JPH10232743A - Leading zone position determining method and disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce a total seek amount and to enable retry at the time of seek error by changing the position of leading zone from all the hard disk devices when recording data from that leading zone. SOLUTION: The recording of data from a leading zone A to a terminal zone H is started. Hard disk devices HDD 0, HDD 2, HDD 4... at the off- numbered positions of eight hard disk devices HDD set odd-numbered zone positions successively shifted one by one to (n/2-1)th one (including 0-order one) as leading zones. Hard disk devices HDD 4, 6... set even-numbered zone positions successively shifted one by one to the n/2-th hard disk device as leading zones. Thus, only the hard disk device HDD 0 becomes dull seek, total seek time can be shortened into about 1/3 to 1/4 and even when any data error appears, retry is enabled while securing timewise continuously.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a head zone position determining method when a plurality of hard disk devices are used in parallel, and a disk array device using the determining method. For details, set the zone to which data is written first for each disk array device, and make this head zone different depending on the hard disk device to reduce the total head acceleration when head seek is performed across zones. In addition, the seek operation time of the disk array device is reduced, and the size and weight of the device are reduced.

[0002]

2. Description of the Related Art A plurality of disk recording / reproducing devices such as hard disks are used to increase redundancy and ensure reliability, and at the same time, increase the data transfer rate to record digital data such as video and audio. It has been proposed to configure a device for reproduction.

For example, when the system is limited to a system for a broadcasting station, in a system that requires a high-quality image such as used in a CM, for example, a digital composite signal D2 has a baseband component of 120M.
It is required that a high transfer rate of about bps can be continuously secured.

In recent years, these systems have been constructed using a data recording / reproducing device called a disk array, taking advantage of the economics of hard disks and the high speed of access.

For example, in a video / audio data recording apparatus as shown in FIG. 11, a plurality of hard disk drives HDD
At the same time, video and audio data are distributed to the respective hard disk devices HDD0 to HDD7 and recorded, and at the same time, these data (0 to
7, 8 to 15,...) Are recorded at the same address of another hard disk drive HDD8.

By doing so, if any of the hard disk devices HDD fails, the original data can be restored from the parity data P, so that redundancy and reliability of the disk array device can be ensured.

In addition to these features, n units (for example, 8 units)
When data is recorded and reproduced on the data recording / reproduction hard disk device by striping, the hard disk device 1
A data recording / reproducing device (disk array device) with a transfer rate and recording capacity about 7 times the effective transfer rate per unit
Can be used.

FIG. 12 shows a schematic view of such a disk array device.

In the frame 10, a plurality of hard disk drives (caddies) H accommodating a plurality of hard disks are provided.
DDs (HDD0 to HDDi) are exchangeably housed. The replacement of the hard disk drive HDD is performed in units of caddies. Although not shown in the frame 10, the SCS
A board for an I interface circuit, a circuit board for a CPU or a buffer memory functioning as an array controller,
A hard disk drive HDD, an SPC circuit board as a SCSI protocol controller, and the like are accommodated, and a connector for inputting / outputting SCSI is mounted on the back side of the frame 10.

In addition, the power supply 11 and the circuit cooling fan 1
2 are provided, and the power supply 11 has a ritual configuration using two power supplies so that even if one power supply fails, it can be backed up.

FIG. 13 shows a general disk array device 20.
2 is a simplified circuit configuration of FIG. The data input / output interface has SCSI, and when recording data, the externally transferred data
PC circuit (SCSI protocol controller) 21
Is once stored in the buffer memory 22 and then striped by the data multiplexer 23. Thereafter, the data is temporarily written to the dedicated buffer memory (FIFO configuration) 24 of each hard disk drive HDD. These controls are performed by the data controller 26 based on a command from the CPU 25.

Thereafter, the data is further recorded on a plurality of cylinder-structured hard disks provided in the hard disk devices HDD0 to HDD7 via the SPC circuit 27.

At the same time as the data processing, a parity operation circuit 28 is provided in association with the data multiplexer 23, and in this example, the parity P for eight data in this example recorded in parallel is calculated. Parity data is temporarily stored in the buffer memory 29 and then SPC
A dedicated hard disk drive HDD 8 via a circuit 30
Is recorded at the corresponding address.

In the hard disk drive HDD, as shown in FIG. 14, 10 to 20 magnetic disks (four in this example) are arranged on the same rotating shaft (spindle) 33 so as to overlap each other. For example, a head arm 35 having a flying head 34 mounted on each of the front and rear surfaces 32 is provided. The plurality of head arms 35 can be moved synchronously in the disk radial direction by a head positioner (positioning device) 36, and data is recorded and reproduced on a predetermined track.

The format for the disk 32 includes, for example, a track T which is divided concentrically as shown in FIG. 15, and a sector S which is obtained by dividing the track T into a sector in the circumferential direction, and an address (address) is given in sector units. Data management.

Of the plurality of disks, a track position signal is recorded on one disk (for example, the uppermost dedicated disk 0) as shown in FIG. Such a servo system for obtaining position information with a dedicated head is called a servo surface servo. Since the head 34 on each magnetic disk 32 (disk 1 to disk n) is located on a track Ti formed on one cylinder, one set of these tracks Ti is called a cylinder.

Recording on the disk is performed in cylinder units in order from the outer track. For example, when all sectors on the upper surface track i of the uppermost disk 1 have been recorded,
Next, the head 34 is switched to the lower surface track i provided on the lower surface side of the same disk to record data. When recording on the upper and lower tracks i is completed, recording is performed on the lower disk. This is repeated in order, starting with the exhaustion of all sectors of the same track on the last disk n, and jumping to the cylinder one inner side of the current cylinder (see FIG. 17).

As described above, in the disk array device, when the recording or reproduction data of one hard disk drive HDD is broken, the data is reconstructed using the parity data of the parity hard disk drive HDD.
Then, a replacement process is performed on a substitute sector for a broken sector based on a command from a higher order, for example, when there is a margin for accessing the hard disk device HDD. This ensures data reliability.

[0019]

As described above, the writing of data to each hard disk drive HDD is performed as a unit of data up to the number of data hard disk drives used in parallel. When one hard disk drive HDD is used, data is simultaneously recorded in parallel at the same address of the data hard disk drives HDD0 to HDD7.

For example, when reading and writing data at address 0 of the outermost track and then reading and writing data at address N of the innermost track as shown in FIG. A full-stroke seek operation (shown by an arrow) is performed from the outer periphery to the innermost periphery.

In this seek operation, a certain probability (for example, 1
Since a data error occurs at a rate of not more than once every 7th power of 0), the seek operation needs to be performed again in this case. The retry requires at least the time required for one rotation of the disk, and at least 8.4 when the rotation speed is 7200 rpm.
It takes msec.

When accessing the area from the outermost circumference to the innermost circumference or from the innermost circumference to the outermost circumference, the full stroke seek usually takes 20 to 21 msec. On the other hand, when accessing an adjacent area as shown in FIG. 19, the access time is very short. Therefore, for example, it is necessary to request an operation of a time (10 msec) longer than the half-stroke seek by one retry count for all the hard disk drives HDD, and the overhead increases.

The probability of retry increases. For example, if the probability of a seek error is once every 10 7th power, it is worse by the number of hard disk drive HDDs. For example, when eight hard disk drive HDDs are used, 10 ^7/8 = 1. An error occurs once every 25 × 10 ⁶ .

On the other hand, when seeking to an adjacent track as shown in FIG. 19, the seek is completed within several msec. In this case, a sufficient time for retrying even if the seek fails is secured.

In general computer processing, data processing requires accuracy such as an error rate. However, since continuity of time is not required as data, the host waits for processing. be able to. Therefore, even if the transfer rate becomes slow due to the processing time due to the retry, there is no particular problem.

However, for video / audio data, continuity of data is required with respect to time. Therefore, if retry is performed during data processing, if data cannot be transferred within the time determined by the system, retry is performed. As a result, the video / audio information will be destroyed for the time that has been exceeded. As a result, image distortion and audio noise occur, resulting in a serious operation defect.

To avoid this, it is sufficient to secure a sufficient time (overhead time) for retrying, but doing so will lower the transfer rate and degrade performance such as image quality. It is conceivable to reduce the transfer rate per unit while securing the transfer rate of the entire system by increasing the number of hard disk drives HDD.
There are problems such as an increase in the cost of the device and an increase in running cost.

In addition, at the time of full stroke seek,
Since all the hard disk drives HDD0 to HDD8 simultaneously perform seek operations, the accelerations during head seek of the hard disk drive HDD itself and the adjacent hard disk drive HDD propagate with each other, and a large acceleration is applied to each hard disk drive HDD. Will join. When the acceleration increases, it affects the head control system, so that off-track is likely to occur, causing a retry operation at the time of seeking, and causing a recording / reproducing error. .

In order to avoid the influence of acceleration, there is a method of hardening the hard disk drive HDD or strengthening the structure of the array device main body. However, when the parts are made thicker to make the hard disk device heavy, the weight increases. However, since the size of a single unit increases, the size and weight of the apparatus are hindered.

Therefore, the present invention has been made to solve such a conventional problem, and a disk array device capable of retrying while reducing overhead time and securing data continuity without increasing the number of devices, and The present invention proposes a method for determining a head zone position that can be used for the above.

[0031]

According to the first aspect of the present invention, in order to solve the above-mentioned problem, when n hard disk drives are used in parallel, the number of hard disk drives can be reduced. When a track is divided into n zones, one of the zones is set as a head zone, data writing is started from this head zone, and when all the hard disk devices perform a seek operation across the zones, the total head of the hard disk device is used. The plurality of leading zone positions are determined so as to minimize the acceleration.

A disk array device according to a second aspect of the present invention has n hard disk devices,
These tracks are divided into n zones, one of the zones is set as a leading zone, data writing is started from the leading zone, and the leading zones of the n hard disk devices are located at different zone positions. Is set.

According to the present invention, when constructing a disk array device using n hard disk devices, all the hard disk devices may start recording / reproducing from the outermost (or innermost) address as a start address. In order to minimize the acceleration due to the head seek operation of the disk array device, the head address position of each hard disk device (that is, the head address position of the head zone)
Are determined differently from each other.

By doing so, the head seek time to the target track is shortened, and the overhead time for retrying can be secured while ensuring the temporal continuity of data.

Further, even when the head is sought to the address in the same zone, the head seek amount and the head seek direction vary depending on the hard disk drive.
The total acceleration of the disk array device is drastically reduced. As a result, the size and weight of the hard disk device itself can be reduced, and the entire device can be reduced in size.

[0036]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment of a head zone position determining method according to the present invention and a disk array apparatus using the same will be described in detail with reference to the drawings.

According to the present invention, as described above, the start address, that is, the leading zone is changed for each hard disk drive HDD to be used, and the hard disk drive HDD is arranged so that acceleration due to head seek can be canceled as much as possible. , While improving data reliability while guaranteeing a high transfer rate.

FIG. 1 is a system diagram of a main part of a disk array device 20 constructed by a plurality of hard disk devices having such a leading zone.
However, in the example shown in the figure, a disk array is constructed using an even number of hard disk drives HDD including a parity hard disk drive. Therefore, there are seven hard disk drives HDD for data recording and reproduction.

In FIG. 1, the description of the same parts as in the prior art is omitted, but the CPU 25 is provided with a memory means (ROM) 32 having a built-in control program, and based on the built-in control prism, the hard disk drive H
Division of a zone (to be described later) for the DD, determination of a leading zone, allocation of an address (logical address) from the determined leading zone, and the like are performed.

Since the number n of hard disk devices HDD to be arrayed is known in advance from the size of the device, the zones to be divided are n zones. Here, for example, 3.
Uses a 4GB hard disk of 5 inches size and 330 commands per hard disk drive.
It is assumed that recording and reproduction are performed in sector units, and the transfer rate at that time has a margin of operation time of about 20 ms as compared with the case of transferring 330 ordinary sectors.

In the present invention, each hard disk drive H is determined based on the number n of hard disk drives HDD to be used.
The head zone of the DD is determined. The allocation of the leading zone is slightly different depending on whether n is an even number or an odd number. An even number (n = 8) of arrays will be described.

When n = 8, each zone is represented by symbols A to H as shown in FIG. 2, and if the leading zone is "A" and the trailing zone is "H", the leading zone A to the trailing zone are assumed. Data recording is started toward H. When the eight hard disk drive HDDs are numbered and represented as shown in FIG.
0, HDD2, HDD4,..., The odd-numbered zone positions sequentially shifted to the (n / 2-1) th (including the 0th) hard disk device are set as head zones, respectively. , Odd-numbered hard disk devices HDD4, HDD6,.
.. Indicates that even-numbered zone positions sequentially shifted for each unit up to the (n / 2) th hard disk drive are set as head zones. Then, the leading zone of the hard disk device which does not satisfy these conditions is selected as a zone position which is n / 2 zones away from the leading zone of the immediately preceding hard disk device.

Therefore, as shown in FIG. 2, the first zone A (A0) of the hard disk drive HDD0 is set to zone 1 which is the outermost peripheral zone, and the next odd-numbered hard disk drive HDD2 is set to third zone 3. (A2).

Since (n / 2-1) = 3, for the third and subsequent hard disk devices, (n / 2-1) = 3
In the odd-numbered hard disk drive HDD4 following 3, the even-numbered zone is selected as the first zone. Therefore, FIG.
In the fifth hard disk drive HDD4, zone 2 is selected as the first zone (A4). In the seventh hard disk drive HDD6, zone 4 is the first zone (A6).

Although n / 2 = 4, since there is no hard disk device corresponding to the next, the head zone is determined according to the following conditions. In other words, for a hard disk drive that does not satisfy the above conditions, a zone that is n / 2 away from the immediately preceding hard disk drive is selected as the head zone.

As a result, in the hard disk drive HDD1 shown in FIG. 2, zone 5 is selected as the first zone (A1), and in hard disk drive HDD3, zone 7 is set as the first zone (A1).
3), the hard disk drive HDD5 is in zone 6
Is selected as the first zone (A5), and zone 8 is selected as the first zone (A7) in the last hard disk drive HDD7.

Therefore, for example, as shown in FIG. 3, the seek amount when straddling the adjacent zone is set to “1”, and the acceleration generated by seeking across the adjacent zone is set to “1” (when the seek direction is on the inner circumferential side, “1”). If the seek is performed from address 0 to address N as shown in the figure, only the hard disk drive HDD0 will be full seek, and the total seek amount of the head will be "14". Incidentally, as shown in FIG. 18, the total seek amount from the address 0 to the address N when the head zone position is the same in all the hard disk devices is "56".

This means that the seek time up to the address N can be reduced to 1/3 to 1/4 of the conventional case, and as a result a sufficient overhead time can be secured.
Even if a data error occurs, it is possible to retry while ensuring temporal continuity of data.

The acceleration of the hard disk drive HDD0 is "+7", while the other hard disk drives HDD1 to HDD7 are all seeking in the outer peripheral direction, and the address N is in the adjacent zone. Is "-1".

FIG. 4 shows the relationship between the accelerations. As a result, the acceleration in the disk array device 20 seeks in a direction to cancel each other out, so that the total sum of the acceleration is theoretically zero. Incidentally, in the related art, the seek directions of all the hard disk devices HDD0 to HDD7 are the same, and the accelerations are all "7". Therefore, the entire device receives a considerably large acceleration.

When the hard disk drive HDD0 is half-seeked, the seek amount reaches the address N / 2 as shown in FIG. 5, and in this case, it is necessary to seek to the zone E. The total seek amount at that time is “32”, which is the same as the conventional case. The sum of the accelerations becomes zero as is clear from the explanatory diagram of FIG. Since the conventional total acceleration is “32”, the acceleration received by the apparatus cannot be reduced even in the case of the conventional half seek.

FIG. 7 shows an example where the seek amount is less than half, for example, when the seek amount is set to "2". In this case, the total seek amount is "24", and as is apparent from the seek direction indicated by the arrow in FIG. 7, the heads move in directions in which the accelerations cancel each other, and the total acceleration at that time becomes zero. Obviously, the acceleration cannot be reduced to zero in the conventional configuration.

When a disk array device is constructed using an even number of hard disk devices, the seek distance can be reduced to n / [2 (n-1)] at worst. In addition, by reducing the seek amount, in the example shown in the figure, retry can be performed for seeks of four zones or less, thereby preventing a data transfer rate from being deteriorated due to a seek error.

FIG. 8 and subsequent figures show a case where an odd number of hard disk devices are used. Similarly to FIG. 2, the odd-numbered hard disk devices are up to the (n / 2−1) th (including the 0th) hard disk device. The odd-numbered zone position sequentially shifted for each vehicle is set as the head zone.

In the even-numbered hard disk devices following the (n / 2-1) -th, even-numbered zone positions sequentially shifted for each unit up to the n / 2-th hard disk device are set as head zones. . Then, the leading zone for the hard disk device that does not satisfy these conditions is selected as a zone position separated by (n / 2 + 1) zones from the leading zone of the immediately preceding hard disk device.

The example of FIG. 8 shows a case where the disk array device 20 is configured by using nine hard disk devices. FIG. 8 shows only the first zone in each of the hard disk devices HDD0 to HDD8. First, odd-numbered hard disk devices HDD0, HDD2, HDD
Up to 4, the odd-numbered zones 1, 3, and 5 are the leading zones A0, A2, and A4, respectively.

When the number of hard disk drives is odd,
Even-numbered hard disk devices are selected as the (n / 2-1) th and subsequent hard disk devices. This is different from FIG. This is because the odd number of hard disk devices HDD to be used does not include any hard disk devices for pairing.

Also, although n / 2 = 4, (n / 2-1)
Since there are only two even-numbered hard disk devices thereafter, only the hard disk devices HDD5 and HDD7 are even-numbered zone 2 and zone 4 is the first zone A5 and A5.
It becomes 7.

Then, the remaining hard disk drive HDD
In this example, the first zone A1, A3, A6, and 5 are separated from each other by five zones (hatched) so that the HDD1, HDD3, HDD6, and HDD8 are paired with the immediately preceding hard disk device.
Selected as A8.

FIG. 9 shows the relationship between the total seek amount and the acceleration when the head zone is set as described above. This example is also an example when the hard disk drive HDD0 performs a full seek from the position of address 0 to the position of address N. At this time, the total seek amount is “15”, which can be much smaller than the conventional total seek amount “56”, so that the retry time can be secured while maintaining the temporal continuity of data. Become.

Since the acceleration is theoretically zero as shown in FIG. 9, the acceleration applied to the disk array device can be drastically reduced.

FIG. 10 shows an example of a half seek operation. In this case, the total seek amount (= 40) is the same as the conventional case, but since the acceleration at this time can be made zero, the acceleration applied to the apparatus can be greatly reduced. .

[0063]

As described above, according to the present invention, when performing parallel processing using a plurality of hard disk devices, a track is divided into zones, and one of the divided zones is set as a head zone, and the head zone is used as a starting zone. In recording data, the position of the head zone is changed in all hard disk devices. According to this, since the total seek amount in all the hard disk devices can be reduced, it is possible to increase the probability that a retry can be performed when a seek error occurs. This can prevent the data transfer rate from deteriorating due to a seek error and improve the reliability of data recording and reproduction.

Further, in the disk array device according to the present invention, since the head zone differs for each hard disk device, the total seek amount can be reduced.
Since the acceleration due to the head seek can be offset each other so that the total acceleration is as small as possible,
The acceleration of the entire disk array device can be significantly reduced as compared with the conventional one.

By canceling the accelerations generated in the hard disk devices, the accelerations of the adjacent hard disk devices do not propagate. Thereby,
A large acceleration is applied to the hard disk device itself, and the head operation is affected by the acceleration, and the probability that the head control system malfunctions is reduced. Therefore, off-track is reduced, and seek error can be improved. As a result, recording and reproduction errors are improved, and the error rate and reliability of the entire apparatus can be improved.

Of course, the present invention is characterized in that the number of parts is not increased by making the entire apparatus robust, and the cost of parts is not increased, and the weight and cost can be reduced.

[Brief description of the drawings]

FIG. 1 is a system diagram of a main part showing an embodiment of a disk array device according to the present invention.

FIG. 2 is a diagram illustrating a relationship between a hard disk device and a leading zone when an even number of hard disk devices are used.

FIG. 3 is a diagram illustrating a relationship between a seek amount and an acceleration during a full seek.

FIG. 4 is an explanatory diagram of acceleration at that time.

FIG. 5 is a diagram showing a relationship between a seek amount and acceleration during a half seek.

FIG. 6 is an explanatory diagram of acceleration at that time.

FIG. 7 is a diagram illustrating a relationship between a seek amount and acceleration when a seek is shorter than a full seek.

FIG. 8 is a diagram showing a relationship between a hard disk device and a leading zone when an odd number of hard disk devices are used.

FIG. 9 is a diagram illustrating a relationship between a seek amount and an acceleration during a full seek.

FIG. 10 is a diagram showing a relationship between a seek amount and acceleration during a half seek.

FIG. 11 is a conceptual diagram of a disk array.

FIG. 12 is a schematic view of a disk array device.

FIG. 13 is a system diagram of a disk array device.

FIG. 14 is a configuration diagram of a hard disk device.

FIG. 15 is a plan view showing the relationship between tracks and sectors.

FIG. 16 is a diagram showing a relationship between a truck and a cylinder.

FIG. 17 is a conceptual diagram of a cylinder.

FIG. 18 is a diagram illustrating a relationship between a seek and a zone.

FIG. 19 is a diagram showing a relationship between a seek and a zone.

[Explanation of symbols]

20: Disk array device, HDD: Hard disk device, 25: CPU, 32: Memory means

Claims (5)

[Claims] 1. When n hard disk drives are used in parallel, a track of the hard disk drive is divided into n zones, and one of the zones is set as a head zone, and data writing is started from the head zone. When all the hard disk drives perform a seek operation across a zone, the plurality of head zone positions are determined so that the total head acceleration of the hard disk devices is minimized. Method. 2. A hard disk drive having n hard disks, wherein a track is divided into n zones, one of the zones is set as a head zone, data writing is started from the head zone, The head array of the n hard disk devices is set at different zone positions from each other. 3. The plurality of head zone positions are set so that the total head acceleration of the hard disk device is minimized when all the hard disk devices perform a seek operation across the zone. Item 3. The disk array device according to Item 2. 4. When n is an even number, the odd-numbered hard disk devices are odd-numbered hard disks sequentially shifted up to the (n / 2−1) th (including the 0th) hard disk device. The second zone position is set as the first zone, and the odd-numbered hard disk devices following the (n / 2−1) th are even-numbered zones sequentially shifted one by one up to the n / 2th hard disk device. Each position is set as a head zone, and a head zone for a hard disk device that does not satisfy these conditions is selected as a zone position n / 2 zones away from a head zone of a hard disk device located immediately before. Item 3. The disk array device according to Item 2. 5. When n is an odd number, the odd-numbered hard disk devices are shifted to the (n / 2−1) th (including the 0th) hard disk device by odd numbers sequentially shifted one by one. Each of the zone positions is set as the first zone, and the even-numbered hard disk devices following the (n / 2−1) th are even-numbered zones sequentially shifted one by one up to the n / 2th hard disk device. Each position is set as a head zone, and the head zone for a hard disk device that does not satisfy these conditions is selected as a zone position that is (n / 2 + 1) zones away from the head zone of the immediately preceding hard disk device. 3. The disk array device according to claim 2, wherein