JPH11126136A - Transfer method and dividing method for data, and information storage controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make efficiently usable an information storage device which is different in access time and storage capacity by making isochronal packets and divided data packets included in one cycle and periodically performing a packet transfer. SOLUTION: A cycle start CS, a header H, and data D1, D2, and D3 between information storage devices are transferred in this order from the head of one cycle and are transferred periodically (A). For example, when the access times of three information storage devices are 1:2:2 as a proportion between them, one packet to the information storage device 1, two packets to the information storage device 2, and two packets to the information storage device 3 are transferred (B). When the storage capacity of the three information storage devices are 2:2:1 as a proportion, on the other hand, two packets to the information storage device 1, two packets to the information device 2, and one packet to the information storage device 3 are transferred in one cycle (C).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing system such as a disk array, and more particularly to a method of transferring data in an information processing system, and a method of dividing data when performing periodic and isochronous packet transfer. And an information storage control device for controlling an information storage device constituting the information processing system.

[0002]

2. Description of the Related Art A disk array divides data into a plurality of disks and writes or reads them all at once to obtain high-speed performance, and enables data recovery even when a disk is broken. This achieves high reliability.

A disk array is a RAID (Redun
dant Arrays of Inexpensiv
e Disks). RAID has six levels from RAID 0 to RAID 5, depending on how data is divided (also called striping) and how data is restored when a disk is broken.

[0004] In RAID 0, data is divided into blocks each consisting of a plurality of bytes. In RAID1, disk mirroring is performed. Two disks are prepared, and the same contents of one disk are written to another disk, and when the disk is broken, data is restored from the other disk.

In RAID 2, data is divided into bits and a hamming code, which is a correction code, is added to the data. The hamming code is distributed and stored on a plurality of disks. In RAID3, data is divided into bits or bytes, and parity is added to the data. The parity is stored on a fixed disk.

In RAID 4, data is divided into blocks each having a plurality of bytes, and parity is stored in a fixed disk. In RAID5, parity is distributed and stored on a plurality of disks.

[0007] Of these, RAID 5 is most frequently used recently. RAID5 will be described with reference to FIG. As shown in FIG. 8, when there is an input / output request from the OS, a certain data string is divided into a plurality of striping sets 1 to n, and the divided data is sent to each disk by the disk array controller. That is, it is stored in the disk array device.

In RAID 5 of FIG. 8, data and parity are dispersedly stored on four disks. Stripes 1 to 3 are stored on disks 1 to 3, respectively, and stripes 1 to 3
Are stored in the disk 4.

[0009] Stripes 4 and 5 are on disks 1 and 2 respectively, and stripe 6 is on disk 4
Then, the parity corresponding to stripes 4 to 6 is stored in the disk 3. Data of stripe 7 and below is stored in the same manner.

The generation of parity and the buffering of data are performed by the disk array controller. In the disk array controller, these processes may be performed mainly by hardware with a dedicated CPU mounted, or may be performed only by software.

Generally, when the processing is performed mainly by hardware, high-speed processing is possible, but the price is high. On the other hand, if the operation is performed by software, dedicated hardware is not required, so that the cost is not high. However, the performance is not so high as compared with the case of mainly hardware.

[0012]

Recently, for example, HD
Digital data for high-resolution display devices such as TVs require a high-speed transfer rate of about 300 [Mbps], and large-scale and high-speed writing and reading of digital data in real time using a disk array. Is performed.

FIG. 9 shows an example of a conventional RAID configuration. FIG.
In a disk array controller, the time required for positioning a magnetic head for writing or reading data to a disk, that is, the access time is uniform for one disk array controller, and the disk capacity for storing data is the same. -A plurality (for example, three) of array devices are arranged and connected.

As shown in FIG. 9, in the interface between the disk array control device and the disk array device, a maximum of 320 [Mbps] is transferred in the case of ultra-wide SCSI, and the maximum is also transmitted to each disk array device. A data transfer speed of 320 [Mbps] is obtained.

As described above, in the disk arrays ranging from RAID0 to RAID5, high reliability and high performance have been realized by using a plurality of disk array devices, ie, hard disk devices, having the same access time and disk capacity.

However, if the data read time and the data write time vary among the hard disk devices,
There is a problem that buffer data overflows and underflows occur, and there is a problem that hard disk devices having the same performance such as access time must be arranged.

Further, when there is a difference in the disk capacity between the hard disk devices, the capacity of the configured hard disk devices is adjusted to the smallest one, and there is a problem that the larger the disk capacity, the more space is created.

Further, even if the disk is "inexpensive", cost becomes a problem if a plurality of hard disk devices are arranged. In addition, there is a problem that the system manufacturer knows the performance, and if the hard disk devices specified by the manufacturer are prepared, the models are limited and the cost is rather high.

On the other hand, in a multimedia communication device and the like, an IEEE 1 standard is used between a communication control device and a plurality of communication devices.
394 interface is considered,
Periodic and isochronous packet transfer is performed.

Here, one second is divided into, for example, 8000 times, and a packet for performing isochronous transfer called a cycle start at a period of 125 [microseconds] is used. Each communication device sets the time by a cycle start.

Following the cycle start packet, an isochronous data packet is transferred. At the beginning of the data packet, a command indicating the direction of writing or reading data and a header indicating the length of one packet are transmitted. Also, asynchronous data packets other than isochronous data packets may be transferred following isochronous data packets.

The packet transfer in the communication device will be described with reference to FIG. As shown in FIG. 10, from the beginning of one cycle, a cycle start (shown as CS in the figure), a header (shown as H in the figure), and data between each communication device (M1, M2, M3 in the figure). ) And asynchronous packets (shown as A1 and A2 in the figure) in this order.

FIG. 10 shows packet transfer between one communication control device and three communication devices.
M2 and M3 are packets respectively corresponding to three communication devices. By performing the packet transfer periodically and at the same time as described above, batch writing and batch reading are realized for a plurality of communication devices.

In view of the foregoing, the present invention provides a method of transferring data between an information storage device such as a disk array control device and an information storage device such as a plurality of disk array devices, a method of dividing data, and a method of dividing data. An object of the present invention is to relate to an information storage control device and to efficiently use information storage devices having different access times and storage capacities.

[0025]

The above object is achieved by a data transfer method, a data division method, and an information storage control device according to the present invention configured as follows.
According to the data transfer method of the present invention, an isochronous packet for adjusting the time of a plurality of information storage devices and data transferred between the information storage control device and each information storage device are transmitted to each information storage device. Correspondingly divided packets and 1
It is characterized in that it is included in a cycle and packet transfer is performed periodically.

Further, the data dividing method of the present invention is a data dividing method for performing periodic and isochronous packet transfer between an information storage control device and a plurality of information storage devices, The number of packets transferred in one cycle is allocated to each information storage device in an integer ratio of the reciprocal of the access time of each information storage device.

Further, the information storage control device of the present invention is an information storage control device for transferring packets to and from a plurality of information storage devices, wherein the number of packets transferred to and from each information storage device is determined. A speed determining means for allocating to an integer ratio of a reciprocal of an access time of each information storage device, and a packet for adjusting the time of a plurality of information storage devices, and a packet indicating a result allocated by the speed determining means, Isochronous transfer means for transferring a packet transferred to and from the information storage device every cycle.

Further, the data dividing method of the present invention is as follows.
A method of dividing data when performing periodic and isochronous packet transfer between an information storage control device and a plurality of information storage devices, wherein the number of packets transferred in one cycle is It is characterized in that an information storage device is allocated to an integer ratio of the storage capacity of each information storage device.

Further, the information storage control device of the present invention is an information storage control device for performing packet transfer with a plurality of information storage devices, wherein the number of packets transferred with each information storage device is determined. Capacity determining means for allocating to an integer ratio of the storage capacity of each information storage device; a packet for adjusting the time of the plurality of information storage devices; a packet indicating a result allocated by the capacity determining means; Isochronous transfer means for transferring a packet transferred to and from the device every period.

FIG. 1 is a diagram illustrating the principle of the present invention. In the packet transfer shown in FIG. 1A, an isochronous packet for synchronizing the times of a plurality of information storage devices, that is, a cycle start, and data transferred between the information storage control device and each information storage device are transmitted. A packet divided so as to correspond to the information storage device is included in one cycle, indicating that packet transfer is performed periodically.

The isochronous data packet transferred following the cycle start packet is write data sent from the information storage control device to each information storage device when writing to the information storage device. In the case of reading from a storage device, it is read data of each information storage device sent from each information storage device to the information storage control device.

At the beginning of the data packet, a command indicating whether to write or read data and a header indicating the length of one packet are transmitted. As shown in FIG. 1A, the cycle start (C
S), the header (H), and data (D1, D2, D3) between the information storage devices are transferred in this order, and these are transferred periodically.

FIG. 1A shows the packet transfer between one information storage control device and three information storage devices, and D1, D2, and D3 indicate the packet transfer to three information storage devices. Each packet corresponds to one packet.

As described above, the packet transfer between the information storage control device and the plurality of information storage devices is performed periodically and at the same time, thereby realizing batch writing and batch reading for the plurality of information storage devices. .

The (B) access time in FIG. 1 indicates the number of packets transferred per cycle between the information storage control device and a plurality of information storage devices based on the access time of each information storage device. A decision is made to indicate that a periodic and isochronous packet transfer will take place.

If the access times of the three information storage devices are T1, T2, and T3, and the numbers of packets transferred in one cycle are P1, P2, and P3, for example, P1 / P2 / P3 is (1/1). T1) vs. (1 / T2) vs. (1 / T3).

For example, assuming that the access times of three information storage devices are 8 [milliseconds], 4 [milliseconds], and 4 [milliseconds] in order, P1: P2: P3 is an integer ratio, and therefore, 1: 2: 2
Becomes Therefore, 1 for the information storage device 1 in one cycle
, Two packets to the information storage device 2, and two packets to the information storage device 3.

On the other hand, FIG. 1C shows the storage capacity of the packet transferred between the information storage control device and the plurality of information storage devices in each cycle based on the storage capacity of each information storage device. The number is determined, indicating that periodic and isochronous packet transfers occur.

For example, if the storage capacities of three information storage devices are C1, C2, and C3, and the numbers of packets transferred in one cycle are P1, P2, and P3, P1: P2: P3
Is C1: C2: C3.

For example, the storage capacities of three information storage devices are sequentially changed to 1 [GB (gigabyte)], 1 [GB], and 500 [M].
B (megabytes)], P1: P2: P3 is an integer ratio, and thus becomes 2: 2: 1. Therefore, in one cycle, two information storage devices 1 and two information storage device 2
And one packet is transferred to the information storage device 3.

FIG. 2 is a block diagram of the information storage control device of the present invention. 2 is an information storage control device, 2 is a speed determining means, 3 is a capacity determining means, and 4 is an isochronous transfer means. FIG. 2A shows an information storage control device 1 including a speed determination process 2 and an isochronous transfer unit 4, and FIG. 2B shows an information storage device including a capacity determination unit 3 and an isochronous transfer unit 4. The control device 1. The information storage control device 1 performs periodic and isochronous packet transfer with a plurality of information storage devices (not shown).

The speed determining means 2 distributes the number of packets transferred to each information storage device to an integer ratio of the reciprocal of the access time of each information storage device. The capacity determining means 3
The number of packets transferred to and from each information storage device is allocated to an integer ratio of the storage capacity of each information storage device.

The isochronous transfer means 4 includes a packet for adjusting the time of a plurality of information storage devices, that is, a cycle start, a packet indicating a result allocated by the speed determination means 2 or the capacity determination means 3, that is, a header,
Packets transferred to and from each information storage device, that is, the number of data packets allocated to each information storage device;
Is transferred every cycle.

As described above, by using the data transfer method, the data division method, and the information storage control device of the present invention, even when the access time and the disk capacity of the information storage device differ depending on the manufacturer and model, the information storage device can be used. Can be used efficiently as an information processing system such as a disk array.

[0045]

Embodiments of the present invention will be described in detail with reference to the drawings. FIG. 3 is a configuration diagram of an information storage control device, that is, a disk array control device, showing an embodiment of the present invention. 3 is a disk array controller of the present invention, 6 is a performance storage unit, 7 is a performance determination unit, 8 is a division / restoration processing unit, and 9 is a control unit.

The disk array controller 1 receives commands for writing and reading data from a computer such as a workstation or a personal computer (not shown), and controls a plurality of information storage devices, that is, a disk array device. , And transfers data to and from these disk array devices, and further transfers data to and from the computer.

The disk array controller 1 comprises a performance storage unit 6, a performance determination unit 7, a division / restoration processing unit 8, and a control unit 9. The performance storage unit 6, the performance determination unit 7, and the control unit 9 correspond to the speed determination unit 2 and the capacity determination unit 3 in FIG.
And the division / restoration processing unit 8 and the control unit 9 correspond to FIG.
And an isochronous transfer unit 4.

When the control unit 9 receives a command from the computer, the control unit 9 checks the access time and the storage capacity, that is, the disk capacity of the disk array device via the performance determination unit 7 and stores it in the memory in the performance storage unit 6. .

The performance determining unit 7 reads the access time and the disk capacity from the memory in the performance storage unit 6 prior to the data transfer for writing and reading to the disk, and transfers the packet to and from each disk array device. The number is determined and stored in a register in the performance storage unit 6.

The performance determining unit 7 determines the number of packets transferred to and from each disk array device as an integer ratio of the reciprocal of the access time of each disk array device or the disk capacity of each disk array device. It is decided to distribute to an integer ratio.

In the division / restoration processing unit 8, when a command to be written to the disc is issued, the data is divided into packets distributed by the performance determination unit 7 and the data is transmitted. When a read command is issued, Then, the performance determining unit 7 requests the data of the number of packets distributed to the disk array device, receives the data packet from each disk array device, and restores the data to the data before being divided.

In both the case of a command to write to the disk and the case of a command to read from the disk, the data transfer involves periodic and isochronous packet transfer. 1 cycle, 1
In the cycle, a cycle start packet for adjusting the time of a plurality of disk array devices, a header packet indicating a result allocated by the performance determining unit 7, and data of the number allocated to each disk array device A packet is generated and transferred to and from each disk array device.

FIG. 4 is a processing flow chart up to the performance determination of the disk array controller of the present invention. As shown in step S410 of FIG. 4, the number S of the disk array devices connected to the disk array control device 1 is first set. Subsequently, as shown in step S412, i = 0 is set and the process proceeds to step S510.

In step S510, a disk is actually accessed to obtain the access time and disk capacity of each disk array device. Then, step S5
Proceed to 12.

In step S512, each disk
The access time is measured for the array device, the measurement result is stored in the memory of the performance storage unit 6 for each disk array device, and the process proceeds to step S514.

In step S514, each disk
For the array device, a comparison is made between the disk capacity obtained by actually accessing the disk and the disk capacity stored in the memory of the performance storage unit 6. After confirming the disk capacity in this way, the process proceeds to step S516.

In step S516, i + 1 is performed, and it is determined whether or not i + 1 has exceeded the number S of connections of the device set first. If i + 1 does not exceed S, the process returns to step S510, and if i + 1 exceeds S, the process proceeds to step S610.

In step S610, i = 0 is set, and the flow advances to step S612. In step S612, the performance is determined for the speed based on the access time of each disk array device, and the performance is determined based on the disk capacity.

Subsequently, the flow advances to step S614, where the determined performance value is stored in the register of the performance storage unit 6. Next, the process proceeds to step S616. In step S616, i + 1 is performed, and it is determined whether i + 1 has exceeded S. If i + 1 does not exceed S, step S612
And if i + 1 exceeds S, the process ends.

FIG. 5 shows the performance parameters determined by the processing flow of FIG. As shown in FIG. 5, in each disk array device, the ratio of the number of packets, that is, the performance ratio, for the speed and capacity is determined for each device number.

The device number of each disk array device is assigned to each manufacturer and model of the hard disk device, and is stored in advance in the memory of the performance storage unit 6 together with the access time and disk capacity of the disk array device for each device number. Have been.

Steps S512 and S in FIG.
At 514, when the disk device is actually accessed to confirm the access time and the disk capacity, in step S612, the number of packets to be transferred per period is proportionally distributed with respect to speed and capacity.

In FIG. 5, item numbers 1 to 3 and item numbers 5 to 6
The proportionality of the reciprocal of the access time of the five disk array devices is divided into 10: 10: 5: 8
Get pair 4. Similarly, the disk capacity is proportionally distributed to obtain a capacity ratio of 12: 10: 5: 6: 4.

FIG. 6 shows the packet transfer of the present invention. FIG.
6A shows the number of transfer packets in one cycle distributed for the speed ratio, and FIG. 6B shows the number of transfer packets distributed for the capacity ratio. In the figure, CS indicates a cycle start indicating the beginning of one cycle, H indicates a header, and D1 to D5 indicate packets for the disks 1 to 5.

FIG. 7 is a diagram for explaining the format of the header used in the packet transfer of the present invention, and is an example of the header. The header is transferred every cycle, following a cycle start, prior to the data packet,
Indicate what is in the data packet.

FIG. 7A shows the number of divisions, and FIG. 7B shows the format of a header for a command. FIG. 7A shows a header packet indicating the number of divisions. This header indicates the number of divisions in order from the top, and the block size is 51.
2 bytes, disks 1 to disk n (where n is the number of disk array devices to be connected)
The number of packets up to is indicated by a code.

FIG. 7B shows a header representing a command.
It is a packet, and the code indicates that this header indicates a command. Write to command code (wr
te), read, control (contro)
l), sense, and the like.

Incidentally, instead of using the header packet representing the command shown in the embodiment, the data packet for each disk array device is included, and the header is used only for time adjustment. No problem.

[0069]

As is apparent from the above description, according to the present invention, even when a RAID is constituted by an information storage device such as a disk having a different disk access time or a different disk capacity depending on the manufacturer or model, each disk or the like can be used. There is an effect that processing can be efficiently performed according to the access time and storage capacity of the information storage device.

[Brief description of the drawings]

FIG. 1 is a principle diagram of the present invention.

FIG. 2 is a configuration diagram of an information storage control device.

FIG. 3 Example

FIG. 4 is a processing flowchart.

FIG. 5 Performance parameters

FIG. 6 is a packet transfer of the present invention.

Fig. 7 Header format

FIG. 8 RAID 5

FIG. 9 is a RAID configuration example.

FIG. 10 Packet transfer

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 information storage device or disk array control device 2 speed determination means 3 capacity determination means 4 isochronous transfer means 6 performance storage section 7 performance determination section 8 division / restoration processing section 9 control section

Claims (5)

[Claims] 1. An isochronous packet for adjusting the time of a plurality of information storage devices between an information storage control device for controlling a plurality of information storage devices and the plurality of information storage devices, and an information storage control device. And a packet obtained by dividing data transferred between each information storage device so as to correspond to each information storage device in one cycle, and performing packet transfer periodically. Transfer method. 2. A data division method for performing periodic and isochronous packet transfer between an information storage control device that controls a plurality of information storage devices and a plurality of information storage devices, comprising: A data division method, wherein the number of packets transferred in a cycle is allocated to each information storage device in an integer ratio of a reciprocal of an access time of each information storage device. 3. An information storage control device for performing packet transfer with a plurality of information storage devices, wherein the number of packets transferred with each information storage device is determined by the access time of each information storage device. Speed determining means for distributing the information to the reciprocal of an integer ratio, and a packet for adjusting the time of the plurality of information storage devices, a packet indicating a result allocated by the speed determining means, and transfer between each information storage device And an isochronous transfer unit for transferring a packet to be transmitted every period. 4. A data division method for performing periodic and isochronous packet transfer between an information storage control device and a plurality of information storage devices, wherein the packet is transferred in one cycle. Allocating the number of data to each information storage device in an integer ratio of the storage capacity of each information storage device. 5. An information storage control device for transferring packets to and from a plurality of information storage devices, the number of packets transferred to each information storage device being determined by the storage capacity of each information storage device. Capacity determining means for allocating to an integer ratio; a packet for adjusting the time of the plurality of information storage devices; a packet indicating a result allocated by the capacity determining means; and a packet transferred between the information storage devices. An information storage control device, comprising: an isochronous transfer unit that transfers a packet and a packet every period.