JPH09160813A - File managing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a file managing device whereby the through-put is improved, writing is executed at a high speed, the fault tolerance is kept and data can be rearranged so as to be suitable for high-speed data reading. SOLUTION: The optimum storing information 105 of a block is added to the respective blocks 104 which are stored in the respective segments SEG of a data area 100 in a transfer source, the blocks 104 are read every segment unit at the time of transferring data so as to be temporarily stored in queues 107, the block for which high-speed reading is requested, is stored in a high- speed area 101 based on optimum storing information 105 by an optimum arrangement information judging device 106, the block required normal-speed reading is stored in a standard area 102, the block which is not troubled by low-speed reading is stored in a low-speed area 103 and also the rearrangement is executed by using a garbage collection or compaction processing time so that the rearrangement of data suitable for high-speed writing and high-speed reading is realized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a file management device, and more particularly to a file management device for writing or reading data in a data storage medium such as a magnetic disk.

[0002]

2. Description of the Related Art LFS (Log-Structured File Sys) is used as a file system for improving high-speed writing and fault tolerance.
tem). In LFS, new data and data to be updated are continuously written in the form of a log. Therefore, since the head of the disk continuously operates, high speed writing is possible. In addition, since old data is also left as a log, fault tolerance is improved.

FIG. 5 is a diagram showing the structure of a log in a conventional LFS. As shown in FIG. 5, in LFS,
The data is managed in blocks of the log 10a. A header (Header par) is included in the log 10a.
t) 11a and a data part (Data part) 12a. The header portion 11a has an ID (identifier), Address (the position of the log), and Size.
(Data size), Version (Version), Date
(Date and time) is included. The data section 12a stores data in which information such as images and sounds is recorded.

In the LFS, not only the data to be newly written but also the data to be updated is continuously written in the free area of the disk in the unit of the log 10a, so that high speed writing is possible. Also, since old data is also saved in units of the log 10a, for example, if a crash occurs, the data can be restored by referring to the log 10a,
Fault tolerance is improved.

In LFS, when the disk area becomes insufficient, or when it is desired to explicitly increase the disk free area, an operation called garbage collection or compaction is performed. Garbage collection or compaction ignores unused data as invalid data, transfers used data as valid data, and transfers only valid data to another area to generate a valid data-only area. . Therefore, after the garbage collection or compaction is completed, the area where the original data is stored can be used as a free area.

FIG. 6 shows how to transfer data when performing garbage collection or compaction in LFS. In FIG. 6, 100 is the original data area, for example, the data area on the disk, 110 is the data area on the transfer destination disk, 104a is the data block on which garbage collection or compaction is performed, for example, data in log units, 107 Indicates a temporary data storage area on the memory. Note that the temporary data storage area 107 on this memory will be described below.
Is called a queue.

A data area of a normal disk is divided into units called segments (SEG), and each divided segment SEG has a constant data capacity. On the disc, the remaining number of free segments SEG is an index indicating how much free space the entire disc has.
In LFS, usually, a plurality of logs are stored in each segment SEG.

As shown in FIG. 6, when performing garbage collection, first, data is sequentially read from the original data area 100 to be garbage-collected in units of segments and temporarily stored in the queue 107. Then, the temporarily stored log 104 a is read from the queue 107 and written in the transfer destination area 110. By repeating the above-described operation, all the logs 104a targeted for garbage collection are transferred from the original data area 100 to the new data area 110.

FIG. 7 is a diagram showing data transfer at the time of garbage collection and compaction in the conventional LFS. In FIG. 7, 100 is an original data area before garbage collection, 110 is a new data area after garbage collection, VLD is a valid log, and DED is an invalid log.

As shown in FIG. 7, valid logs and invalid logs are mixed in each segment SEG of the original data area 100 before garbage collection and compaction are performed. By the data transfer operation shown in FIG. 6, garbage collection is performed on the original data area 100. At this time, the header part 11 of each log 10a
The validity of the data is determined based on the information stored in a, only the valid log VLD is transferred to the new data area 110, and the invalid log DED is discarded.

Therefore, as a result of performing garbage collection and compaction, a new data area 11
At 0, as shown in FIG. 7, only the valid log VLD is continuously written to each segment SEG of the disk, and the invalid log DED is truncated.

[0012]

By the way, the above-mentioned conventional LFS is not suitable for the multimedia file system for the following reason. First, since writing is continued steadily even if there are a plurality of write requests for different purposes, writing throughput is high, but it is difficult to arrange the data suitable for reading (for example, it is not a continuous arrangement). Next, generally, in a disc, higher-speed data input / output can be performed in the outer circumference, but in the above-described LFS, the arrangement according to the characteristics of such a disc cannot be performed. Further, many file systems for multimedia have realized high-speed reading of data, but there is a problem that few consider high-speed writing.

The present invention has been made in view of such circumstances, and an object thereof is to improve the throughput of a file management device by optimally arranging data, and to achieve high-speed writing and high fault tolerance of a conventional LFS. It is an object of the present invention to provide a file management device capable of rearranging data suitable for high-speed data reading while maintaining the above.

[0014]

In order to achieve the above object, the present invention is a file management device for controlling writing of data to a storage medium, wherein the storage medium is at least according to the attribute of the write data. A writing unit that is divided into two or more areas, and the write data is added with arrangement information indicating an area to which the data is to be written, and the write data is stored in a predetermined area indicated by the arrangement information. Have.

Further, according to the present invention, there is provided a file management device for controlling writing of data to a storage medium rotating at a predetermined angular velocity, wherein the storage medium is at least two or more areas based on a distance from a rotation axis. Further, the writing data is added with arrangement information according to the attribute of the writing data, and the writing data is stored in a predetermined area indicated by the arrangement information.

Further, in the present invention, the write data to be read at high speed is stored in a region far from the rotation axis.

Further, the present invention has means for rewriting the arrangement information of the write data according to the change of the attribute.

Further, the present invention has means for rearranging each write data on the above-mentioned storage medium based on the above-mentioned arrangement information added to each write data at the time of garbage collection or compaction processing.

According to the present invention, the storage medium is divided into a plurality of areas according to the attribute of the write data, and each write data has one attribute of the write data, for example, the write data according to the required data transfer rate. The arrangement information indicating the area to be arranged is added.

Further, according to the present invention, the storage medium which rotates at a predetermined angular velocity is divided into, for example, three regions according to the distance from the rotation axis, and based on the arrangement information added to each write data, The data that requires high-speed data transfer is arranged in a region farther from the rotation axis. As a result, the write data is arranged in the area to be arranged according to each attribute. At the time of the first data transfer,
That is, when data is first written to the recording medium, continuous writing is performed like the conventional LFS regardless of the arrangement information added to each write data.

Furthermore, according to the present invention, the write data is rearranged on the storage medium based on the arrangement information of each write data during the garbage collection or compaction processing. This rearrangement is performed as described above, and when the data arrangement location is moved during garbage collection or compaction processing, the arrangement of the write data on the storage medium can be optimized based on the arrangement information of each write data. it can. This allows for optimal placement of the data using garbage collection or compaction processing time.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION First Embodiment FIG. 1 is a diagram showing the structure of a data area of a file management apparatus according to the present invention. In FIG. 1, 101 is a high speed area, 102 is a standard area, and 103 is a low speed area.

Generally, a disk has different input / output speeds of data on the outer and inner circumferences. The data input / output speed becomes higher toward the outer circumference. Therefore, when higher speed data transfer is required, it is preferable to use the outer peripheral area of the disk. In order to realize this, as shown in FIG. 1, the data area of the disk is divided into a high speed area 101, a standard area 102 and a low speed area 103, respectively. When high-speed data transfer is required, the high-speed area 101 is used. When the data transfer speed may be low, the low-speed area 10 is used.
3 is used and the normal transfer rate is sufficient, the data area of the disk can be selectively used according to the purpose such that the standard area 102 is used.

FIG. 2 is a diagram showing a first embodiment of the file management apparatus according to the present invention, and is a diagram showing data transfer at the time of garbage collection and compaction in the file management apparatus according to the present invention. In FIG.
100 is the original data area before transfer, 101 is a high speed area, 102 is a standard area, 103 is a low speed area, 104
Is a data block to be transferred, 105 is optimal placement information indicating one attribute of data in each block, 106 is an optimal placement information determination device, and 107 is a temporary area queue on the memory.

The optimum layout information 105 added to each block 104 is one attribute of each block 104,
For example, it is data indicating an area in which the block 104 is to be arranged according to the required data transfer rate of the block 104. Here, the optimum arrangement information 105 is composed of, for example, 2-bit data. When the 2-bit optimum arrangement information 105 is (01), the high speed area 10
1 and (10) indicates the standard area 102,
In the case of (11), the low speed regions 103 are shown. As a result, the added 2-bit optimum placement information 105 is set to (01) in the block 104 that requires a high data transfer rate, and the added 2 bits in the block 104 that requires a low data transfer rate. The optimum placement information 105 is set to (11), and the added 2-bit optimum placement information 105 is set to (10) in the block 104 which may have a normal data transfer rate.

The optimum layout information determination device 106 reads the optimum layout information 105 added to each block 104, and controls based on this optimum layout information 105 so that each block 104 is arranged in the area to be arranged. . For example, when the optimal placement information 105 added to the block 104 is (01), the block 104 is controlled to be placed in the high-speed area 101, and the block 10
When the optimum placement information 105 added to 4 is (10),
Optimal placement information 1 added to the block 104 by controlling the block 104 to be placed in the standard area 102
When 05 is (11), the block 104 is controlled to be arranged in the low speed area 103.

As shown in FIG. 2, optimal allocation information 105 of data is added to each block 104. At the time of data transfer, data is read from the original data area 100 in units of segments and temporarily stored in the queue 107. Then, each block 104 is read from the queue 107, and the optimal placement information determination device 106 determines to which data area the block should be transferred based on the optimal placement information 105 stored in each block 104. In accordance with the determination result of the optimal placement information determination device 106, the blocks 104 are respectively assigned to the high speed area 10
1, the data is transferred to the standard area 102 or the low speed area 103.

Further, when the required data transfer rate changes depending on the situation, the optimum allocation information 105 added to the block 104 is rewritten so that the block 104 can be used for other areas at the next garbage collection and compaction processing. Relocated to. Further, for the block 104 to which the optimum arrangement information is not added, the block 104 may be transferred to a place convenient for the system. For example, when only the low speed area 103 remains, the block 104 to which the optimum arrangement information is not added may be transferred to this area.

As described above, according to this embodiment, the optimum storage information 105 of the block is stored in each block 104 stored in each segment SEG of the transfer source data area 100. At the time of data transfer, blocks are read in segment units and temporarily stored in the queue 107, and the optimum placement information determination device 106 transfers the blocks requested to be read at high speed based on the optimum storage information 105. The blocks are stored in the high-speed area 101 at the destination, the blocks requested to be read at the normal speed are stored in the standard area 102 at the transfer destination, and the blocks that may be read at low speed are stored in the low-speed area 103 at the transfer destination. Therefore, reading of data according to the required data transfer rate can be realized.

Second Embodiment FIG. 3 is a diagram showing a second embodiment of the file management apparatus according to the present invention, and is a diagram showing the structure of a log. In FIG. 3, 10 is a log, 11 is a log header (Head).
erpart), 12 is the data part of the log (Data p)
art) are shown.

As shown in FIG. 3, the header portion 1 of the log 10
1 is a conventional ID (identifier), Address (position of the log), Size (data size), Vers.
In addition to information such as ion (version) and Date (date and time),
Optimal placement information (New A that newly indicates the attribute of the log 10)
(tribute) is added. Log 10
The optimum placement information newly added to is, for example, log 1
It is composed of 2-bit data indicating the area where the log 10 is to be arranged according to the required transfer rate of 0. For example, the log 10 that should be placed in the high speed area 101
The 2-bit optimum placement information is set to (01) in the header part 11 of the log 10 and the 2-bit optimum placement information is set to (10) in the header part 11 of the log 10 to be placed in the standard area 102. The 2-bit optimum placement information is set to (11) in the header portion 11 of the log 10 to be placed in the low speed area 103.

When logs are transferred during garbage collection and compaction processing, a new allocation area for the log 10 is determined based on this optimum allocation information. For example, when the optimum placement information of the log 10 is (01), the log 10 is placed in the high speed area 101,
If the optimal allocation information of the log 10 is (10), the log 1 concerned
When 0 is allocated in the standard area 102 and the optimum allocation information of the log 10 is (11), the log 10 is in the low speed area 1
It is located at 03.

FIG. 4 is a diagram showing operations of LFS garbage collection and compaction in the file management apparatus according to the present invention. In FIG. 4, 100 is the original data area 10 before garbage collection.
0 and 101 are high-speed areas, 102 is a standard area, and 103
Is for low speed area, VLD is valid log, DED is invalid log,
FST indicates high-speed arrangement information (01), STD indicates standard arrangement information (10), and SLW indicates low-speed arrangement information (11).

As shown in FIG. 4, the valid log VLD and the invalid log DED are mixedly arranged in each segment SEG of the original data area 100 before the garbage collection and compaction are performed.

Garbage collection or compaction on the original data area 100 is performed by the data transfer operation shown in FIG. That is, the log is read from the original data area 100 in segment SEG units and temporarily stored in the queue 107 shown in FIG. 2, for example.

Then, for example, the version (Ver) stored in the header portion 11 of each log 10 stored in the queue 107 is stored.
(Sion) information, the validity of the log is determined, only the valid log VLD is selected, and the invalid log DED is discarded. Only the valid log VLD is processed by the optimum placement information determination device 106 shown in FIG.
Based on the optimum placement information added to the, the high speed area 101, the standard area 102, or the low speed area 103, respectively. That is, the log 10 for which a high transfer rate is requested is re-stored in the high-speed area 101, the log 10 for which the low transfer rate is good is in the low-speed area 103, and the log 10 whose normal transfer rate is good is in the standard area 102. Will be placed.

Further, when the required transfer rate of the predetermined log 10 changes depending on the situation, the optimum allocation information of the log 10 is rewritten, so that the log 10 can be changed to another one at the next garbage collection and compaction processing. Relocated to the area.

Therefore, as a result of garbage collection and compaction, only the valid log VLD is continuously written in each segment SEG of the new area based on the optimum allocation information added to the log, and the invalid log DED is discarded. To be done. This not only relocated each log to a data area according to the requested data transfer rate and made it readable at the requested rate, but garbage collection and compaction truncated the invalid data. The usable data area in the storage medium is expanded.

As described above, according to the present embodiment, in the LFS, the valid log VLD and the invalid log DED stored mixedly in each segment SEG of the original data area 100 are temporarily stored in the memory. The valid log VL is stored based on the information stored in the header of each log.
Only D is selected, and based on the optimum allocation information added to the header part of each log, the high speed area 101, the standard area 102, or the low speed area 103 is selected according to the requested data transfer rate of each log. Stored in the invalid log DE
Since all D's are discarded, not only can each log be read at the required data transfer rate as a result of garbage collection or compaction, but also the available data area on the storage medium can be expanded.

[0040]

As described above, according to the file management apparatus of the present invention, data can be optimally arranged by utilizing time such as garbage collection, and the data throughput in the file management apparatus can be improved. . Further, when applied to LFS, there is an advantage that data can be rearranged suitable for high-speed data reading while maintaining high-speed writing and high fault tolerance unique to LFS.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a data area of a file management device according to the present invention.

FIG. 2 is a diagram showing the first embodiment of the present invention, and is a diagram showing data transfer based on optimum allocation information.

FIG. 3 is a diagram illustrating a second embodiment of the present invention, and is a diagram illustrating a configuration of an LFS log to which optimal placement information is added.

FIG. 4 is a diagram showing LFS garbage collection and compaction in the present invention.

FIG. 5 is a diagram showing a structure of a log in a conventional LFS.

FIG. 6 is a diagram showing data transfer in a conventional LFS.

FIG. 7 is a diagram showing garbage collection and compaction in a conventional LFS.

[Explanation of Codes] 10 ... Log 11 ... Log Head 12 ... Log Data 100 ... Original Data Area 101 ... High Speed Area 102 ... Standard Area 103 ... Low Speed Area 104 ... Log etc. Data Block 105 ... Optimal placement information 106 ... Optimal placement information determination device 107 ... Queue 110 ... Data area of transfer destination VLD ... Valid log DED ... Invalid log FST ... High-speed placement information STD ... Standard placement information SLW ... Low-speed placement information

Claims (5)

[Claims] 1. A file management device for controlling writing of data to a storage medium, wherein the storage medium is divided into at least two or more areas according to an attribute of the write data, and Is a file management device having a writing unit to which layout information indicating an area to which the data is to be written is added and which stores the write data in a predetermined area indicated by the layout information. 2. A file management device for controlling writing of data to a storage medium rotating at a predetermined angular velocity, wherein the storage medium is divided into at least two or more areas based on a distance from a rotation axis, Further, a file management device having write means for adding layout information according to an attribute of the write data to the write data, and storing the write data in a predetermined area indicated by the layout information. 3. The file management device according to claim 2, wherein the write data to be read at high speed is stored in an area far from the rotation axis. 4. The file management device according to claim 2, further comprising means for rewriting the arrangement information of the write data in accordance with a change in the attribute. 5. The file management device according to claim 2, further comprising means for rearranging each write data on the storage medium based on the arrangement information added to each write data at the time of garbage collection or compaction processing. .