JP2001125754A - Data access method for disk array - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a disk array device, with which an access response is accelerated. SOLUTION: A configuration for accessing a filing system recorded in a disk array device composed of plural disk devices is provided with a step 82 for knowing the leading logical address of a partition, which is the set of data handling units in the filing system, a step for knowing the leading logical address of the physical chunk of the disk device, a step S4 for finding a difference as offset when the logical addresses of these partition and chunk are not coincident, a step S5 for setting the sector of offset to the disk device, and a step for performing access by adding the sector length of offset set to the partition having designated data when data access is dessignated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data storage device that functions as a single storage device by using a plurality of storage means such as a group of magnetic disk devices.

[0002]

2. Description of the Related Art Conventionally, a plurality of storage means are provided to form one collective data storage device.
Array devices are known. The disk array device is a “RAID (Redundant Arrays)” devised by the University of California, Berkeley.
f Inexpensive Disks) architecture, and includes a plurality of magnetic disk devices (hereinafter, referred to as disks). The definition of the “RAID architecture” includes RAID levels 0 to RAI.
There are six types of D level 5. In the following description, a RAID level 0 disk array device whose configuration is shown in FIG. 10 will be described as an example.

First, a memory access operation from a computer will be described including its configuration. FIG. 13 shows an example of a disk array device that uses three disks marked (DK # 0 to 2) in the “RAID configuration diagram”. The host computer HST includes a disk array device composed of three disks shown in the figure and a host interface Hos.
Connected via t I / F. The respective disk devices DK # 0 to DK2 are controlled via disk control mechanisms CD # 0 to CD # 2.

In a disk array device composed of a plurality of disks and a disk array control mechanism, a disk address (hereinafter, referred to as a logical address) space viewed from a host computer HST is divided into a plurality of disks DK.
It is arranged to be distributed to the logical address spaces on # 0 to # 2. FIG. 11 shows an example of the relationship between the logical address (hLBA) viewed from the host computer and the logical address (dLBA) on the disk. For example, the host-side logical address hLBA # 04 is the logical address dLBA # 00 of the disk DK # 1. Logical addresses are allocated in units of sectors (512 bytes in FIG. 11). Here, the logical address on DK # 0-2 in the disk array device is dLBA as shown in FIG. 4B, and the logical address as seen from the host computer is as shown in FIGS. 4A and 4B. Let's call it hLBA.

[0005] The disk array device handles the logical address space on DK # 0-2 by dividing it by a certain size, and this is called a chunk. In FIG. 11, the chunk size is 4
It is a sector (2048 bytes). The logical address space as viewed from the host computer is divided (striped) by the size of the chunk by the disk array device and assigned to the logical address space on DK # 0-2. For example, in the example shown in FIG. 11, the logical addresses hLBA # 08 to 11 viewed from the host computer are DK # 2
The logical addresses are assigned to the upper logical addresses dLBA # 00 to 03.

Next, a general “software configuration diagram” shown in FIG. 12 will be described. The host computer has the following three types of software modules. An application (APP) 111; a file system (FS) 112; a driver (DRV) 113. The disk array device has the following three types of software modules. A host interface control unit (HCON) 121, a disk array control unit (RCON) 122, and a disk control unit (DCON) 123.

Next, the operation of each module when performing data access will be described. Software module on host computer a) Application (APP) Data is handled based on the file name (FLNM). FLNM
To request data access to the file system (FS). B) File system (FS) Using the file allocation table from FLNM, the cluster number where the file exists (CNUM)
/ Check the cluster length (CLEN). Here, the “file allocation table” is a means for associating a file name with a cluster. A "cluster" is a unit of data handled by the file system, and is determined separately from a chunk as shown in FIG.
In FIG. 13, the size of the cluster is 4 sectors (2048 bytes). CNUM / CLEN / cluster size /
The logical address (hLBA) / logical block length (hLBL) where the corresponding cluster exists is checked from the leading logical address of the partition. hLBA = CNUM × size of cluster + first logical address of partition (1) Here, “partition” is a set of clusters.
The file system handles data in the disk array device by dividing it into one or more partitions.
Request data access to the driver by designating hLBA / hLBL. C) Driver (DRV) Designates hLBA / hLBL and requests data access to the disk array device (host interface control unit).

Software module on the disk array controller d) Host interface controller (HCON) Requests data access to the disk array controller by designating hLBA / hLBL. E) Disk array control unit (RCON) The hLB / hLBL checks which disk in the disk array device is to be accessed (DKN), and checks the logical address (dLBA) / logical block length (dLBL) of the relevant disk. DKN / dLBA / dLBL
To request data access from the disk controller. F) Disk control unit (DCON) For the disk specified by the DKN, dLBA / d
Request data access by specifying LBL.

The "partition diagram" will be described with reference to FIG. As described above, a partition is a set of clusters. In the case of FIG. 14, the partition PRT # 0
0 is composed of clusters CLT # 00 to CLT # M. In the case of the figure, cluster CLT # 00 is sector L
It is composed of BA # 63 to LBA # 66. Generally, there is no regularity in the head address of the sector included in the cluster. Partition configuration information is stored in the boot sector. In the case shown in the figure, the boot sector LBA # 00 stores "the leading logical address of the partition PRT # 00 is LBA # 63".

[0010] When accessing the cluster CLT # 02, how to obtain the logical address of the sector will be described below. Step 1. Reads boot sector LBA # 00,
Know the starting logical address of the partition. Partition top logical address = LBA # 63 (2) Step 2. The sector length from the head of the partition to the head logical address of the cluster CLT # 02 is obtained by multiplying the cluster size of 4 sectors by the cluster number # 02 as shown in the following (3). 2. Position from start logical address of partition = 8 sectors (3) Step 3. Step 1. And the value obtained in step 2. Are added to obtain the start logical address of the cluster CLT # 02 as in the following (4). Start logical address of cluster CLT # 02 = LBA # 63 + 8 = LBA # 71 (4)

[0011]

In a conventional system of a host computer and a disk array device, since a cluster and a chunk are arranged without being aware of each other, the response to the access of the disk array device becomes slow. There are issues. That is, in FIG. 13, a case where the host computer simultaneously accesses the cluster # 00 and the cluster # 01 will be described. Data of cluster # 00 is DK # 0 (dLBA # 01-03)
And DK # 1 (dLBA # 00) and cluster # 01
Are DK # 1 (dLBA # 01-03) and DK #
2 (dLBA # 00), and both are DK # 1
Disk array device is cluster # 00
And access to cluster # 01 cannot be processed in parallel, and after the processing for access to cluster # 00 (or 01) is completed, cluster # 01 (or 00) is executed. Slows down.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and has a major object to speed up a response to an access from a host computer to a disk array device.

[0013]

SUMMARY OF THE INVENTION A data access method for a disk array according to the present invention is a method for handling data of a file system in a configuration for accessing a file system recorded on a disk array device comprising a plurality of disk devices. A step of knowing a head logical address of a partition which is a set of units; a step of knowing a head logical address of a physical chunk of a disk device;
If the logical addresses of these partitions and the chunks do not match, the difference is determined as an offset, and a step of setting an offset sector in the disk device. And adding the sector length of the set offset to access.

Still further, a driver under the file system of the host computer obtains the offset, and specifies the sector to be accessed by adding the sector length of the offset in the case of data access.

Still further, the host interface control unit of the disk array device obtains the offset, and specifies the sector to be accessed by adding the sector length of the offset in the case of data access.

Further, when the file system handles a plurality of partitions, after calculating the offset amount of the first partition, an offset is added to the last sector address after offset correction of the immediately preceding partition for each of the following partitions. Then, the offset amount of the partition is calculated.

[0017]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 A configuration and operation in which a response delay of disk access is reduced by devising the formation of a logical address will be described. The software configuration diagram in the present embodiment is the general one shown in FIG. That is, FS: file system on the host computer DRV: driver on the host computer On the other hand, the correspondence of the logical address arrangement on the disk of the file is "data arrangement diagram N1" shown in FIG. In FIG. 1A, hLBA '# xx: FS and DR on the host computer
Logical address space between V In FIG. 1B, hLBA # xx: Logical address space between the disk array device and the host computer. In FIG. 1C, dLBA # xx: Logical on the disk in the disk array device. Address space.

Thus, FIG. 1 shows the logical address space (hLBA '# xx) viewed from the FS (via the DRV),
The logical address space (hLBA # xx) viewed from the DRV (via the host interface) and the logical address space (dLBA #) on the disk in the disk array device
xx). The feature is that the driver (DRV) performs an offset indicated by S between FIGS. 1A and 1B. For example, the sector of the logical address hLBA ′ # 06 viewed from the FS is located at the logical address hLBA # 09 viewed from the DRV, and the logical address dLBA # 01 of the disk DK # 2 in the disk array device.
Is stored in

The operation of the high-speed system configured as described above will be described. First, the initialization operation of the disk array device will be described with reference to FIG. At the time of initialization of the disk array device by the host computer in step S1, F
S writes the start logical address (hLBA '# 01) information of the partition in a sector called the boot sector in the disk array device (normally, the start logical address (hLBA'# 00) of the disk array device). In step S2, the DRV obtains the leading logical address (hLBA '# 01) of the partition from the information in the boot sector. In step S3, the DRV determines the start logical address (hLBA '# 01) of this partition and the start logical address (hLBA #) of the chunk including that address.
00), if the leading logical address of the partition does not match the leading logical address of the chunk, the leading logical address (hLBA) of the next chunk is determined in step S4.
# 04, the start logical address (hL) of the offset with respect to
BA # 01) and sector length In this case, (3) is obtained and stored in step S5. The storage location is, for example,
It is setting information storage means on a disk in the array device.
If the head logical addresses match, the process proceeds to step S6.
This is the case when no offset is required, and does nothing.

Since an offset is inserted at the beginning of a partition, the data capacity (hereinafter, storage capacity) that can be stored in the disk array device is reduced by the size of the offset. For example, in the case of the “data layout diagram N1” in FIG. 1, the storage capacity of the disk array device is N + 1 sectors, but in FIG. 1, since the offset amount is 3 sectors, these 3 sectors are reduced and can be used from the FS. Storage capacity is N +
It is not one sector but N-2 sectors. Here, 3 is the size of the offset (sector length). The DRV reports the storage capacity of the disk array device accessed by the FS to the FS by reporting the storage capacity of the disk array device reduced by the amount of the offset from the storage capacity of the disk array device as the storage capacity of the disk array device. Restrict.

Next, the operation at the time of data access will be described with reference to FIG. When the FS requests access in step S11, the DRV checks the logical address (hLBA '# nn) requested by the FS in step S12, and if it is equal to or later than the start logical address (hLBA # 01) of the offset,
That is, if hLBA '# nn> = hLBA # 01, the logical address (hLBA'# nn) is determined in step S13.
And the start logical address (hLBA #) for accessing the disk array device by adding the offset sector length (3)
nn + 3 = hLBA '# nn + 3). In other cases, the logical address requested by the FS in step S15 is used as it is. In step S14, for example, the logical address (hLBA # nn + 3) obtained in S13
Is used to access the disk array device.

In the "data arrangement diagram N1" of FIG.
The case where S accesses cluster # 00 and cluster # 01 simultaneously will be described. The data of cluster # 00 is stored in DK # 1 (dLBA # 00-03)
01 is stored in DK # 2 (dLBA # 00-03). The data of clusters # 00 and 01 are stored on separate disks. Therefore, the disk array device is DK # 0 (dLBA # 00-03),
Access to DK # 1 (dLBA # 00-03) is processed in parallel.

As described above, simultaneous access to different clusters is generally performed in series, but in the present embodiment, two accesses can be processed in parallel. The response performance to access is improved. Also, the configuration is easy, since only the software (DRV) on the host computer is changed without changing the disk array device.

Embodiment 2 FIG. In the above-described embodiment, a case has been described in which the driver, which is software on the host computer, offsets the format of the logical address. In the present embodiment, a case will be described in which this is performed by a host interface control unit which is software on a disk array control device. In the software configuration diagram of FIG. 12, software on the disk array controller is composed of the following three components. HCON: Host interface control unit in the disk array device RCON: Disk array control unit in the disk array device DCON: Disk control unit in the disk array device

The "data layout diagram N2" in FIG. 4 shows the correspondence between the logical address space from the host computer and through the host interface control unit and the logical address space on the disk in the disk array device. FIG.
In FIG. 4A, hLBA # xx: logical address space between the disk array device and the host computer. In FIG. 4B, hLBA ′ # xx: logical address space between HCON and RCON. dLBA # xx: a logical address space on a disk in the disk array device. Thus, FIG. 4 shows the logical address space (hLBA # xx) viewed from the host computer (via the host interface) and the logical address space (hLBA '# xx) viewed from the HCON (via the RCON).
And the logical address space (dLBA # xx) on the disk in the disk array device. FIG.
The feature is that the host interface control unit (HCON) performs an offset indicated by S between (A) and (B). For example, the sector of the logical address hLBA # 06 viewed from the host computer is located at the logical address hLBA '# 09 viewed from the HCON, and
In the array device, the logical address dL of the disk DK # 2
It is stored in BA # 01.

The operation of the high speed system is as follows. First, the initialization operation of the disk array device will be described with reference to FIG. In step S21, at the time of initialization of the disk array device by the host computer, the host computer sets a partition in a sector called a boot sector in the disk array device (usually, the head logical address (hLBA # 00) of the disk array device). Is written at the start logical address (hLBA # 01).
In step S22, the HCON determines the start logical address (hLBA # 0) of the partition from the information in the boot sector.
Find 1). In step S23, the HCON determines the start logical address (hLBA # 01) of this partition,
The head logical address (hLBA '# 00) of the chunk including that address is compared. If the head logical address of the partition and the head logical address of the chunk do not match, in step S24, the head logical address (hLBA's) of the next chunk is determined. The start logical address (hLBA '# 01) and sector length of the offset with respect to'# 04) and the sector length In this case (3) are obtained and stored. The storage location is, for example, setting information storage means on a disk in the disk array device. If there is no offset, it is S26 and nothing is performed.

In order to insert an offset at the beginning of a partition, the amount of data that can be stored in the disk array device (hereinafter referred to as storage capacity) is reduced by the size of the offset as in the previous embodiment. is there. For this reason,
The case where the offset is subtracted from the storage capacity of the disk array device for reporting is the same as in the above embodiment.

The operation at the time of data access will be described with reference to FIG. In step S31, when the host computer requests access, in step S32, the HCON transmits the logical address (hLBA #) requested by the host computer.
nn) and the offset, and logical address hLBA #
If nn is equal to or later than the start logical address (hLBA '# 01), that is, if hLBA # nn> = hLBA'# 01, in step S33, the logical address (hLBA #
nn) plus the offset sector length (3) and RCON
Logical address (hLBA '# nn) for accessing
+ 3 = hLBA # nn + 3) is calculated. Otherwise, in step S35, the logical address requested from the host computer is used as it is. Step S34
Then, for example, the logical address (h
The access to the RCON is executed using LBA '# nn + 3). The case where the host computer accesses the cluster # 00 and the cluster # 01 simultaneously in the “data arrangement diagram N2” of FIG. 4 will be described. The data of cluster # 00 is DK # 1 (dLBA # 00-03)
The data of the cluster # 01 is DK # 2 (dLBA # 0).
0 to 03). The data of clusters # 00 and 01 are stored on separate disks. Therefore, the disk array device operates as DK # 0 (dLBA #
00-03) and access to DK # 1 (dLBA # 00-03) are processed in parallel.

[0029] Even in this case, as in the previous embodiment, two accesses can be processed in parallel without conflicting disks even in simultaneous access to different clusters.
This configuration can be easily performed only by changing the software (HCON) on the disk array without changing the host computer.

Embodiment 3 In the above embodiment, the file system using one partition has been described. In the present embodiment, a method of devising a logical address for reducing a response delay of disk access when a file system includes a plurality of partitions will be described. FIG.
FIG. 8 is a diagram showing how to give an offset when there are two partitions. The case where the number of partitions is expanded to three or more can be handled in the same way as the following two cases. In FIG. 7, the head logical address information of the partition is stored in the boot sector.

The operation in the present embodiment is as follows. First, the initialization operation will be described with reference to FIG. In the figure, the meaning of each symbol will be described along with the description of the operation. Obtaining the offsets # 00 and 01 for the two partitions PRT # 00 and 01 respectively is as follows. A) For the first partition PRT # 00 of the plurality, the start logical address of offset # 00 and the sector length are obtained. Here, pLBA # 0: the leading logical address of PRT # 00 cLBA # n: the leading logical address of the chunk including pLBA # 0 cLBA # n + 1: the leading logical address of the chunk next to cLBA # n oLBA # 0: offset # 00 first logical address oLBL # 0: Sector length of offset # 00.

In step 41, the leading logical address of offset # 00 is made equal to the leading logical address of PRT # 00. oLBA # 0 = pLBA # 0 In step 42, the leading logical address of PRT # 00 is
The head logical address of the chunk including the logical address is compared. In the case of coincidence: That is, in the case of pLBA # 0 == cLBA # n, in step 44, the sector length of the offset # 00 is set to 0. oLBL # 0 = 0 When they do not match: That is, pLBA # 0! = CLBA # n
In step 43, the sector length of offset # 00 is pL
It is assumed that pLBA # 0 is subtracted from the head logical address of the chunk next to the chunk including BA # 0. oLBL # 0 = cLBA # n + 1-pLBA # 0 (5)

B) For the next partition PRT # 01, the leading logical address and sector length of offset # 01 are obtained. Here, pLBA # 1: the leading logical address of PRT # 01 pLBA '# 1: PRT # 0 after applying offset # 00
1 first logical address cLBA # m: first logical address of the chunk including pLBA '# 1 cLBA # m + 1: first logical address of the next chunk after cLBA # m oLBA # 1: first logical address of offset # 01 oLBL # 1 : Sector length of offset # 01. In step 45, the start logical address of PRT # 01 after applying the offset # 00 is obtained from both the start logical address of PRT # 01 and the offset # 00 obtained in A). That is, the value of the following equation (6) is obtained. pLBA '# 1 = pLBA # 1 + oLBL # 0 (6)

In step 46, the start logical address of offset # 01 is the PRT # 0 after application of offset # 00.
1 is equal to the first logical address. oLBA # 1 = pLBA '# 1 Next, the head logical address of the PRT # 01 after applying the offset # 00 is compared with the head logical address of the chunk including the logical address. When they match: pLBA '# 1 == cLBA # m
In step 49, the sector length of offset # 01 is set to 0 in step 49. oLBL # 1 = 0 When they do not match: That is, pLBA '# 1! = CLBA #
m, in step 48, the sector length of offset # 01 is pL
This is obtained by subtracting pLBA '# 1 from the head logical address of the chunk next to the chunk containing BA'# 1. oLBL # 1 = cLBA # m + 1-pLBA '# 1 (7)

The procedure for obtaining the head logical address for actually accessing from the head logical address of the data to be accessed will be described below with reference to FIG. The first logical address of the data to be accessed is defined as the first embodiment. FS for
For DRV, Embodiment 2. In this case, the host computer requests the HCON for the first logical address of the data requested. The first logical address to be actually accessed is determined according to the first embodiment. In the case of (1), the DRV is applied to the disk array device. H for
CON is the leading logical address of the data requested for RCON. Here, aLBA: leading logical address of data to be accessed bLBA: logical address temporary storage means rLBA: leading logical address of data to be actually accessed oLBA # 0: leading logical address of offset # 00 oLBL # 0: offset # 00 OLBA # 1: Start logical address of offset # 01 oLBL # 1: Sector length of offset # 01

In step 51, aLBA and oLBA # 0
Compare. If aLBA <oLBA # 0, it is determined in step 53 that rLBA = aLBA. If aLBA> = oLBA # 0, in step 52, bLBA = aLBA + oLBL # 0 (8). In step 54, if the result of step 61 is aLBA> = oLBA # 0, bLB
A and oLBA # 1 are compared. bLBA <oLBA # 1
In step 56, rLBA = bLBA in step 56. If rLBA> = oLBA # 1, at step 55 rLBA = bLBA + oLBL # 1 (9) Thus, even if there are a plurality of partitions,
By appropriately adding an offset to the beginning of each partition, a cluster boundary and a chunk boundary can be matched, and the response performance to access is improved.

[0037]

As described above, according to the present invention, since the access is performed by matching the addresses of the partition and the chunk, the response performance can be improved without changing the hardware.

[Brief description of the drawings]

FIG. 1 is a diagram showing an arrangement relationship of logical addresses of a file system according to a first embodiment of the present invention.

FIG. 2 is an operation flowchart at the time of initialization according to the first embodiment;

FIG. 3 is an operation flowchart of data access according to the first embodiment.

FIG. 4 is a diagram showing an arrangement relationship of logical addresses of a file system according to a second embodiment of the present invention.

FIG. 5 is an operation flowchart at the time of initialization in the second embodiment.

FIG. 6 is an operation flowchart of data access according to the second embodiment.

FIG. 7 is a diagram illustrating an offset relationship between a partition and a cluster according to the third embodiment of the present invention.

FIG. 8 is an operation flowchart at the time of initialization in the third embodiment.

FIG. 9 is an operation flowchart of data access according to the third embodiment.

FIG. 10 is a diagram showing a conventional RAID configuration.

FIG. 11 is a diagram showing an arrangement of chunks in a conventional disk device.

FIG. 12 is a diagram showing a general software configuration.

FIG. 13 is a diagram showing an arrangement relationship of logical addresses in a conventional file system.

FIG. 14 is a diagram showing a conventional relationship between partitions and clusters.

[Explanation of symbols]

110 software on host computer, 111
Application (APP), 112 file system (FS), 113 driver (DRV), 120
Software on disk array controller, 121
Host interface control unit (HCON), 122
Disk array controller (RCON), 123 Disk controller (DCON), 131 DK # 0, 132 D
K # 1, 133 DK # 2, S2 partition head address calculation step, S3 partition and chunk head logical address comparison step, S4 offset calculation step, S5 offset storage (setting) step, S12 logical address and offset comparison step,
S14 Step of accessing with offset addition value, S
22 Partition head address calculation step, S2
3 Step of comparing the start logical address of the partition and the chunk, S24 Offset calculation step, S25
Offset storage (setting) step, S32 Logical address and offset comparison step, S34 Step of accessing with offset addition value.

Claims (4)

[Claims] A disk comprising a plurality of disk devices.
In a configuration for accessing a file system recorded in an array device, a step of knowing a head logical address of a partition, which is a set of data handling units of the file system, and a step of knowing a head logical address of a physical chunk of the disk device If the logical addresses of the partition and the chunk do not match, the difference is obtained as an offset, and an offset sector is set in the disk device. When data access is specified, there is specified data. A data access method for a disk array, comprising a step of adding a sector length of the set offset to a partition for access. 2. The method according to claim 1, wherein a driver under the file system of the host computer obtains an offset, and in the case of data access, adds a sector length of the offset to designate a sector to be accessed. Item 2. The data access method for a disk array according to Item 1. 3. The data processing system according to claim 1, wherein the host interface control unit of the disk array device obtains the offset, and specifies the sector to be accessed by adding the sector length of the offset in the case of data access. Item 2. The data access method for a disk array according to Item 1. 4. When the file system handles a plurality of partitions, an offset is added to the last sector address after offset correction of the immediately preceding partition for each of the following partitions after calculating the offset amount of the first partition. 2. The data access method for a disk array according to claim 1, wherein an offset amount of said partition is calculated.