time. However, there is a problem that the data is lost and the reliability of the device is reduced.

The present invention has been made in order to solve the above-mentioned problem, and detects the retry due to the error occurrence of the disk and the occurrence of the internal calibration by the response time delay of the corresponding disk, and the other disk from the corresponding disk. And a data buffer for maintaining the data transfer to the host while executing the data restoration.

Further, the addition of a disk device can be realized very easily, and further, when a disk device fails, a multimedia recording system which does not affect the other disk devices is realized.

[0015]

A multimedia recording system according to the present invention has the following elements. (A) a recording unit for recording data and redundant data; (b) a control device for accessing the data recorded in the recording unit, recording the time at which access to the data recorded in the recording unit should be terminated. The timeout table,
A control device, comprising: a data generating unit that accesses the redundant data recorded in the recording unit and generates data when the access does not end even after the time stored in the timeout table has elapsed.

The recording unit is a RAID (Redundant Arrays of Inexpensive Disks) system composed of a plurality of disk devices, and the control device is a RAID controller.

The time-out table is characterized in that the time is stored for each user who requests data access.

Further, the control device is further characterized in that the time to be stored in the time-out table is set based on the data transfer rate required by the user who requested the data access.

Further, the control device is further characterized in that the time to be stored in the timeout table is set based on the response time required by the user who requested the data access.

Further, the control device is further characterized by setting a time to be stored in the time-out table based on a response time and a data transfer rate required by a user who requests data access.

Further, the recording section includes a data recording section for recording data and a redundant data recording section for recording redundant data, and the control device accesses the data recording section at the start of data access, and at the same time, the redundant data is recorded. It is characterized by accessing the recording unit.

Further, the control device further comprises a data buffer area for temporarily storing the data obtained by accessing the recording section and the redundant data, and the control device is provided in an empty area of the data buffer area. Based on this, the time to be stored in the timeout table is set.

The multimedia recording system according to the present invention has the following elements. (A) a plurality of disk devices in which a group of multimedia data is recorded in one disk device, (b) a redundant disk device in which redundant data of a plurality of multimedia data recorded in the plurality of disk devices is recorded, (c) )
A control device comprising a connection port connecting the plurality of disk devices and the redundant disk device, and accessing the plurality of disk devices and the redundant disk device as a RAID system.

Further, the control device further comprises an expansion port for expanding a new disk device in which a group of multimedia data is recorded, and when the new disk device is expanded, the control device records in the redundant disk device. Redundant data recalculating means for calculating new redundant data by using the redundant data thus created and multimedia data recorded in the new disk device and recording the new redundant data in the redundant disk device.

Further, when a failure occurs in any one of the plurality of disk devices, the control device prohibits access to the multimedia data recorded in the failed disk device. It is characterized by having means.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION

Embodiment 1. In this embodiment, data is divided and stored in a plurality of disk devices, and when there is no response within the preset expected response time of the timeout table when internal calibration or error of the constituent disk occurs, the data is stored in the disk. A description will be given below of a multimedia recording system that cancels the above request, obtains requested data from another component disk, and continues data transfer.

FIG. 1 is a system configuration diagram showing an example of a multimedia recording system according to the present invention. In FIG. 1, a multimedia storage device 106 and a user 108, which is a client terminal, are connected to a host system 130.
Connected by The multimedia storage device 106 includes the control device 102 and the storage unit 1 that is a recording unit.
00 and a data buffer 105. The storage unit 100 stores data in a mirror configuration. Data is stored in the disk device 100a, and redundant data is stored in the disk device 100b. The control device 102 accesses the disk device 100a according to an instruction from the user 108 and acquires data. At this time, the control device 102 refers to the preset time-out table 103 according to the continuous transfer rate or the maximum response time requested by the user 108, and determines whether or not there is a response from the disk device within the response expected time for each user. Manage response time. If there is no response within the expected response time, the data generation unit 104 acquires redundant data from the disk device 100b. The control device 102 stores the accessed data in the data buffer 105 (stores redundant data in the data buffer 105 when there is no response within the expected response time). Then, the data is transmitted from the data buffer 105 to the host system 130.

FIG. 2 is a diagram for explaining the operation of the control device in the system configuration of FIG. In FIG. 2, the timeout table 103 is a table that stores a response expected time for each user. Although not shown in the control device 102 of FIG. 1, the disk configuration table 121 is also a table stored in the control device 102. In the disk configuration table 121, the transfer speed between the disk device and the buffer, average seek time, 1 track read time,
Each information of the rotation waiting time and the number of error retries is stored. Although there are two disk devices connected to the control device 102, since both disk devices have the same configuration, there is only one disk configuration table 121. In addition, the timeout table 103 is defined by the continuous transfer rate or the maximum response time required for each user who uses the multimedia recording system according to the present invention.
The expected response time is set in advance. If not set, 30 ms is set.

FIG. 3 is a flow chart showing the procedure of data read processing in the multimedia recording system. FIG. 4 is a flowchart showing the procedure of the timeout process in FIG.

Next, the procedure for reading data from the system using the multimedia recording system will be described with reference to FIGS. 1 and 2 and according to the flowcharts of FIGS. 3 and 4. First, when the user accesses the multimedia recording system, the timeout table 1
The expected response time is set according to the continuous transfer rate or maximum response time required for 03. Even if the value has been set, but the set value is desired to be changed, the expected response time is newly input and the timeout table 103 is updated (S1).
In the multimedia recording system of this embodiment, the initial value of the expected response time is 30 ms.
Next, the disk controller 120 confirms that a value is set in the disk configuration table 121 (S
2). If no value is set in the disk configuration table 121, it is not possible to check whether the expected response time of the specified timeout table 103 is a realizable value. Further, in the example described in the third embodiment,
The expected response time set in the timeout table 103 is calculated using the value stored in the disk configuration table 121. In the second embodiment, it is possible to check whether the expected response time is a realizable value by the same calculation as the expected response time. For this reason, a value must be set in advance in the disk configuration table 121. The timeout table 103 and the disk configuration table 121 are assumed to exist in a non-volatile memory such as an EEPROM or a disk system area. The disk controller 120 checks the disk configuration table 121 (S2), and if the response expected time set in the timeout table 103 is not a proper value, it is not shown in FIG. Display a warning. When a proper value is set in the timeout table 103, the control device 102 waits for a Read or Write request from the host. When there is a Read or Write request from the host (S
3), the disk controller 120 obtains the expected response time corresponding to the user in the timeout table 103, and the timer A managed by the disk controller 120
The expected response time obtained in step S4 is set (S4). And
The disk controller 120 is a disk device 100a.
A Read command or a Write command is issued to (S5). The disk controller 120 monitors the time until a response is returned after issuing the Read command or the Write command to the disk device 100a. Within the expected response time set in the timer A, the disk device 100a
If there is more response (S6), in the case of Read processing, the read data is stored in the data buffer 105. In the case of Write processing, data is written from the data buffer 105 to the disk device 100a. And
Wait for Read / Write request from host. When there is no response from the disk device 100a even after the value of the timer A has passed (S7), when the request is a read request, the disk controller 120 instructs the data generation means 104 to read redundant data. Data generation means 1
04 accesses the disk device 100b to read redundant data.

In the multimedia storage system shown in FIG. 1, since data is recorded in a mirror configuration, even if the data cannot be read, if the data is obtained from the disk device 100b storing the redundant data. , You can get the necessary data. The data generating means 104 stores the read redundant data in the data buffer 1.
05. The data buffer 105 transfers the acquired data to the user who has issued the Read request (S8). After that, the data generating means 104 activates the timeout process of FIG. 4 (S9). When the time-out process ends, the control device 102 reads the Read from the host.
Wait for / Write request again.

Next, the procedure of the timeout process will be described with reference to the flowchart of FIG. Disk controller 120
Sets a timer B for disconnecting the disk device 100a which has not responded (S10). Here, the error retry time is set to 200 ms and is set in the timer B.
The timer B is a time limit for waiting for a response from a disk device in which a failure has occurred, and an arbitrary value is set in advance in the system. Then, it is monitored whether there is a (timer B) response within 200 ms (S11, S
12). If there is no response within 200 ms, the disk is disconnected and a failure report is made (S17). 200 ms
If there is a response within that, it is judged whether the retry process is OK or not according to the response from the disk device 100a (S1).
3). Then, if the data can be restored by the retry function of the disk device (retry OK), an error log is output (S15). Although not shown, the error log exists in the error management area of the control device 102. If the retry process is not OK, the disk device 1
Redundant data acquired from 00b
Is written to and data is restored (S14). Then, the error log is output (S15). After that, the number of error occurrences is checked, and if the number of error occurrences allowed in the system exceeds the specified value, the disk device 100a in which the failure has occurred is disconnected from the system and a failure report is made (S17). In order to ensure the safety of the system, the above specified value sets a limit on the number of failures that occur in the entire system, and when failures occur beyond the limit,
This is for disconnecting the faulty part from the system. If the number of error occurrences is within the specified value, the timeout process ends. It should be noted that the above specified values are prepared for ensuring the safety of the system and are not always necessary in the present invention.

Further, in the above example, the access to the redundant disk is started after the timeout is detected, but the access to the redundant disk may be executed in advance. In this case, since the storage unit 100 has a mirror configuration, the same data can be obtained from the disk device 100a and the disk device 100b. Therefore, the disk controller 120
The data from the disk device having the faster response is acquired and stored in the data buffer 105. For example, the disk controller 120 simultaneously accesses the disk device 100a and the disk device 100b,
When the response from b is fast, the data is acquired from the disk device 100b and stored in the data buffer 105. Further, the host system 130 connects a plurality of users, and the storage unit 100 is shared by the plurality of users. Therefore, when a plurality of users access the recording unit 100 at the same time, an access wait occurs. However, waiting for access can be avoided by previously determining the disk device to be accessed for each user. For example, the user 1 decides in advance to access the disk device 100a. Also, the user 2 decides in advance to access the disk device 100b. By doing so, even if the user 1 and the user 2 try to access the storage unit 100 at the same time, the disk controller 120 responds to the read command from the user 1 by
The disk device 100a is accessed first. Then, if there is no response from the disk device 100a within the expected response time, the disk device 100b is accessed to acquire the data. Also, in response to a read command from the user 2, the disk controller 120 first sends the disk device 1
Access 00b. Then, if there is no response from the disk device 100b within the expected response time, the disk device 100a is accessed to acquire the data.

Further, the disk controller 120 is
The usage status of the disk device 100a and the disk device 100b may be determined and either disk device may be accessed. For example, when the disk device 100a is in use, the disk controller 120 accesses the disk device 100b and acquires redundant data. In this way, the disk controller 120 responds quickly to access requests from users.

As described above, the data generating means according to the present invention, if there is no response from the disk device within the expected response time, if the disk device can be restored by the retry function of the disk device, the disk device will retry. The function automatically recovers data. If the disk device cannot be restored by the retry function, the redundant data is read from the disk device that stores the redundant data, and the redundant data is rewritten to the disk device in which the error has occurred to recover the data. It can be performed. Further, by simultaneously accessing the disk device storing the data and the disk device storing the redundant data, the data restoration can be executed immediately.

Although the data reading has been described above, when writing data, S8 in FIG. 3 is performed.
Is changed to a process of writing the data received from the host to the disk device 100b storing the redundant data. Further, in the flow chart of FIG. 4, the data received from the host in the process of S14 is stored in the disk device 100.
Write to a. Other than that, the process is the same as the read process described above.

Embodiment 2. In this embodiment, the operation of the data generating means when the disk configuration is the RAID configuration will be described below. FIG. 5 is a system configuration diagram showing an example of a multimedia recording system according to the present invention. In FIG. 5, the host system 130
Connects a plurality of users 108 via a communication line 109. In addition, the host system 130 uses the communication line 1
The control device 102 is connected via 07. The control device 102 is connected to the storage unit 100 having a RAID configuration. The disk configuration in this embodiment is 5 in the horizontal direction and 3 in the vertical direction, and the RAID level is 4. The redundant data is assumed to be stored in the disk device 100g. SPC133 is SCSIPr
It is an otcor controller. SPC133
Connects the storage unit 100 and the data generation unit 104. The data generation means 104 includes 5 from CH0 to CH4.
One channel 132 is provided. Data buffer 10
Reference numeral 5 denotes the host I / F control unit 131 and the data generation means 10.
4 are connected. The data buffer 105 stores the write data transmitted from the host system 130 and the data read by the data generation means 104. The data buffer 105 is composed of a FIFO. The system control MPU 134 acquires and decodes the access command transmitted from the host system 130 via the host I / F control unit 131. Further, it sends a decoding completion response to the host I / F control unit 131. Furthermore, system control M
The PU 134 gives a channel switching instruction to the data generating means 104, or gives an instruction to a disk device for rewriting the restored data. In addition, system control M
The PU 134 issues an I / O issue instruction to the SPC 133 and receives a completion signal from the SPC 133. Furthermore, the system control MPU 134 includes a timeout table 103 that stores a response expected time for each user.

FIG. 6 shows the data generating means 10 in FIG.
4 is a circuit diagram of FIG. In FIG. 6, reference numeral 135 denotes a switch, and when the response from the disk device which issued the I / O command is not within the response expected time, the data restoration circuit 137.
It is a switch for switching the line to input more data. The buffer 136 is the disk device 100c.
This is a storage area for storing data input from 100 g. The switch 138 is a system control MPU 13
C by the channel switching instruction transmitted from
The connection with the five channels 132 from H0 to CH4 is switched. The data restoration circuit 137 restores the data based on the data received from the buffer 136, and retransmits the restored data to the buffer 136 for the disk device in which the error has occurred.

FIG. 7 is a diagram showing the setting contents of the timeout table and the disk configuration table in the system configuration of FIG. The timeout table 103 and the disk configuration table 121 are stored in the system control MPU 134.

In the system configuration of FIG. 5, the disk configuration table 121 is set in advance as shown in FIG. 7 (B). For example, a case where the user 1 accesses the data stored in the storage unit 100 via the host system 130 will be described with reference to the flowchart of FIG. First, the user 1 sets the expected response time to “22 ms” in the timeout table 103 (S1). The response expected time is the time from when the disk devices 100c to 100g are accessed by the system control MPU 134 until the response is returned to the data buffer 105. The control device 102 checks whether the set expected response time is a realizable time (S2). In this embodiment, a total of 6 from the disk devices 100c to 100f.
It is assumed that 0 KB of data is transferred at one time. Also,
It is assumed that necessary data is stored in one track of each of the disk devices 100c to 100f. The content of the disk configuration table 121 is as shown in FIG. 7B, and is “10 MB / s”.
Is the transfer speed between one disk device and the data buffer, and if there are four disk devices, it is "10MB / s.
With the capacity of “× 4”, the transfer between the disk device and the data buffer can be performed. Therefore, the disk device 100
c, 100d, 100e, 100f and data buffer 1
The maximum achievable continuous transfer time between 05 and
Including (when the acquired data is not stored in the next track, that is, when the track is randomly accessed), 60 KB / (10 MB / s × 4 units) +16 ms + 4 ms
= 21.5 ms. Here, as shown in FIG.
ms is an average seek time, and 4 ms is a rotation waiting time.
In addition, if the data to be acquired is stored in the next track, that is, in the case of sequential access,
Only one track seek, 60 KB / (10 MB / s × 4 units) +0.6 ms = 2.
It is required to be 1 ms. Here, as shown in FIG.
6 ms is the next track seek time. The disk configuration table 121 in FIG. 2 stores the average seek time and the next track seek time, which are the seek times required to read the data. Depending on the type of disk device, there are a disk device in which the seek time for reading data and a seek time for writing data are the same time, and a disk device in which the seek time is different. In this embodiment, as long as the seek time for reading data is stored in the disk configuration table 121, either type of disk device may be used. Since the expected response time set by the user 1 is “22 ms”, it can be seen that it is a realizable value. Next, the control device 102 receives a Read / Write request from the user 1 via the host system 130 (S3). The host I / F control unit 131 sends the Read / Write request received from the host system 130 to the system control unit MPU 134. The system control MPU 134 decodes the received instruction and sends a completion response to the host I / F control unit 131. Transmission / reception between the host I / F control unit 131 and the system control MPU 134 is performed via 134a.

The system control MPU 134 is shown in FIG.
The expected response time “22 ms” of the user 1 is acquired from the timeout table 103 of FIG.
4). Then, via 134b, the system control MPU1
34 instructs the data generation means 104 to switch the channel. Since the disk device 100g is for parity, the system control MPU 134 instructs the data generation means 104 to switch the channel so as to disconnect the disk device 100g. The data generation means 104 follows the channel switching instruction from the system control MPU 134, and switches 1 via the switch 1b.
The switch is switched so as to disconnect the connection of the disk device 100g with respect to 38. Furthermore, system control MP
The U 134 sends an I / O command to the SPC 133 to the disk device that stores the corresponding data via 134c.
An O issue instruction is given (S5). The SPC 133 reads the data from the disk devices 100c to 100f according to the I / O issue instruction from the system control MPU 134. When the data reading from the disk devices 100c to 100f is completed, the SPC 133 reports the reading completion to the system control MPU 134 via 133c. On the other hand, the system control MPU 134 activates the timer after issuing an I / O issue instruction to the SPC 133.
Wait for the read completion report from 133. System control M
When the PU 134 receives the read completion report from the SPC 133 earlier than the value of the timer A, the PU 134 waits again for a Read / Write request from the host system 130 (S
6). The system control MPU 134 acquires data from the redundant data stored in the disk device 100g when there is no read completion report from the SPC 133 even after the response expected time set in the timer A is exceeded (S7) (S7).
8). For example, if a failure occurs in the disk device 100c,
It is assumed that reading cannot be performed from the disk device 100c. The system control MPU 134 instructs the data generation means 104 via 134b to generate data from redundant data. Data generation means 104
When receiving the instruction from the system control MPU 134, the switch disconnects the disk device 100c using the switch 138 and switches the switch to connect the disk device 100g. Then, based on the data read from the disk device 100d from the disk device 100d, the data restoration circuit 137 restores the data that was to be read from the disk device 100c.

A method of reproducing data using redundant data will be described with reference to FIG. Data D is stored in the disk device 100c.
0, data D1 in the disk device 100d, data D2 in the disk device 100e,
Data f3 is stored in f, and redundant data DP is stored in the disk device 100g. When the exclusive OR is obtained using these data, it becomes “0” as shown in FIG. Therefore, when a failure occurs in the disk device 100c, the data D0 stored in the disk device 100c can be obtained by the formula of FIG. 8 (B). Then, the data generating means 104 causes the buffer 13 in which the restored data is connected to the disk device 100c.
6 is input to the disk device 100c by using the switch 135, and the switch is switched so that data is input from the data restoration circuit 137. Further, the data generating means 10
4 disconnects the connection of the disk device 100c by using the switch 138 and switches to the input from the disk device 100g. Then, the data generation unit 104 transmits the restored data to the data buffer 105 and the data read from the disk device 100d to the disk device 100f. The host I / F control unit 131 sends a data read completion report to the system control MPU 134. This is done via 134a. The data buffer 105 is also transmitted via the host I / F control unit 131.
The data is transferred from the host system 130 to the host system 130 (S8). After that, the data generating means 104 activates the timeout processing according to the flowchart shown in FIG. 4 (S
9). After the time-out processing ends, the control device 102
Is read / write from the host system 130.
Wait for the request again.

The timeout process will be described with reference to the flowchart of FIG. First, system control MPU134
Gives an instruction to the data generating means 104 to disconnect the disk device 100c in which the error has occurred. Then, the timer B is set. In this embodiment, the timer B is set to "200 ms". The timer B is a time limit for waiting for a response from a disk device in which a failure has occurred, and an arbitrary value is set in advance in the system. The system control MPU 134 waits for a response from the disk device 100c until the set timer B is exceeded (S11, S12). If there is no response from the disk device 100c even if the timer B is exceeded, the system control MPU 134 sends the user 1 via the host system 130.
Report the problem to. Further, the failure report is also made to the operator who is monitoring the system (S17).

When there is a response from the disk device 100c, or when the disk device 100c can be restored by the retry function of the disk device (S13), the system control MPU 134 outputs an error log to the error management area (S15). If the disk device 100 cannot be restored by the retry function of the disk device, the system control MPU 134 instructs the SPC 133 to write the data restored by the data restoration circuit 137 to the disk device 100c. The SPC 133 rewrites the restoration data stored in the buffer 136 to the disk device 100c (S14). Then, the system control MPU1
34 outputs an error log to the error management area (S
15). The error management area is the system control MPU.
It is assumed to be included in the control unit 102 or the control device 102. Next, the system control MPU134
Determines whether the number of errors exceeds the specified value. In order to ensure the safety of the system, the above-mentioned specified value is to set a limit on the number of failures that occur in the entire system, and when a failure occurs in excess of the limit, the failure location is separated from the system. If the specified value is exceeded, the disk device 100c is disconnected from the system and a failure report is made (S17). The above specified value is
It is prepared in order to ensure the safety of the system and is not absolutely necessary in the present invention.

As described above, in this embodiment, the disk configuration is the RAID configuration. Even if the disk has a RAID configuration, the expected response time can be set for each user as in the first embodiment. Further, the user can obtain the data within the set expected response time. Further, even if a failure occurs in the disk device, it is possible to restore the data by using the data from the disk device in which the failure has not occurred.

In the second embodiment, since the disk has the RAID structure, the disk devices 100c to 1c
The access control for 00f is performed by the system control MPU1.
34 is doing. That is, when the data stored in the storage unit 100 is accessed from the host system 130, the system control MPU 134 causes the disk device 1 to access the data.
Which of the disk devices from 00c to 100g is to be accessed is determined. Therefore, in order to acquire redundant data at the same time as data, it is necessary to customize the function of the system control MPU 134. Also, RA
Since it has an ID structure, it is not possible to generate data only with redundant data. For example, to generate the data stored in the disk device 100c, the disk device 1
The data stored in 00d to 100g is required. For this reason, unlike the mirror configuration of the first embodiment, data cannot be generated only by acquiring data from the faster response time of the disk device 100a and the disk device 100b. Further, as shown in FIG. 5, when the host system 130 connects a plurality of users, access to the storage unit 100 may occur frequently and the data buffer 105 may run out of free space. In this case, by providing a plurality of data buffers 105, it is possible to deal with frequent access to the storage unit 100. Also,
Depending on the capacity of the memory and the number of users connected, a data buffer may be provided for each user. In this way, the access to the storage unit 100 from the user is dealt with quickly.

Embodiment 3. In this embodiment, a method for setting the timeout table based on the transfer rate request management table in the same system configuration as that of the first embodiment shown in FIG. 1 will be described below. FIG. 9 is a diagram showing an operation when the control device in FIG. 1 includes a transfer rate request management table. In the figure, the control device 102 uses the transfer rate request management table 12
2 and the buffer capacity table 123. The other components are the same as those in FIG. 2 described in the first embodiment, and thus the description thereof will be omitted.

FIG. 10 is a diagram showing setting contents of each table included in the control device of FIG. In this embodiment,
The total capacity of the data buffer 105 is set to 1 MB, and in order to evenly distribute the buffer to each user, it is set to 100 KB per user. Then, the buffer capacity table 1
The buffer size for each user is set in 23. In this embodiment, it is assumed that it is set in advance as shown in FIG.

In the above-mentioned first and second embodiments,
The user has set the expected response time in the timeout table 103. In this embodiment, the user sets the transfer rate in the transfer rate request management table 122 without directly setting the expected response time in the timeout table. The transfer rate is a data transfer rate between the data buffer 105 and the host system 130. Then, the disk controller 120 of the control device 102 makes the disk configuration table 121 and the transfer rate request management table 12
2 and the buffer capacity table 123, the expected response time is calculated for each user, and the timeout table 103 is calculated.
Set the calculated value to.

A method of calculating the expected response time for each user will be described below with reference to FIG. The transfer rate request management table 122 is set for each user when the system is started. This corresponds to the process of S1 in the flowchart of FIG. 3 described in the first and second embodiments. In the processing of S1, the time-out table was set, but in this embodiment, the transfer rate request management table 1
The transfer speed is set to 22. The transfer rate is a rate at which data is transferred from the data buffer 105 to the host system 130. Initial transfer rate is 1MB / s
It is. A setting example of the transfer rate request management table 122 is shown in FIG. For example, user 1 has a transfer rate of 1
MB / s is required. The buffer capacity is 10
It is 0 KB. This allows 100 KB / 1 MB / s between the data buffer 105 and the host system 130.
Data transfer of 100 KB is performed in a time of 100 ms. In the third embodiment, it is desired to continuously transfer data from the data buffer 105 to the host system 130. Therefore, the disk controller 120 accesses the disk device 100a to acquire data and stores the acquired data in the data buffer 105 so that the amount of data stored in the data buffer 105 does not become 0 bytes. . To access the disk device 100a, referring to the disk configuration table 121 in FIG. 2, average 16 ms (average seek time) +4 m
A seek time of s (waiting for rotation) = 20 ms is required. Therefore, for 20 ms after the start of seek, the disk device 1
Data cannot be transferred from 00a to the data buffer 105. In other words, to transfer 100 KB of data from the disk device 100a to the data buffer 105,
B (capacity of the data buffer 105) / 10 MB / s (transfer time between the disk device and the buffer: from the disk configuration table 121 of FIG. 2) can be calculated to be 10 ms, but actually, including the average seek time, 10 ms + 20 ms =
It will take 30 ms. Therefore, when the expected response time is calculated, 100 KB / 1 MB / s− (100 KB / 10 MB / s
+4 ms + 16 ms) = 70 ms. That is, in order to continuously transfer data from the data buffer 105 to the host system 130 at the speed requested by the user 1, the disk controller 120
70ms after accessing the disk device 100a
A response is returned from the disk device 100a until the time elapses, and it is monitored that data transfer from the disk device 100a to the data buffer 105 can be started. Similarly, when the response expected time of the user 2, the user 3, ..., The user n is calculated, it is obtained as shown in FIG. Further, the buffer capacity can be changed for each user as shown in FIG. 10 (C). If there is a difference in the amount of data to be processed for each user in advance, it is possible to allocate an appropriate buffer capacity to each user and improve the processing efficiency. When the buffer capacity is set as shown in FIG. 10C, the expected response time of each user is as shown in FIG.

As described above, the expected response time is calculated based on the data transfer rate requested by the user and set in the timeout table, so that the buffer can be used more efficiently.

Embodiment 4 FIG. In this embodiment, a method for setting the timeout table based on the response time request management table in the same system configuration as that of FIG. 1 of the first embodiment will be described below.

FIG. 12 is a diagram showing an operation when the control device in FIG. 9 has a response time request management table. In the figure, the control device 102 newly includes a response time request management table 124. The response time request management table 124 is set for each user, and is set at system startup. The disk controller 120 uses the response time request management table 124 and the disk configuration table 121.
The expected response time is calculated for each user, and the calculated value is set in the timeout table 103.

FIG. 13 shows the response time request management table 12
FIG. 4 is a diagram showing setting contents of No. 4; According to FIG. 13, for example, the user 1 sets the response time as “100 ms”. The response time is the disk controller 120.
Is the time from when the access request is issued to the disk device 100a until the data is completely transferred to the data buffer 105. However, since it is assumed that data is continuously transferred from the data buffer 105 to the host system 130, when calculating the expected response time, the seek time during which data is not transferred from the disk device 100a to the data buffer 105 is calculated. Must be deducted from the response time set by user 1. Therefore, when the response expected time is calculated based on the response time and the disk configuration table 121, it is calculated as 100 ms- (16 ms + 4 ms) = 80 ms. Similarly, user 2, user 3, ...
When the expected response time of the user n is calculated, it is obtained as shown in the timeout table 103 of FIG.

As described above, since the expected response time is set based on the response time requested by the user, it is possible to provide the service according to the strength of the user's request.

Embodiment 5 In this embodiment, the expected response time is calculated based on the transfer rate request management table and the response time request management table in the same system configuration as that of the first embodiment shown in FIG. 1, and the calculated value is small. The method of setting the other value in the timeout table will be described below. FIG. 14 is a diagram showing an operation when the control device in FIG. 1 includes a transfer rate request management table and a response time request management table. In the figure, the control device 102 newly includes a transfer rate request management table 122, a buffer capacity table 123, and a response time request management table 124. The other constituent elements are the same as those in FIG. 2 described in the first embodiment, and thus the description thereof will be omitted. FIG. 15 is a diagram showing setting contents of the timeout table.

The user selects the transfer rate request management table 122 and the response time request management table 124 when the system is started.
To, set an arbitrary value. And the control device 1
02 refers to the transfer rate request management table 122, the response time request management table 124, the disk configuration table 121, and the buffer capacity table 123, and calculates the response expected time for each user. Then, the calculated expected response time is set in the timeout table 103. For example, a method of calculating the expected response time for the user 1 will be described. According to FIG. 10A, the transfer speed of the user 1 is 1 MB /
s. Further, the buffer capacity is 100 KB according to FIG. When the expected response time is calculated based on the above value and the disk configuration table 121, the expected response time of the user 1 is 100 KB / 1 MB / s− (100 KB / 10 MB / s
+4 ms + 16 ms) = 70 ms. Moreover, according to FIG. 13, the response time of the user 1 is 100 ms. As a result, the expected response time is calculated as 100 ms- (16 ms + 4 ms) = 80 ms. In this embodiment, the smaller value is the expected response time. Therefore, the expected response time of the user 1 is set to 70 ms. Similarly, when the expected response time is calculated for the user 2 and the user 3, the transfer speed for the user 2 is calculated to be 20 ms, and the response time is 1
It is required to be 80 ms. Therefore, the expected response time of the user 2 is set to 20 ms. For user 3, the expected response time of 20 ms is calculated from the transfer rate,
The expected response time is 19 ms from the response time. Therefore, the expected response time of the user 3 is set to 19 ms. In this way, the expected response time can be calculated based on each of the transfer rate and the response time, and either one can be selected and set in the timeout table. In addition, in the third to fifth embodiments described above, the storage unit having the mirror configuration is taken as an example, but as in the second embodiment, the RAI is used.
Even with the D configuration, the expected response time can be obtained from the transfer rate and the response time.

In the first to fifth embodiments, the value set in the timer A is referred to the value in the timeout table,
It was setting the obtained value. However, the timer value A may be changed depending on the data buffer capacity. In this case, the buffer capacity table 123 stores the value of the buffer capacity that can be used for each user, as well as the free buffer capacity, as shown in FIG. The disk controller 120 sets and updates the free buffer capacity. The disk controller 120 refers to the data buffer 105 at an arbitrary timing, investigates the free buffer capacity for each user, and sets and updates the free buffer capacity in the buffer capacity table 123. The disk controller 120 determines the free buffer capacity of the buffer capacity table 123, the speed at which data is transferred from the data buffer 105 to the host, and the data buffer 1 from the storage unit 100.
The expected response time is calculated based on the transfer rate of data to 05.

As described above, by constructing the system as in the first to fifth embodiments,
Since the time from I / O issuance to response can be specified, the maximum response time can be guaranteed. In Japanese Patent Laid-Open No. 2-81123 in Conventional Example 2, since the data delay is detected and the data is restored, the response time cannot be guaranteed. Further, according to the present invention, the data buffer for maintaining the data transfer to the host allows the data transfer to be continuously performed even while the data is being repaired. In Japanese Patent Laid-Open No. 2-81123, which is a conventional example, data transfer is interrupted during data delay detection. In addition, as a conventional example, JP-A-2-811
In No. 23, parallel data transfer is assumed. In the disk device according to the present invention, the NRZ (nonreturn)
n-to-zero code) is connected to the hard disk controller by data, does not have a data buffer inside the disk, and when I / O command from the hard disk controller moves to the target cylinder or hard disk, data transfer is performed. Once started, some sector error determination (ECC) was performed by the hard disk controller after data transfer. JP-A-2-8
In No. 1123, the data transfer speed deteriorates at the start of repairing by error determination after data transfer. Therefore, the delay of data transfer is detected as a means for detecting a failure prior to the error determination by the ECC. Japanese Unexamined Patent Publication No. 2-81123 is effective only in a system (parallel data transfer disk) in which data transfer from a disk is carried out in synchronization with a plurality of disks. However, in the method disclosed in Japanese Patent Laid-Open No. 81231/1990, the time delay of the data delay detection timer is set as follows: minimum time: time for preventing misdetection of synchronization deviation For example, if 1% synchronization deviation is allowed at 3600 rpm, ts is ts min. = 16.6 ms × 1% = 166 μs Maximum time: Time until the processing start of the next sector in which the synchronization deviation is detected (If this time is exceeded, a sector synchronization deviation with another disk will occur and a data transfer delay of one rotation will occur. For example, assuming that there are 60 sectors at 3600 rpm, ts max. = 16.6 ms / 60 = 276 μs, and ts can be set to about 200 μs, but the correction of the failed sector affects the next sector (because the data buffer from the failed disk is empty), so that data is continuously written. There is a data reliability problem (a single sector failure affects subsequent operation). In order to avoid this, it is necessary to shorten the data delay detection time, but this is impossible due to the problem of erroneous detection of synchronization deviation. From the above, it is impossible to implement the failure repair by detecting the data delay, and there is a problem in reliability. Moreover, the application is limited to the parallel transfer disk.

Embodiment 6 FIG. In this embodiment, a plurality of disk devices in which a group of multimedia is recorded in one disk device and a redundant disk device in which redundant data of a plurality of multimedia data recorded in the plurality of disk devices are recorded are provided. Furthermore, in a control device that accesses the above-mentioned plurality of disk devices and redundant disk devices as a RAID system, when a new disk device is added to the expansion port, the redundant data recorded in the disk device and the new data are recorded in the new disk device. Calculate new redundant data using the multimedia data
The redundant data recalculating means for recalculating the redundant data recorded in the redundant disk device will be described below.

FIG. 16 is a system configuration diagram showing the system configuration of the present invention. In FIG. 16, RAI
The D control device 213 is a port 2 for connecting the disk device.
A plurality of 14 are provided. The disk device 210 is connected to the port 214. The disk device 210 stores one set of multimedia data. The redundant data of the multimedia data stored in the plurality of disk devices 210 is stored in the redundant disk device 211 and connected to the port 214. The plurality of disk devices 210 and the redundant disk device 211 form a RAID system 212. Further, the RAID control device 213 includes a redundant data recalculation unit 215 and an access prohibition unit 216.
The redundant data recalculation unit 215 uses the RAID system 21.
When the disk device 210 is newly added to the second disk unit 2, the redundant disk device 21 among the multimedia data stored in the newly added disk device 210 and the disk devices which constitute the RAID system 212 from the beginning.
It is a means for newly calculating redundant data from the redundant data stored in 1. Also, access prohibition means 216
When an error occurs in the disk device 210 constituting the RAID system 212, is a means for controlling so that the RAID control device 213 does not access the disk device in which the error has occurred.

FIG. 17 is a diagram for explaining a method of calculating redundant data. FIG. 18 is a diagram for explaining the addition of disk devices. FIG. 19 is a diagram for explaining the addition of disk devices.

The RAID system 212 shown in FIG.
It is assumed to be configured with ID level 4. In FIG.
A specific connection example between the disk device and the disk control device is shown. As shown in FIG. 18A, the RAID control device 213 has three disk devices 210 (# 0, # 1, # 2).
And the redundant disk device 211 (#P) are connected. Then, as shown in FIG. 18B, the disk device 21
0 (# 3) is newly added to the RAID control device 213. The expansion of the disk device is performed by connecting to the port 214 provided in the RAID control device 213. After the disk is added, the redundant data recalculation unit 215 reads the data (D3) of the added disk device and also reads the old parity (DPO) from the redundant disk device 211. Then, the redundant data recalculating means 215 calculates new redundant data (DPN) based on the above data. For calculation of redundant data, refer to FIG. According to FIG. 17, in the RAID system, as shown in FIG. 17A, each disk #k (k = 0,
, 2, ...) Based on the data Dk (k = 0, 1, 2, ...) At the same address stored in the disk, the exclusive OR is 0. In FIG. 18A, 4 of # 0, # 1, # 2 and #P (parity disk) are used.
The two disk devices 210 and 211 are connected. When the data stored in each of the disk devices 210 is used as D0, D1, D2, DPO (parity data) and the exclusive OR is calculated based on these data, it becomes 0 as shown in FIG. 17B. Then, as shown in FIG. 18B, when the disk device 210 of # 3 is added (# 3
It is assumed that the disk device 210 stores the data D3. ), The equation of FIG. 17 (C) must hold. In the expression of FIG. 17C, # 0 to # 3 and #P
Based on the data D0, D1, D2, D3, DPN stored in each disk device 210 and disk device 211 of the After adding 210,
The newly requested parity data is shown. The new parity data is stored in the #P disk device 211. ). From the equations in FIG. 17 (B) and FIG. 17 (C) above,
The formula of FIG. 17D is derived. As a result, the new parity data DPN is obtained by combining DPO (parity data before adding the disk device 210 of # 3) and D3 by exclusive OR. The redundant data recalculation unit 215 uses the DP as shown in FIG.
The new parity data DPN is calculated by combining O and D3 by exclusive OR.

The redundant data recalculating means 21 described above
5 is provided in the RAID controller 213, as shown in FIG. The redundant data recalculation means 215
It may be outside the RAID system. FIG. 19 shows a case where the redundant data recalculation unit 215 is outside the RAID system. Redundant data recalculation means 215 is RA
If it is outside the ID system, new parity data is written in a new disk device for another parity data, and a disk device is added. And replace.

The case where the redundant data recalculating means 215 is outside the RAID controller 213 will be described below with reference to FIG. According to FIG. 19A, the RAID controller 2
13 includes disk devices 210 of # 0, # 1, and # 2,
The # P1 disk device 211 for storing parity data is connected. When the # 3 disk device 210 is newly added to the RAID control device 213, the new parity data DPN must be obtained in advance by the redundant data recalculation means 215. Therefore, FIG.
As shown in (A), the redundant data recalculation unit 215 stores the old parity data in the dual port # P1.
The disk device 211 of # 3 is connected, and the disk device 210 of # 3, which is to be added, and the disk device 211 of # P2 for storing new parity data are also connected. After the connection, the redundant data recalculation unit 215 reads the old parity data (DPO) from the disk device 211 of # P1 in which the data D3 and the old parity data are stored from the disk device 210 of # 3. Then, an exclusive OR is calculated based on DPO and D3 to obtain new parity data DPN. The DPN is written in the new parity data storage disk device 211 of # P2. This completes the creation of new parity data due to the addition of disk devices. After that, as shown in FIG. 19B, the RAID control device 213 has the disk device 210 of # 3 and the disk device 211 of # P2 in which new parity data is stored.
Connect. The disk device 211 of # P1 in which the old parity data is stored is the RAID controller 2
Separate from 13. Then, the disk device 211 of # P1 is stored in preparation for the next expansion of the disk device.

The method in which the RAID controller 213 described above is provided with the redundant data recalculation means 215 is
Once the redundant data recalculating means has been incorporated into the ID system, no special resource is required. Therefore, it is advantageous in terms of cost. However, it is necessary to access the disk device from the host system and rewrite the new parity data in parallel. Therefore, the access performance is temporarily reduced. Further, for example, until the rewriting of the new parity data is completed, the data cannot be restored even if one disk device fails. Further, in the method in which the redundant data recalculation means 215 exists outside the RAID control device 213,
The parity data will be updated at the same time when the disk device is added. Therefore, it is possible to prevent the system performance and reliability from being degraded. However, there is an economical demerit that a means for calculating and writing new parity data must be provided separately from the RAID control device and that a separate parity disk for replacement must be prepared separately. is there. A user who uses this system can select one from two methods according to the purpose of use.

Further, in the system configuration as shown in FIG. 16, the video A is stored in the disk unit 210 of # 0,
The video B is stored in the # 1 disk device 210, the video C is stored in the # 2 disk device 210, and the parity data of the data stored in each disk device is #P.
Stored in the redundant disk device 211 of, for example, even if a failure occurs in the disk device 210 of # 1, unless the user accesses the data of video B, the disk device 210 of # 1 is not accessed. , It does not reduce the performance efficiency of the system. Further, even if the user tries to access the data of video B, the redundant data stored in the redundant disk device 211 of #P,
The data of video B can be reproduced from the data stored in the disk devices 210 of # 0 and # 2. Therefore, the user can access the video B data even if the # 1 disk device 210 fails.

As described above, in this embodiment, a group of multimedia data is recorded in one disk device. For this reason, even if a failure occurs in another disk device, if the data stored in the failed disk device is not accessed, the necessary multimedia data will not be reproduced without performing the data reproduction process. Can be obtained. In addition, the control unit was equipped with an additional port. Therefore, new multimedia data can be added to the system by adding a new disk to the expansion port. Further, redundant data recalculating means is provided outside the control device or the control device. Therefore, even if new multimedia data is added to the system, redundant data can be easily calculated.

Embodiment 7 In this embodiment, FIG.
The access prohibition unit 216 included in the RAID control device 213 of No. 6 will be described. Access prohibition means 21
Reference numeral 6 is a means for disabling access to the multimedia data stored in the failed disk device 210 when the disk device 210 connected to the RAID control device 213 fails.

For example, the access prohibition means 216 judges how many disk devices the read command is continuously received, and judges whether the access is large or small. There are several possible methods for determining how many disk devices the read command has arrived. For example, the following three methods are possible. First, the first method analyzes which disk device the command currently queued is. As a second method, it is checked which disk device the command of a certain number of times (20 times, etc.) processed in the past was for. Furthermore, as a third method, the past fixed time (1
Check which disk the command processed for 0 seconds etc. is for. Any one of the above three methods
As a result of judging by one of the two methods, for example, if there is access to three or more disk devices, “there is a lot of access”
Therefore, if the number is two or less, the processing is switched as "low access". If it is determined that the number of accesses is small, the data stored in the failed disk device is restored from the data stored in another disk device to reproduce the video data. Further, when it is determined that “there is a lot of accesses”, restoring the data stored in the failed disk device reduces the efficiency of the system. Therefore, until the number of accesses decreases (the above three methods are used to determine whether the number of accesses is low)
The video data stored in the failed disk device is rendered unreproducible, and the access prohibition means 216 prohibits access to the failed disk device and does not perform data restoration processing. Data stored in another disk device can be reproduced without any trouble. In this way, the access prohibition means makes it possible to limit the effect to only one video when the disk device fails.

As described above, the access prohibition means is
Prohibits access to multimedia data recorded on the failed disk device. Therefore, even if a failure occurs in the disk device, the reproduction process using the redundant data does not occur.

In the multimedia recording system of the present invention described in the above-described first to seventh embodiments, unlike the time-out used when a failure occurs, the time for ending the data access is recorded. It has a timeout table. The time to end the access to this data is the time that must be kept in order to effectively output or display multimedia data such as image data and audio data. If the access does not end even after the time for ending the access of this data has passed, it should be necessary for outputting or displaying the multimedia data by reproducing and outputting the data using redundant data. Continue the transfer process.

The multimedia recording system of the present invention is applied to a RAID system.

The time-out table stores a time-out time for each user, and the time can be set according to the type of multimedia data required by the user and the degree of request.

The time-out period is determined based on the data transfer rate required by the user.

The time-out period is determined based on the response time required by the user.

The time-out period is determined based on the data transfer rate and response time required by the user.

Further, the control device accesses both the data recording section and the redundant data recording section at the start of data access. As a result, as described above, the process of reproducing and outputting the data using the redundant data can be speeded up.

The time-out period is determined based on the free area of the data buffer area.

Further, the multimedia recording system of the present invention constitutes a RAID system by a plurality of disk devices. Each disk unit has one movie and one
One video data is stored as a group of multimedia data. When the user requests movie or video data, the data requested by the user can be provided by accessing one disk device. If a set of data is distributed and recorded on multiple disks and a failure occurs in a certain disk device, the data recorded on the failed disk will be restored, It must be played with the data. The reproduction process using the redundant data puts a heavy load on the system. In addition, when a group of multimedia data is distributed and recorded in a plurality of disk devices, if a certain disk device fails, all multimedia data must be accessed by a reproduction process using redundant data. . This redundant data reproduction process reduces the efficiency of the system. Since the multimedia recording system of the present invention records a group of multimedia data on one disk, even if one disk device fails, the multimedia data recorded on another disk device is redundant data. It can be provided to the user without performing any reproduction processing.

Further, the multimedia recording system of the present invention has an extension port, and a new disk in which a set of multimedia data is recorded can be connected to the extension port. The redundant data recalculation means recalculates new redundant data by using the data and the redundant data of the newly added disk device. As described above, the multimedia recording system of the present invention enables recording of new data without performing any distributed recording on a plurality of disk devices when recording new multimedia data.

Further, the multimedia recording system of the present invention can prohibit access to the multimedia data recorded in a disk device when a failure occurs in the disk device. By prohibiting the access, it is not necessary to perform the reproduction process of the multimedia data recorded in the failed disk device, and the system load is not increased. When a set of multimedia data is distributed and recorded in a plurality of disk devices, the reproduction process must be performed by using the redundant data. By recording in one disk unit, if a failure occurs,
The simple processing of prohibiting access to the multimedia data recorded in the disk device in which the failure has occurred makes it possible to provide the system efficiency as before.

[0083]

As described above, according to the present invention, the time for ending the access of the multimedia data is stored, and when the access is not ended within that time, the redundant data is used for the data. Is generated, multimedia data can be reliably accessed in time.

Further, according to the present invention, multimedia data can be stored in the RAID system.

Further, according to the present invention, a time-out time can be set for each user, and a flexible time-out process can be performed.

Further, according to the present invention, since the time-out time is set based on the data transfer rate, the buffer can be used efficiently.

Further, according to the present invention, the time-out time is set based on the response time requested by the user, so that it is possible to provide the service according to the strength of the user's request.

According to the present invention, the smaller one of the time-out time based on the data transfer rate and the time-out time based on the response time requested by the user is set as the time-out time. Therefore, it is possible to efficiently use the buffer and provide a service according to the strength of the user's request.

Further, according to the present invention, the control device acquires the data at the start of access and simultaneously acquires the redundant data. Therefore, the data reproduction performed using the redundant data can be immediately executed.

Further, according to the present invention, the time for timeout is set based on the free area of the data buffer area. Therefore, the buffer can be used efficiently.

Further, according to the present invention, since a group of multimedia devices are recorded in one disk device, even if a failure occurs in another disk device, the multimedia is reproduced without performing the data reproducing process. The data can be played back.

Further, according to the present invention, new multimedia data can be added to the system by adding a new disk to the additional port.

Further, according to the present invention, since access to the multimedia data recorded in the disk device in which the failure has occurred is prohibited, even if the disk device fails, the reproduction processing by the redundant data does not occur.

[Brief description of drawings]

FIG. 1 is a system configuration diagram showing an example of a multimedia recording system according to the present invention.

FIG. 2 is a diagram illustrating an operation of a control device in the system configuration of FIG.

FIG. 3 is a flowchart showing a procedure of data reading processing in the multimedia recording system.

FIG. 4 is a flowchart showing a procedure of timeout processing in FIG.

FIG. 5 is a system configuration diagram showing an example of a multimedia recording system according to the present invention.

FIG. 6 is a circuit diagram of the data generating means in FIG.

7 is a diagram showing setting contents of a timeout table and a disk configuration table in the system configuration of FIG.

FIG. 8 is a diagram illustrating data reproduction by redundant data.

9 is a diagram showing an operation when the control device in FIG. 1 includes a transfer rate request management table.

FIG. 10 is a diagram showing setting contents of each table included in the control device of FIG. 9.

FIG. 11 is a diagram showing setting contents of a timeout table.

FIG. 12 is a diagram showing an operation when the control device in FIG. 1 includes a response time request management table.

13 is a diagram showing setting contents of a response time request management table in FIG.

FIG. 14 is a diagram showing an operation when the control device in FIG. 1 includes a transfer rate request management table and a response time request management table.

FIG. 15 is a diagram showing setting contents of a timeout table in FIG.

FIG. 16 is a system configuration diagram of a multimedia recording system according to the present invention.

FIG. 17 is a diagram illustrating reproduction of redundant data.

FIG. 18 is a diagram for explaining disk expansion according to the sixth embodiment.

FIG. 19 is a diagram for explaining disk expansion according to the sixth embodiment.

FIG. 20 is a diagram showing a conventional recording system for continuously transferring a large amount of data.

FIG. 21 is a block diagram showing an embodiment of a conventional parallel data transfer system and apparatus.

22 is a block diagram of a parity generator / data corrector in FIG. 21. FIG.

23 is a circuit diagram of the correction control unit 14 of FIG.

FIG. 24 is a timing chart illustrating the operation of the circuit diagram of FIG. 23.

FIG. 25 is a diagram showing a data recording state in a conventional RAID level 3.

FIG. 26 is a diagram showing a data recording state when one is added to the disk configuration in FIG. 25.

[Explanation of symbols]

100 storage unit, 100a data, 100b redundant data, 102 control device, 103 timeout table, 104 data generation means, 105 data buffer, 106 multimedia storage device, 107 communication line, 108 user, 109 communication line, 120 disk controller, 121 Disk configuration table, 1
22 transfer rate request management table, 123 buffer capacity table, 124 response time request management table, 13
0 host system, 131 host I / F controller, 1
32 channels, 133 SPC, 134 system control MPU, 135 switch, 136 buffer, 13
7 data recovery circuit, 138 switch, 210 disk device, 211 redundant disk device, 212 RAID
System, 213 RAID control device, 214 port, 215 redundant data recalculation means, 216 access prohibition means.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kazuhiko Ito 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Sanryo Electric Co., Ltd. (72) Sei Sei Yamamoto 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Inside Ryo Electric Co., Ltd.

Claims (11)

[Claims] 1. A multimedia recording system having the following elements: (a) a recording unit for recording data and redundant data; (b) a control device for accessing the data recorded in the recording unit, If the access does not end even after the time stored in the time-out table and the time-out table, the redundant data recorded in the recording part is recorded. A control device comprising: a data generation unit that accesses and generates data. 2. The recording unit is a RAID (Redundant Arrays of Inexpensive Disks) system composed of a plurality of disk devices, and the control device is a RAID controller. Recording system for multimedia. 3. The multimedia recording system according to claim 1, wherein the time-out table stores a time for each user who requests data access. 4. The multi-function system according to claim 3, wherein the control device further sets a time to be stored in the timeout table based on a data transfer rate required by a user who requests data access. Recording system for media. 5. The multimedia according to claim 3, wherein the control device further sets a time to be stored in the timeout table based on a response time required by a user who requests data access. Recording system. 6. The control device further sets a time to be stored in the time-out table based on a response time and a data transfer rate required by a user who requests data access. 3. The recording system for multimedia as described in 3. 7. The recording section comprises a data recording section for recording data and a redundant data recording section for recording redundant data, wherein the control device accesses the data recording section at the start of access to the data and The recording system for multimedia according to claim 1, wherein the recording unit is accessed. 8. The control device further comprises a data buffer area for temporarily storing data obtained by accessing the recording unit and redundant data, and the control device is provided in an empty area of the data buffer area. The multimedia recording system according to claim 1, wherein the time to be stored in the time-out table is set based on the time-out table. 9. A multimedia recording system having the following elements: (a) a plurality of disk devices that record a group of multimedia data in one disk device; and (b) a plurality of disk devices that record in the plurality of disk devices. A redundant disk device for recording redundant data of multimedia data, (c)
A control device comprising a connection port connecting the plurality of disk devices and the redundant disk device, and accessing the plurality of disk devices and the redundant disk device as a RAID system. 10. The control device further comprises an extension port for adding a new disk device in which a group of multimedia data is recorded, and when the new disk device is added, the control device records in the redundant disk device. 6. Redundant data recalculating means for calculating new redundant data using the redundant data thus created and multimedia data recorded in the new disk device and recording it in the redundant disk device. 9. The recording system for multimedia according to item 9. 11. The control device further disables access to the multimedia data recorded in the failed disk device when any one of the plurality of disk devices fails. 10. The multimedia recording system according to claim 9, further comprising an access prohibiting unit.