JP2001184176A - Disk controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an effect of enhancement of performance by a log structured data writing method simply by preparing a minimum memory and idle area. SOLUTION: This disk controller 12 controls disk devices 110-0, 110-1 in which plural identifiable logical disks are allocated from a host computer 20 and the smaller number of buffer areas than the plural logical disks are allocated. The disk controller 12 secures unused buffer areas and switches the disk controller to the log mode in which an area formed by integrating an area on the disk device to which a specified logical disk is allocated and a secured buffer area is used as a data storage area when switching of the logical disks from use in a normal mode to the use in a log mode is instructed from the host computer 20. In addition, a corresponding buffer area is released from the logical disk when the logical disk in the log mode is switched to the use in the normal mode.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk controller for controlling data input / output to / from a disk device typified by a magnetic disk drive in accordance with a request from a host device. The present invention relates to a disk control device suitable for use in a disk array device (disk storage system) of medium class or higher.

[0002]

2. Description of the Related Art Recently, in a disk control apparatus for controlling data input / output to / from a disk apparatus according to a request from a host apparatus (host computer), for example, Japanese Patent Application Laid-Open No.
As described in Japanese Patent No. 3235 or the like, write data from a host device is stored in a write buffer in a non-volatile memory to a certain capacity, and the stored data of a certain capacity is stored in the disk controller. It has been proposed to apply a method in which data to be updated (old data of the same logical address) on a connected disk device is sequentially and sequentially written to a free area different from an area where the data is held. Hereinafter, such a method is called a log structured data writing method, and its operation mode is called a log mode. In addition to this method, a conventional general method of writing data to a physical address on a disk device corresponding to a logical address of write data from a host device on a one-to-one basis is referred to as a normal data writing method. And the operation mode is called a normal mode.

Further, in a disk control device to which the log structured data writing method is applied, a conversion table for associating a logical address (visible from the host device) handled by the host device with a physical address on the disk device is nonvolatile or volatile. Having in the memory of the nature.

Conventionally, in a disk control apparatus to which the log structured data writing method is applied, when an arbitrary logical disk viewed from a host device is used in the log structured data writing method, the capacity of the logical disk is further increased as a free area. An area to which the required capacity of several tens percent is added is secured in advance, and this area is fixedly used.

FIG. 17 shows an example. Here, the disk devices 310 constituting the disk arrays 31-0 and 31-1 are described.
-0, 310-1 logical disks LU0, L0 using the log structured data writing method using the areas S0, S1.
When U1 is defined, the relevant logical disks LU0, LU1
The capacity that can be seen from the host device is set to a value that is several tens of percent smaller than S0 and S1, so that the free space is reduced to S0 and S1
It is secured and realized within.

[0006]

According to the above-described log structured data writing method, the writing from the host computer is
When frequently performed on random logical addresses with a small size, an excellent feature is that write performance is improved by collectively sequentially writing to a continuous area of the disk device regardless of the logical addresses. Have.

However, the log structured data writing method is not versatile. When the data transfer size required by one writing is large, when the ratio of writing data to consecutive logical addresses is large, or when data is read out. Depending on how the disk device is used, such as when the ratio is high, the effect of improving the write performance may be small or the performance may be degraded.

Therefore, when it is expected that the write performance will decrease due to use in the log mode, it is conceivable to use the logical disk used in the log mode in the normal mode. However, in a disk controller using the conventional log structured data writing method, if the logical disk used in the log mode is temporarily returned to the normal mode, the capacity of the logical disk as seen from the host device is reduced. It becomes the physical capacity of the disk device itself, and the capacity is changed (increased) as compared with the capacity used in the log mode. Conversely, if the logical disk used in the normal mode is changed to the log mode, the capacity changes (decreases).

Therefore, in the disk control device to which the conventional log structured data writing method is applied, in order to switch between the log mode and the normal mode, all data stored in the logical disk is discarded and initialized. In practice, it has been difficult to realize switching between the log mode and the normal mode.

[0010] In addition, there are two other problems in the log structured data writing method itself. The first is that the logical address (visible to the host device) handled by the host device and the physical address when actually written to the disk device are one-to-one.
This is a problem caused by the fact that a conversion table for associating a logical address with a physical address is required because the conversion table is not uniquely determined. This problem is described below.

First, the capacity (size) of the conversion table
Increases in proportion to the capacity of the disk device connected to the disk controller. Further, since this conversion table is referred to each time an access is made from the host device, it is necessary to hold the conversion table in a nonvolatile or volatile memory for speeding up. In the case of a medium / large disk array device, the memory capacity required for the conversion table becomes enormous, and there is a problem that it is difficult to realize or very expensive. Therefore, it can be said that the log structured data writing method is suitable for a relatively small disk array device (disk system).

Second, when the log structured data writing method is applied, the capacity of the logical disk seen from the host device is reduced by several tens percent relative to the physical capacity of the disk device connected to the disk control device. It is a problem. This capacity loss is not negligible in a large-scale disk array device even in consideration of the performance improvement effect.

[0013] As described above, in the prior art to which the log structured data writing method is applied, the effect is most effective in a usage in which the data transfer size is small and a large number of data writings with random logical addresses occur. On the other hand, when the data transfer size required by one write is large, or when the ratio of writing data to consecutive logical addresses is large, or when the ratio of reading data is large, the effect is not so large. Can't
On the contrary, there is a possibility that the performance may be deteriorated.

Further, when the physical capacity of the disk device is increased, there is a problem that an extra area of the disk is required as a necessary memory and a free area.

The present invention has been made in consideration of the above circumstances, and has as its object to provide a disk which can obtain the effect of performance improvement by a log structured data writing method only by preparing a minimum memory and a free area. It is to provide a control device.

Another object of the present invention is to store user data in a normal mode and a log mode at any time when the tendency of access from a host device changes while using a disk device. An object of the present invention is to provide a disk control device that can maintain the performance of a disk device in an optimum state by switching and using the disk control device as it is.

[0017]

A disk controller according to the present invention includes a plurality of disk devices to which a plurality of logical disks identifiable by a host device are allocated and to which a smaller number of buffer areas are allocated than the plurality of logical disks. A disk controller that controls and uses both a log mode in which a log structured data writing method is applied and a normal mode in which a normal data writing method is applied, and when the logical disk is used in the normal mode, the logical disk is allocated. A normal mode disk use unit that uses the area on the disk device as the data storage area as it is, and, if the logical disk is used in the log mode, the area on the disk device to which the logical disk is allocated and the buffer area. Log mode to use as a data storage area Disk use means and first mode switching means for switching a logical disk used in the normal mode to use in the log mode, and a buffer not used when the logical disk is used for the log mode. A first area for securing the area and causing the log mode disk using means to use an area obtained by combining the area on the disk device to which the logical disk is allocated and the secured unused buffer area as the data storage area. Mode switching means, and second mode switching means for switching the logical disk used in the log mode to use in the normal mode, the log being combined with the area on the disk device to which the logical disk is allocated. Releases the buffer area used in the mode, and allocates the logical disk Only the area on the disk device is characterized in that a second mode switching means for use in the normal mode Disk unit.

In such a configuration, the area on the disk device to which the logical disk recognizable (visible) from the host device is allocated can be used by switching between the log mode and the normal mode. The number of buffer areas is smaller than the number of logical disks because the relationship with the buffer area that provides the free area necessary for use in the mode is not fixed, and the buffer area can be used in combination with the area of any logical disk. In other words, it is possible to reduce the free disk capacity prepared separately from the user disk capacity (logical disk capacity), minimize the disk device capacity loss due to the buffer area, and reduce the logical disk capacity relative to the physical disk capacity. To increase the ratio of disk space and to make effective use of disk drive space. Kill.

Also, unlike the prior art, not all of the disk device including the area of the disk device that is not suitable for the log mode is used in the log mode. The size of the conversion table for associating physical addresses can be reduced. That is, the capacity of the memory in which the conversion table and the like are placed can be minimized. Such an effect is larger in a disk array device having a larger disk capacity, and matches the recent trend of rapidly increasing the capacity of a disk device. Also, since only the area of the disk device suitable for the log mode (log structured data writing method) can be used in the log mode, even if other areas are used in the normal mode, the log structured data writing method can be used. The advantage of the method is not lost.

Further, according to the present invention, when the logical disk is switched from using the log mode to using the normal mode in the second mode switching means, the area on the disk device to which the logical disk is allocated is determined by the second mode switching means. By extracting all the latest data stored in the buffer area used as the data storage area together with the area, the logical disk in which the logical address and the physical address are associated one-to-one is extracted. Data transfer means for moving data to the allocated area on the disk device, and buffer area releasing means for releasing the buffer area after the data movement by the data moving means is completed. Here, in order to associate the logical address with the physical address on a one-to-one basis, the extracted latest data may be rearranged (sorted) in the order of the logical address.

In such a configuration, it is possible to switch from the log mode to the normal mode while retaining the user data, and release a buffer area that is no longer required by this switch from the logical disk used in the log mode. Therefore, the other logical disks can be switched from the normal mode to the log mode using the released buffer area.

Further, according to the present invention, for each logical disk, operation mode information indicating whether the logical disk is used in a log mode or a normal mode, and a log mode is indicated by the operation mode information. If there is, the operation mode management table in which the entry information including the used buffer area information indicating the buffer area used in the log mode is registered, while the first mode is added.
The mode switching means includes an unused buffer area search means for searching for an unused buffer area with reference to the operation mode management table, and a logical disk for switching the logical disk from the normal mode to the log mode. First operation mode management table updating means for updating the entry information of the operation mode management table corresponding to the disk is provided, and the second mode switching means switches the logical disk from the log mode to the normal mode. In this case, a second operation mode management table updating means for updating the entry information of the operation mode management table corresponding to the logical disk is provided.

In such a configuration, an unused (empty) buffer area required for switching the logical disk from the normal mode to the log mode can be easily searched.

Further, according to the present invention, a plurality of logical disks identifiable by the host device are allocated, a plurality of buffer areas are allocated, and the area of the logical disk is divided into a plurality of slices of a predetermined size and managed. A disk controller for controlling a disk device, comprising: a normal mode using a normal mode disk using unit and a log mode using a log mode disk using unit in the slice unit; and a first and a second mode. The switching between the normal mode and the log mode by the switching unit is performed in units of the slice. Here, when the operation mode management table is applied, the entry information of the table indicates whether the slice is used in the log mode or the normal mode in units of slices for each logical disk. When the log mode is indicated by the mode information and the operation mode information, the entry information including the used buffer area information indicating the buffer area used in the log mode may be used.

In such a configuration, the use in the normal mode and the log mode and the switching between the normal mode and the log mode are performed not in units of logical disks but in units of slices obtained by dividing the logical disk. The effect of improving performance by the log structured data writing method can be obtained. Only the part where access is concentrated can be selectively used in the log mode, and the advantage of the log structured data writing method can be maximized. .

Further, according to the present invention, the first mode switching means detects an upper candidate of a slice suitable for the log mode from each of the slices based on a predetermined condition, and selects a normal mode among the detected candidates. Is provided with a first selecting means for selecting a candidate in (1) as a slice to be switched to the log mode.

Further, according to the present invention, the second mode switching means detects lower-level candidates of a slice unsuitable for the log mode from the respective slices based on a predetermined condition, and sets the log mode among the detected candidates. Is provided with a second selecting means for selecting a candidate in the above as a slice to be switched to the normal mode.

In such a configuration, even if the tendency of access from the host device changes and the slice at which the effect of improving the write performance by the log structured data writing method is obtained changes, the slice suitable for the log mode can be obtained. Alternatively, the above change can be dealt with by appropriately detecting a slice that is not suitable for the log mode, so that the performance of the disk device can always be maintained in an optimal state. Here, the above detection processing may be performed at the time of a request from the host device according to the user operation, or may be performed periodically. Further, mode switching may be automatically performed according to the above detection result, or a switching candidate may be notified to the user through the host device to prompt the user to perform a switching instruction operation. Further, for the above detection processing, statistical information of at least the transfer size and the frequency of write access, which is the statistical information of the access request to the disk device from the host device, is stored in slice units for each logical disk. A statistical information table to be updated, and a statistical information table updating means for updating corresponding statistical information of the statistical information table in response to an access request from the host device to the disk device, based on the statistical information table. It is good to carry out the above detection processing.

[0029]

Embodiments of the present invention will be described below with reference to the drawings.

[First Embodiment] FIG. 1 shows the configuration of a disk array device (disk storage system) provided with a disk control device to which a log structured data writing method according to a first embodiment of the present invention is applied. It is a block diagram.

The disk array device 10 of FIG. 1 includes a plurality of sets, for example, two sets of disk arrays 11-0 and 11-1 and a host computer (host device) 20 in response to a request from the host computer 20. Array 11-0 or 11
And a disk controller (disk array controller) 12 for controlling data input / output to / from the disk controller. The disk arrays 11-0 and 11-1 include a plurality of disk devices (here, magnetic disk devices) 110-0 and 110, respectively.
It consists of -1.

The disk controller 12 has a rewritable nonvolatile memory 120, for example. In addition to the data write buffer 121 and the address conversion table 122 necessary for applying the log structured data writing method, the nonvolatile memory 120 has an operation necessary for realizing switching between the log mode and the normal mode. Each area of the mode management table 123 is reserved.

When data is written by the log structured data writing method, the physical capacity of the disk device actually required is described in the column of [Prior Art] and the column of [Problems to be Solved by the Invention]. As described above, the capacity of the logical disk (the minimum unit of the disk device seen from the host device) identified by the host computer (disk capacity that can be logically read and written by the host device) is usually several tens percent or more. On the other hand, in the case of the normal data writing method, the logical disk capacity and the physical disk capacity are almost the same. This is because, in the case of the log structured data writing method, both old data and new data corresponding to the same logical address can exist on the disk device.

In the disk controller 12 of the present embodiment,
As shown in FIG. 2, the logical disk LU1 identified by the host computer 20 is placed on an arbitrary disk array connected to the disk controller 12, for example, the disk array 11-1 (the disk device 110-0 constituting the disk array 11-1). A configuration having a buffer area B0 managed (secured) separately from the area described above is applied. In this example, the entire area S0 on the disk array 11-0 (the disk device 110-0 constituting the disk array 11-0) is defined as a logical disk LU0 that can be identified from the host computer 20, and the disk device 11-1 (the disk device constituting the disk array 11-1) is defined. 110-1) The area S1 excluding a part of the host computer 2 is the host computer 2.
It is defined as a logical disk LU1 that can be identified from 0. Then, a partial area on the disk array 11-1 (the disk device 110-1 configuring the same), that is, a part excluding the area S1, is secured as the buffer area B0.
The capacity of the buffer area B0 is the difference between the logical disk capacity seen from the host computer 20 and the actually required physical disk capacity. The buffer area B0 is used only in the log mode.

In this embodiment, the regions S0, S
1, the normal data writing method is applied to the areas S0 and S1 on the disk arrays 11-0 and 11-1 (the disk devices 110-0 and 110-1 constituting the logical arrays LU0 and LU1) to which the logical disks LU0 and LU1 are allocated. This is the required capacity. That is, when the logical disks LU0 and LU1 are used in the normal mode, the areas S0 and S1 are used as they are, and the buffer area B0 is used as the logical disk LU0 or LU.
It will not be used as part of one. Therefore, the capacities of the logical disks LU0 and LU1 do not change between the log mode and the normal mode.

The host computer 20 has a logical disk LU
0 and LU1 are identified (recognized) as follows. For each logical disk number indicating a plurality of logical disks that can be managed by itself, the host computer 20 inquires the disk controller 12 of the disk array device 10 about the capacity of the corresponding logical disk. In response to this inquiry, the disk controller 12 determines whether there is a corresponding logical disk and, if so, the capacity of the logical disk.
Notify the host computer 20.

In the example of FIG. 2, only for the inquiry by the logical disk number indicating LU0, LU1, the existence and capacity (S0, S1) of the corresponding logical disk LU0, LU1
Of the disk controller 12 to the host computer 2
Returned to 0. Thus, the host computer 20 can identify the logical disks LU0 and LU1.

FIG. 3 shows a disk array 1 when all of the logical disks LU0 and LU1 are used in the normal mode.
1-0, 11-1 (disk devices 110-0, 11-1
0-1) Indicates the usage status of the area above. The hatched areas, that is, the areas S0 and S1 are the parts that are actually used, and the buffer area B0 is unused (unused).

FIG. 4A shows an example of the contents of the operation mode management table 123 in the state shown in FIG. Each entry of the operation mode management table 123 includes a logical disk number setting field (logical disk number field) as logical disk identification information indicating a logical disk defined (allocated) on the disk array device 10;
A setting field (operation mode field) of operation mode information indicating whether the logical disk is used in the log mode or the normal mode, and a used buffer area information indicating a buffer area used in the mode. It has a buffer number setting field (used buffer number field). However, when the operation mode is the normal mode, the buffer area is not used as described above, and thus “unused” information indicating that the buffer area is not used is set in the used buffer number field.

As described above, in the state of FIG. 3, that is, in a state where both the logical disks LU0 and LU1 are used in the normal mode, as shown in FIG.
In each of the operation mode fields in the entry in which the logical disk number indicating LU1 is set, information indicating the normal mode is set, and in the used buffer number field,
In each case, information indicating “unused” is set.

Here, from the state where both of the logical disks LU0 and LU1 are used in the normal mode as shown in FIG. 3, one of the logical disks, for example, LU0, is switched to use in the log mode. FIG.
And a flowchart of FIG.

First, the host computer 20 stores the logical disk L
When the user wants to switch the logical disk LU0 to use in the log mode while U0 is used in the normal mode, instructs the disk controller 12 of the disk array device 10 to change the operation mode of the LU0 to the log mode. Send. The disk controller 12 is a host computer 20
When a mode switching instruction is received from, it is checked whether the instructed mode is the log mode or the normal mode (step S11).

If switching to the log mode is instructed as in this example, the disk controller 12
Checks whether the current mode of LU0 is the normal mode with reference to the operation mode management table 123 (step S12). If the LU 0 is in the normal mode as in the present embodiment (see FIG. 4A), the disk control device 12 performs a switching process to a log mode described in detail below (step S13).

First, the disk control device 12 refers to the used buffer number field of the operation mode management table 123 to thereby indicate the buffer area indicated by the buffer number not registered in the operation mode management table 123, ie, the unused buffer area. (Step S21).
In this example, the buffer area existing in the disk array device 10 is one of B0 (on the disk array 11-1). This buffer area B0 is not registered as being used, as shown in FIG. Therefore, in step S21, the buffer area B0 is searched as an unused buffer area.

When the unused buffer area B0 can be searched for (step S22), the disk controller 12 sets the unused buffer area B0 to the logical disk LU0 whose mode is to be switched, as shown in FIG. Together with the allocated area S0, that is, the area S0 used in the normal mode, the data storage area of the logical disk LU0 in the log mode is set as an access data storage area (step S23). As a result, it is possible to secure the capacity for the free area required in the log mode.

At the same time, the disk controller 12 updates the contents of the entry relating to the logical disk LU0 in the operation mode management table 123 from the state shown in FIG. 4A as shown in FIG. 4B (step S24).

The buffer area B0 allocated to the logical disk LU0 is used only as the capacity of the free area. Therefore, the capacity of the logical disk LU0 seen from the host computer 20 does not change by switching from the normal mode to the log mode. Further, the data used up to that time in the normal mode remains in the area S0 as it is. In the case of the normal mode, the logical address and the physical address on the disk device are associated one-to-one. Therefore, the disk control device 12 can uniquely create the address conversion table 122 indicating the correspondence between the logical address and the physical address at this time for the logical disk LU0 (step S25).

FIG. 8 shows an example of a simplified data structure of the address conversion table 122. Here, a physical address corresponding to the logical address is recorded for the logical disk LU0. After creating the address conversion table 122 for the logical disk LU0, the disk control device 12 generates the address conversion table 122 every time a data write request from the host computer 20 occurs while using the corresponding logical disk LU0 in the log mode. Address conversion table 1
Update 22.

On the other hand, when the unused buffer area cannot be searched in the search processing in step S21 (step S2).
2), the disk controller 12 notifies the host computer 20 that switching to the log mode is not possible (step S)
26). Such a state is, for example, the logical disk LU1
Is used in the log mode and only one buffer area B0 is allocated to the LU1, and another logical disk LU0 is to be switched to use in the log mode. That is, the number of logical disks exceeding the number of buffer areas cannot be set to the log mode.

When the LU0 instructed to switch to the log mode is not in the normal mode, that is, when it is already used in the log mode (steps S11 and S1).
2), the disk controller 12 terminates the process without doing anything. At this time, the host computer 20 may be notified that the designated LU0 is already in the log mode.

Next, a process for switching a logical disk used in the log mode to use in the normal mode will be described with reference to the flowcharts of FIGS. explain.

First, the host computer 20 stores the logical disk L
When it is desired to switch the logical disk LU0 to use in the normal mode while U0 is used in the log mode, an instruction to switch the operation mode of LU0 to the normal mode is sent to the disk controller 12. Upon receiving the mode switching instruction from the host computer 20, the disk controller 12 checks whether the instructed mode is the log mode or the normal mode (Step S).
11).

If switching to the normal mode is instructed as in this example, the disk controller 12
Checks whether the current mode of LU0 is the log mode with reference to the operation mode management table 123 (step S14). If the LU 0 is in the log mode as in the present embodiment (see FIG. 4B), the disk control device 12 performs a process of switching to the normal mode (step S15) described in detail below.

First, the disk controller 12 extracts all the valid latest data randomly stored in the data storage area corresponding to the logical disk LU0 used in the log mode, that is, the area of S0 + B0. The data is rearranged in the order of the addresses, and the data is moved (restored) to the area S0, which is newly used in the normal mode, where the logical address and the physical address of the area S0 are associated one-to-one (step S31). ).

FIG. 10 shows an example of the rearrangement in step S31. In the log mode, data corresponding to a logical address is stored at an arbitrary address in the area of S0 and B0. Therefore, the disk control device 12 searches the entire address conversion table 122 for the logical disk LU0 in the above-mentioned step S31, and arranges the addresses S0 so as to be arranged in ascending order of the logical address from the head of the S0 area.
And rearrangement of valid latest data stored in the area B0.

When all the rearrangement (data movement) in step S31 is completed, no valid data exists in the buffer area B0. This area B0 does not need to be used in the normal mode. Therefore, the disk controller 12 releases the buffer area B0 allocated and used for the logical disk LU0 in the log mode from the logical disk LU0, and releases the logical disk LU0 to the area S.
Switching to the normal mode using only 0 (step S3
2, S33).

At the same time, the disk controller 12 updates the contents of the entry relating to the logical disk LU0 in the operation mode management table 123 from the state shown in FIG. 4B as shown in FIG. 4A (step S34). In this state,
That is, when the mode is switched to the normal mode, the address conversion table 122 relating to the logical disk LU0 previously used in the log mode becomes unnecessary.

When the buffer area B0 is released from the logical disk LU0 as described above, the other logical disk LU1 is used by using the buffer area B0.
Can be switched from normal mode to (used in) log mode. As described above, the upper limit of the logical disks that can be used in the log mode matches the number of buffer areas, and in the present embodiment, there is only one.

The switching between the normal mode and the log mode described above, in particular, the switching from the log mode to the normal mode is not frequently performed. The reason is that when the mode is switched from the log mode to the normal mode, the rearrangement processing of all the valid latest data stored in the area allocated to the logical disk used in the log mode occurs, so that the switching is completed. This takes a considerable amount of time. In the present embodiment, in the mode switching, a mode switching instruction is transmitted from the host computer 20 to the disk control device 12 in the disk array device 10 in accordance with a command input operation or a switch operation from the user to the host computer 20. It is done by being done.

When the LU0 instructed to switch to the normal mode is not in the log mode, that is, when the LU0 is already used in the normal mode (steps S11 and S1).
4), the disk controller 12 terminates the process without doing anything. At this time, the host computer 20 may be notified that the designated LU0 is already in the normal mode.

As described above, in the first embodiment of the present invention, since the capacities of the logical disks LU0 and LU1 seen from the host computer 20 do not change when used in either the log mode or the normal mode, Switching between the log mode and the normal mode can be realized without discarding data stored in the logical disk.

In this embodiment, a part (one) of the plurality of logical disks (LU0, LU1) can be used in the log mode and the other (the other) can be used in the normal mode. In contrast to the configuration described above, the memory capacity necessary for the address conversion table essential for use in the log mode and the buffer area (that is, an empty area where data is written collectively regardless of the data position to be originally updated) ) And can be reduced.

[Second Embodiment] FIG. 11 shows a second embodiment of the present invention.
FIG. 10 is a block diagram illustrating a configuration of a disk array device (disk storage system) including a disk control device to which the log structured data writing method according to the embodiment is applied.
1 are given the same reference numerals for convenience.

A feature of the disk array device 10 of FIG. 11 is that the areas S0 and S1 on the disk arrays 11-0 and 11-1 (the disk devices 110-0 and 110-1 constituting the disk arrays) are shown in FIG. , SL00-S in units of predetermined size
L0n and SL10 to SL1m are divided and managed, and switching between the normal mode and the log mode is also performed in units of the slice. Here, in the area of the disk array 11-1 (the disk device 110-1 constituting the disk array 110-1), the area other than the S1 (the area corresponding to the buffer area B0 in the first embodiment) is changed to a predetermined size. It is divided into p buffer areas B00 to B0p and managed. The capacity of the buffer area B0i (i = 0 to p) is several tens of percent of the capacity of each slice SL00 to SL0n and SL10 to SL1m. The number (p + 1) of the buffer areas B0i is smaller than the total number of slices (n + m + 2) ((p + 1) <(n + m + 2)). Note that the buffer areas B00 to B0p are distributed to the disk array 11-0 (upper disk device 110-0) and the disk array 11-1 (upper disk device 110-1), and the sizes of S0 and S1 are substantially reduced. It is possible to make them equal.

In the present embodiment, in order to realize the mode switching in the slice unit, the operation mode management table 12 in the non-volatile memory 120 of the disk controller 12 is used.
Unlike the first embodiment, the third embodiment employs a data structure in which entry information is stored in slice units for each logical disk.

In order to manage the mode switching in slice units, an area of a statistical information table 124 described later is secured in the nonvolatile memory 120 of the disk control device 12.

Except for such mode switching in slice units and management in slice units, the use of the buffer area and the rearrangement of data when switching from the log mode to the normal mode are performed in the first embodiment. This is performed in the same manner as in the embodiment.

Now, the statistical information table 124 in FIG.
Is used to hold statistical information of access status in slice units for each logical disk. The statistical information in the statistical information table 124 is updated by the disk control device 12 each time the host computer 20 accesses the corresponding slice.

FIG. 13 shows an example of the data structure of the statistical information table 124. The statistical information held in the statistical information table 124 is collected by the disk control device 12 in slice units, and includes (1) the number of write commands (the number of writes), (2) the number of read commands (the number of reads),
(3) It consists of information on the average data transfer size (average size) specified by the write command.

Next, switching of the mode to the log mode in the slice unit according to the present embodiment will be described with reference to the flowcharts of FIGS. First, the disk control device 12 refers to the statistical information held in the statistical information table 124 for all slices of the logical disks LU0 and LU1, and determines the slice in which access that can obtain the effect of the log mode has occurred most effectively. Are extracted in order from the one in which is obtained (steps S41 and S4).
2). In other words, slice candidates suitable for log mode are
Only N items are extracted in order from the top candidate.

The effect of improving the performance in the log mode increases as data writing to random logical addresses with a small transfer size occurs more frequently. This condition is met because the occurrence ratio of write commands is higher than that of read commands (that is, the number of writes> the number of reads), and the number of writes is greater than the reference number (that is, the absolute number of write commands is greater). ), A slice in which an access has occurred in which the average size (the average data transfer size specified by the write command) is smaller than the reference size. Therefore, the disk control device 12 determines in steps S41 and S41
At 42, for slices that meet the above conditions,
The number of write commands is higher, the number of writes is larger, and the average size is smaller.
It is extracted as a top candidate that can obtain the best performance improvement effect in the log mode. Here, using predetermined coefficients k1 to k3, (k1 × write command occurrence ratio +
The calculation of (k2 × number of writes + k3 × average size) is performed to calculate the log mode optimality indicating the degree suitable for the log mode, and the top N items having the greatest optimality are extracted.

When the disk controller 12 extracts the upper N slices suitable for the log mode, the disk controller 12 refers to the operation mode management table 123 for the extracted slices, and selects the slice used in the normal mode from the slices. Select (Step S43). Then, the disk controller 12 determines, for each of the selected slices,
Switching processing for switching to use in the log mode is performed (step S44).

Here, for simplicity of explanation, N = 1
It is assumed that the slice SL0i in the area S0 to which the logical disk LU0 is allocated has been extracted and selected. Hereinafter, the switching process for switching the slice SL0i to use in the log mode will be described with reference to the flowchart in FIG.

First, the disk control device 12 searches for an unused buffer area with reference to the operation mode management table 123 (step S51).

If an unused buffer area, for example, B0
If j can be retrieved (step S52), the disk controller 12 sets the area of the slice B0j (the first
Like the buffer area B0 for S0 in the embodiment of the present invention, the data storage area in the log mode is combined with the previously searched area of the slice SL0i (step S5).
3). As a result, the capacity for the free area (slice) required in the log mode can be secured. At this time, the disk control device 12 updates the contents of the entry related to the slice SL0i in the operation mode management table 123 (Step S54). Further, the disk control device 12 creates and registers the entry information of the address conversion table 122 indicating the correspondence between the logical address and the physical address used in the log mode (Step S55).

On the other hand, when an unused buffer area cannot be searched, that is, since all buffer areas S00 to S0p are used together with the same number of arbitrary slices in the log mode, the unused buffer area is searched. If not, the disk controller 12 refers to the operation mode management table 123 to search for all the slices currently operating in the log mode, and refers to the statistical information table 124 based on the slices. The slice that deviates from the above condition, that is, the slice with the smallest log mode optimality, is searched (steps S56 and S57). Then, the disk controller 12 switches the searched slice to the normal mode, thereby releasing the buffer area and securing it as an unused buffer area (step S5).
8). The following operation is the same as when the unused buffer area can be searched, and thereafter, the above steps S53 to S55
The slice SL0i selected in step S43 and having a high log mode effect is used in the log mode. As can be seen, the number of slices exceeding the number p of buffer areas cannot be set to the log mode.

If the slice SL0i is not already selected in the log mode in step S43 because the slice SL0i is already in the log mode, the processing for switching to the log mode for the slice SL0i (step S44) ) Is not performed.

In the above, the case where the value of N is 1 has been described. However, the same can be applied to the case of 2 or more (plural), and the normal mode is selected from the upper N slices suitable for the log mode. Only the slice needs to be switched to the log mode.

Next, the processing at the time of switching the mode to the normal mode in slice units according to the present embodiment will be described with reference to FIG.
This will be described with reference to the flowchart of FIG. First, the disk control device 12 refers to the statistical information held in the statistical information table 124 for all slices of the logical disks LU0 and LU1, and an access that does not satisfy the above conditions and the effect of the log mode cannot be obtained occurs. Slice
Only M items are extracted in order from the one with the least effect (steps S61 and S62). That is, only M slice candidates that are inappropriate for the log mode are extracted in order from the lowest candidate (the slice having the smallest log mode optimality).

When the disk controller 12 extracts the lower M slices unsuitable for the log mode, the disk controller 12 refers to the operation mode management table 123 for the extracted slices, and selects the slice used in the log mode from the slices. Is selected (step S63). Then, the disk controller 12 determines, for each of the selected slices,
A switching process for switching to use in the normal mode is performed (step S64).

Here, for the sake of simplicity, M = 1
It is assumed that the slice SL0i in the area S0 to which the logical disk LU0 is allocated has been extracted and selected. The process of switching the slice SL0i to use in the normal mode (step S64) is the same as the switching process according to the flowchart of FIG. 9 in the first embodiment except that the switching is performed in slice units. Done in

First, assuming that the buffer area used in the log mode in accordance with slice SL0i is B0j,
All valid latest data stored at random in the area of SL0i + B0j is extracted and rearranged in the order of the logical address, and the logical address and the physical address of the slice SL0i are newly associated with each other. Data is moved to the slice SL0i used in the normal mode.

When all the data is moved to SL0i used in the normal mode, the buffer area B0j used up to that time in the log mode is released, and the B0j is used.
Is switched to use in the normal mode. At this time, the contents of the entry regarding the slice SL0i in the operation mode management table 123 are updated.

As described above, in the second embodiment of the present invention, the mode switching on a slice-by-slice basis is applied, and the slice in which the access that gives the effect of the log mode most occurs is switched to the use in the log mode. Slices that have been accessed with the least effect of log mode are now switched to use in normal mode, so even if the tendency of access from the host computer changes while using the disk device, In addition, it is possible to switch and use the mode suitable for the access tendency while retaining the user data for each slice, and to maintain the performance of the disk device in an optimal state.

Further, since the switching between the normal mode and the log mode is performed in slice units, that is, the mode switching is performed in a unit smaller than the capacity of the logical disk, the time required for switching between the normal mode and the log mode is obtained. Can be shortened.

The switching of the slice mode between the normal mode and the log mode based on the statistical information as described above can be performed, for example, according to a command input operation or a switch operation from the host computer 20 by the user. This is performed at the timing when the mode switching instruction is sent to the control device 12. In addition,
The disk control device 12 may periodically search for statistical information, and automatically switch to a log mode for slices that provide more effects. In addition, when a statistical information search is periodically performed and a slice with a higher effect is found, the user is notified via the host computer 20 that switching is more preferable in terms of performance. The operation may be prompted.

The data write buffer 121, the address conversion table 122, the operation mode management table 1
23 and the statistical information table 124 are placed on a high-speed volatile memory such as a RAM and used, so that table access and the like can be speeded up. In this case, the data write buffer 121 and the address conversion table 12 are used so that the data is not lost even when the power supply of the device is cut off.
2. A copy (backup) of the operation mode management table 123 and the statistical information table 124 may be stored in a nonvolatile memory or a nonvolatile storage medium such as a disk device.

In the first and second embodiments described above, the case where both are applied to mode switching in a disk array device has been described. However, the same applies to mode switching in a disk storage system having a plurality of disk devices. Applicable. However, the effect of mode switching is greater for a disk array device having a larger disk capacity.

[0089]

As described above, according to the present invention, the log structured data writing method is applied to a part of a plurality of logical disks, and the normal data writing method is applied to the remaining logical disks. Therefore, the effect of improving the performance by the log structured data writing method can be obtained while minimizing the memory capacity and buffer area required for the address translation table essential for use in the log mode. This effect is larger in a disk device having a larger disk capacity, and matches the recent trend of rapid increase in the capacity of disk devices. In particular, the greatest effect can be obtained when a logical disk on which accesses for which the log structured data writing method is most effective is concentrated among a plurality of logical disks is used in the log mode.

Further, according to the present invention, while managing a logical disk by dividing it into slices of a predetermined size, statistical information indicating the tendency of access from the host device is collected in slice units, and slices are collected based on the statistical information. Since switching between normal mode and log mode is performed in units,
If the tendency of access from the host device changes while using the disk device, it is possible to switch between the normal mode and the log mode at any time while retaining the user data, and use the disk mode. The performance of the device can be maintained in an optimal state.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a disk array device (disk storage system) including a disk control device to which a log structured data writing method according to a first embodiment of the present invention is applied.

FIG. 2 shows a disk array 11- applied to the embodiment.
0, 11-1 (disk devices 110-0, 11-
FIG. 4 is a diagram showing an example of the relationship between the area 1) and the logical disk and buffer areas.

FIG. 3 shows logical disks LU0 and L0 in the embodiment.
The figure which shows the state at the time of using both U1 in the normal mode.

FIG. 4 is an exemplary view showing an example of contents of an operation mode management table 123 applied in the embodiment.

FIG. 5 is an exemplary flowchart for explaining the flow of processing at the time of mode switching in the embodiment.

FIG. 6 is an exemplary flowchart for explaining a detailed procedure of a switching process to a log mode in the embodiment.

FIG. 7 is a diagram showing a state in which the logical disk LU0 is switched to use in a log mode in the embodiment.

FIG. 8 is an address conversion table 12 according to the embodiment;
FIG. 3 is a diagram showing an example of the data structure of FIG.

FIG. 9 is an exemplary flowchart for explaining a detailed procedure of a switching process to a normal mode in the embodiment.

FIG. 10 is a diagram illustrating an example of a process of rearranging data at the time of switching to the normal mode.
FIG. 8 is a diagram for explaining an example of switching from 0) in log mode to use in normal mode.

FIG. 11 is a block diagram showing a configuration of a disk array device (disk storage system) including a disk control device to which a log structured data writing method according to a second embodiment of the present invention is applied.

FIG. 12 is a disk array 11 applied in the embodiment;
-0, 11-1 (Disk devices 110-0, 11-
FIG. 3 is a diagram illustrating an example of a relationship between an area 1) and a logical disk, and a buffer area and a slice.

FIG. 13 is a statistical information table 1 applied in the embodiment.
The figure which shows the example of 24 data structures.

FIG. 14 is an exemplary flowchart for explaining processing at the time of switching slice units to the log mode in the embodiment.

FIG. 15 is an exemplary flowchart for explaining a detailed procedure of switching processing to a log mode in the embodiment.

FIG. 16 is a flowchart for explaining processing at the time of switching slice units to a normal mode in the embodiment.

FIG. 17 is a view for explaining a conventional technique.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 ... Disk array apparatus (disk storage system) 11-0, 11-1 ... Disk array 12 ... Disk controller 110-0, 110-1 ... Disk apparatus 120 ... Non-volatile memory 121 ... Data write buffer 122 ... Address conversion table 123: operation mode management table 124: statistical information table

Claims (11)

[Claims] 1. A method for writing a log-structured data by controlling a plurality of disk devices to which a plurality of logical disks identifiable from a host device are allocated and to which a smaller number of buffer areas are allocated than the plurality of logical disks. A disk control device using both a log mode and a normal mode applying a normal data writing method, wherein when using the logical disk in the normal mode, an area on the disk device where the logical disk is allocated is Means for using a normal mode disk that is used as it is as a data storage area; and when the logical disk is used in the log mode, data storage is performed by combining the area on the disk device to which the logical disk is allocated and the buffer area. Use log mode disk to use as area And a first mode switching means for switching the logical disk used in the normal mode to use in the log mode, which is not used when the logical disk is switched to use in the log mode. The log mode disk use unit secures the buffer area of the logical mode and sets an area obtained by combining the area on the disk device to which the logical disk is allocated and the secured unused buffer area as the data storage area. First mode switching means for use, and second mode switching means for switching the logical disk used in the log mode to use in the normal mode, wherein the disk to which the logical disk is assigned Release the buffer area used in the log mode together with the area on the device, A second mode switching unit that causes the normal mode disk using unit to use only the area on the disk device to which the logical disk is allocated. 2. The system according to claim 1, wherein the second mode switching unit is configured to, when switching the logical disk from using the log mode to using the normal mode, use an area on the disk device to which the logical disk is allocated. All the latest data stored in the buffer area used as the data storage area together with the area is extracted, and the logical address and the physical address are associated one-to-one. A data moving means for moving data to an area on the disk device to which a disk is allocated, and a buffer area releasing means for releasing the buffer area after the data moving by the data moving means is completed. Item 2. The disk control device according to Item 1. 3. A case in which, for each logical disk, operation mode information indicating whether the logical disk is used in a log mode or a normal mode, and a log mode is indicated by the operation mode information. Further comprising an operation mode management table in which entry information including the used buffer area information indicating the buffer area used in the log mode is registered, wherein the first mode switching means is configured to execute the operation mode management. An unused buffer area searching means for searching the unused buffer area by referring to a table, and the operation mode corresponding to the logical disk when the logical disk is switched from the normal mode to the log mode. First operation mode management table updating means for updating entry information of the management table, Serial second mode switching unit, when switching the logical disk from the logging mode to the normal mode,
3. The disk control device according to claim 2, further comprising a second operation mode management table updating unit that updates entry information of the operation mode management table corresponding to the logical disk. 4. A plurality of disk devices, wherein a plurality of logical disks identifiable by a host device are allocated and a plurality of buffer areas are allocated, and the logical disk area is divided into a plurality of slices of a predetermined size and managed. A disk controller that switches between a log mode in which a log structured data writing method is applied and a normal mode in which a normal data writing method is applied, wherein the slice in the logical disk is used in the normal mode. The normal mode disk using means for directly using the area on the disk device to which the slice is allocated as the data storage area; and, when using the slice in the logical disk in the log mode, The allocated area on the disk unit and the buffer Log mode disk use means for using the area as a data storage area, and first mode switching means for switching the slice used in the normal mode to use in the log mode, When switching to the use in the log mode, the unused buffer area is secured, and the combined area of the area on the disk device to which the slice is allocated and the secured unused buffer area is set. First mode switching means for causing the log mode disk use means to use the data storage area, and second mode switching means for switching the slice used in the log mode to use in the normal mode. And the log on the disk device together with the area on the disk device to which the slice is allocated. A second mode switching unit that releases the buffer area used in the mode and causes the normal mode disk using unit to use only the area on the disk device to which the slice is allocated. Disk controller. 5. The second mode switching means, when switching the slice from use in the log mode to use in the normal mode, an area on the disk device to which the slice is allocated and the area All the latest data stored in the buffer area used as the data storage area is extracted together with the data, and the slice in which the logical address and the physical address are associated one-to-one is allocated. 5. A data transfer means for transferring data to an area on the disk device, and a buffer area release means for releasing the buffer area after completion of data transfer by the data transfer means. Disk controller. 6. An operation mode information indicating whether the slice is used in a log mode or a normal mode in the slice unit for each logical disk, and a log mode is indicated by the operation mode information. The operation mode management table in which entry information including the used buffer area information indicating the buffer area used in the log mode is registered, and the first mode switching means includes: An unused buffer area search unit that searches for the unused buffer area by referring to an operation mode management table; and an operation corresponding to the slice when the slice is switched from the normal mode to the use in the log mode. First operation mode management table updating means for updating the entry information of the mode management table. The second mode switching unit updates the entry information of the operation mode management table corresponding to the slice when the slice is switched from the log mode to the normal mode. 6. The disk control device according to claim 5, further comprising means. 7. The first mode switching means detects an upper candidate of the slice suitable for the log mode from the respective slices based on a predetermined condition, and switches the normal mode among the detected candidates to the normal mode. 7. The disk control device according to claim 6, further comprising: a first selection unit that selects a candidate as a slice to be switched to the log mode. 8. The second mode switching means detects lower-level candidates of a slice unsuitable for the log mode based on a predetermined condition from among the slices, and switches the log mode to the log mode among the detected candidates. 8. The disk control device according to claim 7, further comprising a second selection unit that selects a candidate as a slice to be switched to the normal mode. 9. A method according to claim 1, wherein said first mode switching means determines a predetermined condition from slices in said log mode when said unused buffer area search means fails to search for said unused buffer area. The most inappropriate slice in the log mode is detected based on the above, and the slice is switched to use in the normal mode by the second mode switching means, so that a corresponding buffer area is secured as the unused buffer area. 7. The disk control device according to claim 6, further comprising a buffer area securing unit that performs the operation. 10. The statistical information of an access request from the host device to the disk device, the statistical information being capable of extracting at least an average data transfer size and a write access frequency is stored in the slice unit for each logical disk. A statistical information table that is held, and a statistical information table updating unit that updates statistical information corresponding to the statistical information table in response to an access request to the disk device from the host device, 8. The disk according to claim 7, wherein the selecting unit detects a slice having a smaller average data transfer size and a higher frequency of the write access as the upper candidate by referring to the statistical information table. Control device. 11. Statistical information of an access request to the disk device from the host device, wherein at least an average data transfer size and write access frequency can be extracted in the slice unit for each logical disk. A statistical information table that is held, and a statistical information table updating unit that updates statistical information corresponding to the statistical information table in response to an access request to the disk device from the host device, The selecting unit detects a slice having a smaller average data transfer size and a higher frequency of the write access as the higher-order candidate by referring to the statistical information table. By referring to the information table, as the top candidate, the average data transfer size is larger 9. The disk control device according to claim 8, wherein a slice having a lower write access frequency is detected.