JP2003263276A - Disk system and disk access method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a disk system to execute write-through processing in a system using an address conversion map with a method suitable for fault- tolerance improvement. <P>SOLUTION: A disk controller 20 has an address conversion map 23 which links logic addresses to physical addresses, and an available memory management table 24 which manages available memory areas of a disk 30. When a write is requested to the disk 30, an access controller 21 checks whether or not the physical address linked to a specified logic address exists with reference to the address conversion map 23. If it exists, the write of data is executed to the physical address. On the other hand, if not, an available memory area of the disk 30 is retrieved using the available memory management table 24 and then registered to the address conversion map 23. Subsequently, data is written to the physical address and the update contents of the address conversion map 23 is recorded in the disk 30. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk system for managing correspondence between a logical address and a physical address by using an address translation map and a disk access method for this disk system, and more particularly to this dynamic update. The present invention relates to a disk system and a disk access method for realizing write through processing in a system using an address translation map in a procedure suitable for improving fault tolerance.

[0002]

2. Description of the Related Art In recent years, work has been computerized in various industries, and various devices have been devised to process each work as quickly as possible. And, one of them is speeding up of disk access.

As a technique for increasing the speed of this disk access, the performance of a system irrelevant to the address at which data writing is requested by the host device is disclosed, for example, in Japanese Patent Laid-Open No. 11-53235. There is known a technique in which a logical address is associated with a physical address and the correspondence is managed by an address conversion map on a cache memory.

Further, in the system utilizing this address translation map, the write completion is notified to the host system when the write data is stored in the cache memory,
It is common to perform so-called write-back processing, in which actual writing to the disk is executed at a predetermined timing such as when the cache memory is full. Then, by performing this write-back processing, the speed of disk access can be further increased.

[0005]

By the way, in this write-back process, so-called dirty data, which is present only in the cache memory and is not yet reflected in the disk, is temporarily generated, so that the system is unexpectedly abruptly generated. There is a risk that these will be lost if is down. For this reason, recently, for a system requiring high reliability, there is an increasing demand for storing write data in a cache memory and a disk almost at the same time, so that a so-called write-through process can be selected.

However, for fault tolerance, when performing write-through processing in a system that uses an address translation map, it is important to manage this dynamically updated address translation map reliably and efficiently. Depending on the coping method, the system performance may be significantly reduced.

The present invention has been made in consideration of such circumstances, and the write-through processing in a system using this dynamically updated address translation map is executed in a procedure suitable for improving fault tolerance. It is an object of the present invention to provide a disk system and a disk access method that realize the above.

[0008]

In order to achieve the above-mentioned object, the present invention requires writing to a disk in a disk system that manages the correspondence between logical addresses and physical addresses using an address conversion map. At this time, the determination unit that determines whether or not any physical address is associated with the specified logical address by referring to the address conversion map, and any physical address is associated by the determination unit. When it is determined that there is no, search the free space of the disk,
A writing unit that registers the physical address in the address conversion map in association with the logical address, immediately writes data to the physical address, and writes the updated content of the address conversion map to the disk. And is provided.

In this disk system, in principle, writing is performed on the currently assigned physical address without changing the address translation map, and if there is no physical address assigned to the logical address, A free area for physical address is allocated on the spot, and the changed portion of the address conversion map is written to the disk together with the data.

In other words, this disk system executes the write-through processing in the system using this dynamically updated address translation map in a procedure suitable for improving fault tolerance so as not to update the address translation map as much as possible. Realize what you do.

Further, this eliminates the need to save the address translation map in the disk at the time of shutdown, so that the time required for shutdown can be shortened.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing the configuration of a disk controller according to this embodiment.

As shown in FIG. 1, the disk controller 20 of this embodiment is interposed between the host computer 10 and the disk 30, and receives a write request or a read request from the host computer 10 to receive the disk 30.
Data is written to or read from the disk 30. The disk controller 20 includes an access control unit 21, a cache memory 2
2. It has an address conversion map 23 and a free area management table 24.

When the host computer 10 requests writing of data, the disk controller 20 notifies the host computer 10 of the completion of writing when the data is stored in the cache memory 22,
A so-called write-back mode in which actual writing to the disk 30 is executed at a predetermined timing such as when the cache memory 22 becomes full, and so-called write-through in which write data is stored in the cache memory 22 and the disk 30 almost simultaneously. There are two operation modes including a mode and the operation mode is set in advance by an instruction from the host computer 10. The following are the features of this disk controller 20,
The operation principle in the write through mode will be described.

The access control unit 21 controls access to the disk 30 in response to a request from the host computer 10. For that purpose, the cache memory 22, the address translation map 23, and the free space management table 24 are provided. It also manages. The cache memory 22 is a storage medium for temporarily storing a part of the cache memory 22 for the purpose of shortening the response time when reading data from the disk 30, for example. In the address conversion map 23, the host computer 10 uses the disk 3
It is a table for managing the correspondence between a logical address for pointing each block in the virtual address space defined for 0 and a physical address for pointing each block in the real address space of the disk 30. The free area management table 24 is a table for managing free blocks of the disk 30.

When receiving the write request, the access control unit 21 determines the physical position on the disk 30 to actually write the data by using the address conversion map 23 and the free space management table 24 and receives the write request. The process of writing data to that position is performed, and if necessary,
The address translation map 23 is rewritten. FIG. 2 is a conceptual diagram showing an example of the address conversion map 23.

The contents of the address conversion map 23 are read from the disk 30 when the system is started up, that is, copied from the address conversion map 31 to the address conversion map 23. As shown in FIG. The correspondence between the logical address and the physical address is held.
In the example of FIG. 2, the physical address 10 is assigned to the logical address 0,
The logical address 1 is associated with the physical address 20, and the logical address 2 is not associated with the physical address.

Here, it is assumed that the host computer 10 issues a write request for 3 blocks from logical addresses 0. In this case, the access control unit 21 of the disk controller 20 processes this writing in the procedure shown below.

First, the access control section 21 refers to the address conversion map 23 to check which physical address the logical address 0 designated by the write request corresponds to. As a result, the access control unit 21 knows that the physical address is 10, and writes the data of the first block, that is, the logical address 0 to this physical address 10.

Subsequently, the access control unit 21 checks which physical address the logical address 1 of the second block corresponds to, similarly to the first block, and writes the data to the physical address 20 obtained by this. I do.

Then, the access control unit 21 checks to which physical address the logical address 2 of the third block corresponds, as in the first and second blocks, and the physical address still corresponds to this logical address 2. Know that it is not attached. Therefore, the access control unit 21 searches the free space management table 24 for a free space in the disk 30, writes data in the free space, and sets the address conversion map 23 to associate the physical address with the logical address 2. Update. At the same time, the access control unit 21 reflects the updated portion, or the minimum access unit to the disk 30 and the block including the updated portion, in the address translation map 31 of the disk 30.

Here, it is assumed that the physical address 21 immediately after the physical address 20 corresponding to the logical address 1 is an empty area. Then, in such a case, the access control unit 21 selects this physical address 21 and realizes high-speed processing.

FIG. 3 is a diagram showing how data is written to the disk 30 at this time, and FIG. 4 is a diagram showing the address translation map 23 after this data writing. Thus, the disk controller 20 writes data to the physical address if the physical address is associated with the logical address specified at the time of the write request, and if the physical address is not yet associated, A free area is searched and associated, and data is written, the address conversion map 23 is updated, and the updated portion is reflected on the disk 30.

FIG. 5 is a flow chart showing the operation procedure of the disk controller 20.

When the write request is received from the host computer 10, the access control unit 21 of the disk controller 20 first refers to the address conversion map 23 (step A1) and assigns one of the logical addresses designated by the write request. It is checked whether or not the physical address is associated (step A2).

If the physical address is associated (YES in step A2), the access control section 21 writes the data to the physical address (step A3). On the other hand, if the physical address is not associated yet (NO in step A2), the access control unit 21 refers to the free area management table 24 to search for a free area in the disk 30 (step A4). Then, the access control unit 2
The No. 1 writes data in this searched area (step A5), updates the address conversion map 23 (step A6), and writes the changed portion to the disk 30 (step A7).

After that, the access control unit 21 checks whether or not the data to be written remains (step A8), and if there is (YES at step A8), the process from step A1 is repeated for the next data. And
If not (NO in step A8), this process ends.

That is, this disk controller 20
Realizes that the write-through process is executed in a procedure suitable for improving fault tolerance so that the address translation map 23 is not updated as much as possible, and it is not necessary to save the address translation map 23 in the disk 30 at the time of shutdown. Reduce the time required for shutdown.

It should be noted that when the host computer 10 issues a write request for 3 blocks from logical addresses 0 and the address conversion map is in the state shown in FIG. The writing of the minute data may be executed by the procedure shown in FIG.

That is, the access control unit 21 checks the physical addresses of the logical addresses 0 to 2 and finds that each is 2
When it is known that the data is 4, 16, 23, this writing is executed in the order of 16, 23, 24, that is, the data of the logical addresses 1, 2, 0 according to the physical arrangement. In this way, writing of logical addresses 2 and 0 is writing to physically continuous areas (physical addresses 23 and 24), and therefore can be processed by one instruction.

Next, with reference to FIG. 8 to FIG. 15, various measures for speeding up which the access control unit 21 carries out when associating any physical address with a logical address will be described. . In the following, it is assumed that the host computer 10 issues a write request for three blocks from logical addresses 0, as described above.

FIG. 8 is a conceptual diagram showing an example of the address conversion map 23. In the example of FIG. 8, logical addresses 0 to 2
No physical address is associated with any of the above. Therefore, as shown in FIG. 9, the access control unit 21 searches for continuous free areas of the same size and associates the physical addresses 100 to 102 with the logical addresses 0 to 2, respectively. FIG. 9 is a conceptual diagram showing data writing to a physical disk.

Next, consider the case where the address conversion map 23 is in the state shown in FIG. That is, logical address 0
This is a case where the physical address 20 is associated with the logical addresses, but the physical addresses are not associated with the logical addresses 1 and 2. Further, at this time, as shown in FIG. 11, it is assumed that the disk 30 has an empty area for two blocks following the physical address 20. In this case, the access control unit 21 associates the physical addresses 21 and 22 of the empty areas of two blocks following the physical address 20 with the logical addresses 1 and 2, respectively.

Next, referring to FIGS. 12 and 13, let us consider a case where a plurality of physical disks are managed by logical addresses.

Here, it is assumed that three disks 30, HDD1 to HDD3, are managed. Further, as shown in FIG. 12, the address conversion map 23 represents a state in which the physical address HDD1-0 is associated with the logical address 0 and the physical addresses are not yet associated with the logical addresses 1 and 2. . Also, at this time, the disk HDD
It is assumed that the area for two blocks following the physical address HDD1-0 in 1 is an empty area. In this case, as shown in FIG. 13, the access control unit 21 associates the physical addresses HDD1-1 and HDD2 of two blocks following the physical address HDD1-0 with the logical addresses 1 and 2, respectively.

In the case of the specification in which the load distribution to each disk is prioritized, the access control unit 21 sets the logical address 0 to the physical address HDD 1-0 as shown in FIG.
It is also effective to associate the logical addresses 1 and 2 with the vacant areas of the physical addresses HDD 2 and 3, respectively.

Next, consider a case where the disk HDDs 1 to 3 are striped every 4 blocks. Here, again, it is assumed that the address conversion map 23 is in the state shown in FIG.

In this case, the access control unit 21 has the function shown in FIG.
As shown in (1), a free area is searched from a block of the same stripe as the physical address HDD1-0 associated with the logical address 0, and the empty areas are associated with the logical addresses 1 and 2, respectively. That is, the block in the stripe for which subsequent access has been determined is the search target. As a result, the recalculation of the parity can be integrated at one time, and the overhead of the disk system can be significantly reduced.

In this way, the access control unit 21 makes various efforts for speeding up when associating a physical address with a logical address.

The invention of the present application is not limited to the above-described embodiment, and can be variously modified at the stage of implementation without departing from the scope of the invention. Furthermore, the embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent elements are deleted from all the constituent elements shown in the embodiment, the problem described in the section of the problem to be solved by the invention can be solved,
When the effects described in the section of the effects of the invention can be obtained, a structure in which this constituent element is deleted can be extracted as the invention.

[0041]

As described in detail above, according to the present invention, in principle, writing is performed on the currently assigned physical address without changing the address translation map, and it is also assigned to the logical address. If the physical address does not exist, a free area of the physical address is allocated on the spot, and the changed portion of the address translation map is written to the disk together with the data.

As a result, in the write-through processing in the system using the dynamically updated address translation map, the address translation map is not updated as much as possible.
It realizes to execute in a procedure suitable for improving fault tolerance, and shortens the time required for shutdown because it is not necessary to save the address translation map to disk at shutdown.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a disk controller according to an embodiment of the present invention.

FIG. 2 is a first conceptual diagram showing an example of an address translation map of the same embodiment.

FIG. 3 is a first diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 4 is a diagram showing an address conversion map after data writing by the disk controller of the embodiment.

FIG. 5 is an exemplary flowchart showing an operation procedure of the disk controller of the embodiment.

FIG. 6 is a second conceptual diagram showing an example of an address translation map of the same embodiment.

FIG. 7 is a second diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 8 is a third conceptual diagram showing an example of an address translation map of the same embodiment.

FIG. 9 is a third diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 10 is a fourth conceptual diagram showing an example of an address translation map of the same embodiment.

FIG. 11 is a fourth diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 12 is a fifth conceptual diagram showing an example of an address translation map of the same embodiment.

FIG. 13 is a fifth diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 14 is a sixth diagram showing how data is written to a disk by the disk controller of the embodiment.

FIG. 15 is a seventh diagram showing how data is written to a disk by the disk controller of the embodiment.

[Explanation of symbols]

10 ... Host computer 20 ... Disk controller 21 ... Access control unit 22 ... Cache memory 23, 31 ... Address conversion map 24 ... Free space management table 30 ... Disc

Claims (8)

[Claims] 1. A disk system that manages the correspondence between a logical address and a physical address using an address translation map, and when a write request is made to the disk, one of the physical addresses is associated with the specified logical address. Determination means for determining whether or not the address conversion map is referred to, and when it is determined by the determination means that no physical address is associated, a free area of the disk is searched for. , The physical address is associated with the logical address, registered in the address conversion map, the data is immediately written to the physical address, and the updated content of the address conversion map is written in the disk. A disk system having means. 2. When the logical addresses that are not associated with any of the physical addresses are continuous, the write control means searches the continuous free area of the disk and selects the physical addresses that are associated with them. The disk system according to claim 1, further comprising: 3. The write control means is configured such that a logical address associated with any one of the physical addresses and a logical address not associated with any of the physical addresses are continuous, and the logical address of the preceding destination. 2. The disk system according to claim 1, further comprising means for allocating the free area to the subsequent logical address when the area of the disk adjacent to the physical address associated with is a free area. 4. The disk system is a system for managing a plurality of the disks and arranging data on each disk, wherein the write control means, when associating a physical address with a plurality of consecutive logical addresses, 3. A continuous area within the same disc is set as a search target.
The described disk system. 5. The disk system is a system for managing a plurality of the disks and arranging data on each disk, wherein the write control means associates a physical address with a logical address and an access amount of the entire system. The disk system according to claim 1, wherein the disk system is distributed to a plurality of disks so that the data is distributed. 6. The disk system according to claim 2, wherein the write control means includes means for executing writing of a plurality of data in the order of physical address arrangement regardless of the order of request. 7. The disk system is a system for managing a plurality of the disks and arranging data on each disk to be striped while sharing parity with all the other disks, wherein the write control means comprises: 3. The disk system according to claim 2, wherein when associating a logical address with a physical address, an area within a stripe whose subsequent access has been determined is a search target. 8. A disk access method of a disk system for managing correspondence between a logical address and a physical address by using an address conversion map, wherein when a write to the disk is requested, one of the specified logical addresses is added. Determining whether the physical addresses of the above are associated with each other by referring to the address conversion map, and when it is determined that none of the physical addresses are associated with each other, the free area of the disk is searched. Registering the physical address in the address translation map in association with the logical address, and immediately writing data to the physical address; and updating the address translation map to the disk. A disk access method comprising the step of recording.