JPH07248971A - Data reading processor - Google Patents

Abstract

PURPOSE:To provide a data reading processor which improves the access efficiency in consideration of the block that is pre-read by a pre-reading cache mechanism when the data are read out of such a storage as a magnetic disk that has the pre-reading cache mechanism. CONSTITUTION:In reference to a storage that has a cache mechanism and pre-reads a block to hold it in a cache, a cache data recognizing part 30 stores the pre-reading block information 32 to show the pre-reading block that is decided by the block read last. Then a reading processing part 31 carries out with preference the reading request of the block included in the information 32 among plural blocks when it carries out the reading requests of those blocks to the storage. Then the part 31 carries out the reading requests of the blocks which are not included in the information 32.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data read processing device for reading data from a storage device such as a magnetic disk having a prefetch cache mechanism.

[0002]

2. Description of the Related Art It is well known to provide a read-ahead cache mechanism in a magnetic disk storage device or the like, for example, a storage device that stores data in a rotating disk storage medium. SC
It is standardized in the SI standard as a READ AHEAD CACHE mechanism.

In a storage device having such a read-ahead cache mechanism, when a block of data requested by a read request is read, a predetermined number of blocks at storage positions following the last block are automatically read-ahead and read-ahead. Store in the cache of the cache mechanism.

For example, as shown in FIG. 2 (a), assuming that the first block to the 50th block can be stored in consecutive storage positions of the disk storage medium 1, for example, there is a read request for the 9th to 12th blocks. It is assumed that this block is read from the disk storage medium 1 and transferred to the request source.

In this case, for example, 8 blocks are automatically read from the 13th block and those 13th to 20th blocks are stored in the cache 2. When this operation is finished, the head for reading stored information on the disk storage medium 1 is at the position for reading the next 21st block.

When there is a next read request, the storage device identifies whether or not the requested block is pre-read in the cache, and if it is pre-read, the corresponding block is read from the cache without reading from the disk storage medium. Send to requester.

That is, in the state of the example of FIG.
If there is a read request for the 17th to 20th blocks, the blocks are not read from the disk storage medium 1 but are read from the cache 2 and sent to the request source.

By doing so, if the required block is in the cache, it can be read in a much shorter access time than in the case of the disk storage medium.
As is apparent from the above operation, the read-ahead cache mechanism is most effective when the blocks at consecutive storage locations are divided into several blocks and sequentially read.

On the other hand, as is well known, there are often read requests for discontinuous plural blocks. In that case, if the first request block is not prefetched in the cache, the prefetch block is not used, but the read request is not used. In, no consideration is given to blocks in the cache.

Further, when each storage device of the duplicated storage device has a prefetch cache mechanism, access is not performed in consideration of the blocks stored in those cache mechanisms.

It is an object of the present invention to provide a data read processing device which improves the access efficiency by taking into consideration the block that is prefetched in the prefetch cache mechanism of the storage device.

[0012]

FIG. 1 is a block diagram showing the configuration of the present invention. The figure shows the configuration of a data read processing device, which is a storage device having a mechanism for storing information on a storage medium, moving a head to select a storage position, and reading the stored contents, and having a cache mechanism. , If the block specified by the read request is held in the cache, read it from the cache; if not, read it from the storage medium and request the read block Is a device for reading a block from a storage device that is configured to read a predetermined number of blocks at a storage position subsequent to the block last read on the storage medium as a prefetch block and retain the block in the cache. Yes, first
In the invention, the cache data recognition unit 30 and the read processing unit 31 are included.

Each time the cache data recognizing unit 30 makes a read request to the storage device, the cache data recognizing unit 30 stores the read-ahead block information 32 indicating the read-ahead block determined from the last read block.

When a read request for a plurality of blocks is made to the storage device, the read processing unit 31 gives priority to a read request for the block contained in the prefetch block information 32 among the plurality of blocks, and then A read request for the block not included in the prefetch block information is issued.

In the second aspect of the invention, the cache data recognizing section 30 holds the storage position of the block following the prefetch block indicated by the prefetch block information as the head position 33.

When a read request is made for a block that is not included in the prefetch block information 32, the read processing unit 31 makes a read request in order from the block at the storage position closest to the head position 33.

In the third aspect of the present invention, the cache data recognizing unit 40 is a device for reading the block from the storage device of the system that operates as a duplicated volume by storing the same information in the two storage devices. And a device selection unit 44 and a read processing unit 41.

The cache data recognizing unit 40 has the same prefetch block information 42a, 42b and head positions 43a, 43 as described above.
and b are held for each storage device. Device selection unit 44
When reading a block from the storage device,
Read-ahead block information including the block that needs to be read
If the storage device corresponding to 42 is in an unused state, select the storage device, and if there is no pre-read block information including the block that needs to be read, then correspond to the head position 43 that is closest to the block that needs to be read. If the storage device to be used is not in use, the storage device is selected.

The read processing unit 41 issues a read request to the selected storage device.

[0020]

According to the data read processing device of the present invention, at the request source which makes a data read request to the storage device having the read-ahead cache mechanism, for example, the read-ahead block information for identifying the block which is expected to be read ahead in the cache is used, for example. At the end of each read request, for example, the block number of the last block is stored.

At the time of a read request to the storage device, it is discriminated whether or not there is a block included in the prefetch block in the request block, for example, 8 blocks following the block number held as prefetch block information, for example. , If there is a corresponding one, issue a read request for that block first.

Next, when a read request for another block is issued, the head position is immediately after the prefetch block,
Schedule the read requests in order from the block closest to the head position.

Also in the case of the dual storage device, if there is a request to read a block in the cache of any one of the storage devices, the storage device is used to access.
As described above, for a storage device having a read-ahead cache mechanism, information for identifying the state of the storage device is created and stored based on the read request issued by itself, and the state of the storage device is effectively used by the information. Thus, the access time for reading data can be shortened.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 3 is a diagram showing an example of the processing flow of the data read processing device of the first and second inventions.

When the read processing unit 31 of FIG. 1A receives a block read request from the data request source, it refers to the prefetch block information 32 in processing step 10 of FIG. 3 and includes the prefetch block in the request block. Identify if there are any

As the prefetch block information 32, the cache data recognizing unit 30 holds, for example, the number L of the last read block as described later, so that the cache capacity of the prefetch cache mechanism of the storage device is, for example, If there are 8 blocks, a request block satisfying L <request block number ≦ L + 8 is searched.

As a result, if there are blocks satisfying the above equation, those block numbers are arranged in ascending order in processing step 11, and a read request for those blocks is issued to the storage device.

Next, in processing step 12, it is discriminated whether or not there is a remaining block in the read request, and if there is no residue, the process proceeds to processing step 15. If there is any remaining residue, the head position 33 is referred to in processing step 13 and the remaining block Number of head position 33
Are arranged in the order close to the contents of, and a read request for those blocks is issued to the storage device in processing step 14.

If it is identified in the previous processing step 10 that there is no request block included in the prefetch block information,
Immediately, the process of process step 13 is started. When the issuance of the read request to the storage device is completed as described above, the cache data recognizing unit 30 determines the block number of the last read request in the prefetch block information 32 and the block number 9 blocks ahead of the head position in the processing step 15. Remember as 33.

By the above processing, for example, with respect to the storage device in the state of FIG. 2A, the read processing unit 31 receives read requests for the 50th, 20th, and 10th blocks by designating the blocks in this order. To do.

At this time, by executing the previous read request,
The cache data recognition unit 30 should store the block number “12” in the prefetch block information 32 and the block number “21” in the head position 33.

Therefore, the read processing unit 31 identifies that the 20th block of the request blocks is included in the prefetch block, first issues a read request for the 20th block to the storage device, and then the remaining 50th block. For the 10th block and the 10th block, a read request is issued to the storage device so that the 10th block close to the head position "21" is read first and then the 50th block is read.

FIG. 4 is a diagram showing an example of the flow of processing of the third invention. When the read processing unit 41 of FIG. 1B receives a block read request from the data request source, the device selection unit
44 is notified, and in the processing step 20 of FIG. 4, the device selection unit 44 identifies whether or not the prefetch block according to the prefetch block information 42a and 42b includes the request block.

If any of the request blocks is included, then in process step 21 it is determined whether the storage device corresponding to the information is in use to execute another request, and if not, the storage device is in use. Select.

If there is no prefetch block information including the request block in processing step 20, processing step
At 22, one of the head positions 43a and 43b having the block number closest to any of the requested blocks is selected.

In processing step 23, the selected head position
It is identified whether the storage device corresponding to 43a or 43b is in use, and if not, that storage device is selected. Processing step
21, processing step 23, if the device is in use, processing step
In 24, if the storage device on the other side is not in use, select the storage device on the other side.If the storage device on the other side is also in use, wait for the end of use and select the storage device on the side that is in the in-use state early. select.

When the device selection unit 44 selects a storage device by any of the above judgments, it notifies the read processing unit 41, and the read processing unit 41 issues a read request to the storage device in processing step 25. In that case, the reading order of the requested blocks is adjusted in the same manner as described with reference to FIG.

When the issuance of the read request to the storage device is completed as described above, the cache data recognizing unit 40 determines the block number of the last read request corresponding to the selected storage device in the processing step 26 as prefetch block information. 42a
Or 42b, and the block number 9 blocks ahead is stored in the head position 43a or 43b.

For example, as shown in FIG. 2B, there are duplicated storage devices 3a and 3b, and each storage device has storage media 1a and 1b.
b, the cache 2a and 2b of the read-ahead cache mechanism.

Here, as shown in the figure, in the storage device 3a, the third to tenth blocks are pre-read in the cache 2a, so that the head on the storage medium 1a is in the eleventh block. In 3b, the cache 2b has 11th to 18th
Suppose the block is read ahead, so the head is in block 19.

At this time, the information provided corresponding to the storage devices 3a and 3b is "2" for the prefetch block information 42a, "11" for the head position 43a, "10" for the prefetch block information 42b, and the head. "19" is stored in the position 43b.

If there is a read request for the 11th block, the head of the storage device 3a is located on the 11th block by the above processing, but the storage device 3b prefetches the 11th block into the cache. Therefore, a read request is issued to the storage device 3b unless it is in use.

[0043]

As is apparent from the above description, according to the present invention, for a storage device having a read-ahead cache mechanism, information for identifying the state of the storage device is created and stored based on a read request issued by itself. However, there is a remarkable industrial effect that the access time for data reading can be shortened by effectively utilizing the state of the storage device by the information.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of the present invention.

FIG. 2 is a diagram illustrating a read-ahead cache mechanism.

FIG. 3 is a flowchart of the process of the present invention (No. 1)

FIG. 4 is a flowchart of the process of the present invention (No. 2)

[Explanation of symbols]

 1, 1a, 1b Disk storage medium 2, 2a, 2b Cache 3, 3a, 3b Storage device 10-15, 20-26 Processing step 30, 40 Cache data recognition unit 31, 41 Read processing unit 32, 42a, 42b Look-ahead block Information 33, 43a, 43b Head position 44 Device selector

Claims (3)

[Claims] 1. A storage device having a mechanism for storing information on a storage medium, moving a head to select a storage position, and reading the stored contents, which has a cache mechanism and is designated by a read request. When the block is held in the cache, the block is read from the cache. When the block is not held in the cache, the block is read from the storage medium and the read block is transferred to the request source. A cache data recognizing unit for reading a block from a storage device configured to read a predetermined number of blocks at a storage position subsequent to the last read block as a prefetch block and hold the block in the cache. (30) and a read processing unit (31), and the cache data recognition unit (30) makes a read request to the storage device. Read-ahead block information that indicates door in, the tip reading block, which is determined from the last read block
(32) is stored, and when the read processing unit (31) issues a read request for a plurality of blocks to the storage device, the block included in the prefetch block information (32) among the plurality of blocks The data read processing device is configured so that the read request for the block that is not included in the prefetch block information is issued first, and then the read request for the block that is not included in the prefetch block information. 2. The cache data recognition unit (30) holds a storage position of a block following the prefetch block indicated by the prefetch block information (32) as a head position, and the read processing unit (31), The prefetch block information (3
2. When a read request is made for a block not included in 2), the read request is made in order from the block at the storage position closest to the head position (33). Data reading processor. 3. A cache data recognizing unit (40) for reading blocks from the storage device of a system that stores the same information in two storage devices according to claim 1 and operates as a duplicated volume. ), A device selection unit (44), and a read processing unit (41), and the cache data recognition unit (40) has prefetch block information (42a, 42b) and claim 2 according to claim 1. Head position
(43a, 43b) for each storage device, and when the device selection unit (44) reads a block from the storage device, the read-ahead block information (block including the block that needs to be read) If the storage device corresponding to 42a, 42b) is in an unused state, the storage device is selected, and if there is no prefetch block information including the block that needs to be read, the head that is closest to the block that needs to be read. If the storage device corresponding to the position (43a, 43b) is not in use, the storage device is selected, and the read processing unit (41) makes a read request to the selected storage device. A data read processing device characterized by being configured as follows.