JPH07104940A - Disk device - Google Patents

Abstract

PURPOSE:To provide a disk device in which a preserving capability (hit ratio) in a storage means 2 of data whose reading is requested from a host control means 10 to the disk device is improved. CONSTITUTION:This device is provided with a reading means 4 which reads disk storage data, transferring means 5 which transfers the read data to a host control means 10, presuming means 6 which presumes a read command to be transmitted the next by the host control means 10, first reading means 9 which first reads the disk storage data, and storage means 2 which records the read data. The presuming means 6 is provided with a first discriminating means 6a which discriminates that the previous and this time command is read commands and a second discriminating means 6b which discriminates the continuous relation of the end address of the previous read command with the leading address of the this time read command. Thus, the read command to be transmitted the next by the host control means 10 and the leading address of the command can be presumed according to the discriminated results of the first and second discriminating means 6n and 6b.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk device, and more particularly to predicting data requested to be read from the host device (host control means) which is a host device to the disk device, and preliminarily storing these data in the disk device. The present invention relates to a disk device whose operation is controlled so as to be stored in a storage means in the device.

[0002]

2. Description of the Related Art Conventionally, in a disk device in which the above-mentioned operation control is performed, that is, in a look-ahead control disk device for prefetching, the data transferred from the disk device to the host control means is cached in the disk device. If the host controller issues a data read request to the disk device and the requested data is stored in the cache memory, the data is stored in the cache memory. When the requested data is not stored in the cache memory, the data is read out and transferred to the host control unit. When the requested data is not stored in the cache memory, the data is newly read out from the disk and transferred to the host control unit. To control the operation like
(Referred to as operation control means), and the next data of the data transferred from the disk device to the host control means will be requested to be read from the host control means to the disk device next time. There are known two operation control means for controlling the operation so that the data next to the data transferred by the host control means is stored in the cache memory in the disk device (hereinafter referred to as the second operation control means). It was

[0003]

However, in both the known first operation control means and the known second operation control means, the data to be stored in the cache memory (storage means) in the disk device is Since each is uniquely determined, the probability (hit rate) that the data requested to be read from the disk drive from the host control means is stored in the cache memory cannot be increased to a certain level or higher. However, there was a problem that the hit rate was naturally limited.

The present invention eliminates the above-mentioned problems, and its object is the probability that the data requested to be read from the disk device by the host control means is stored in the storage means (hit rate). ) Is provided.

[0005]

In order to achieve the above object, the present invention provides a read means for reading data recorded on a disk based on a read command supplied from a host control means, and the read means. Transfer means for transferring the stored data to the host control means, a guessing means for guessing a read command to be sent next by the host control means, and a pre-reading means for pre-reading the information recorded on the disk by the output of the guessing means. And a storage device for recording the data read by the prefetching device and / or the reading device, and a control device for controlling the entire device. The first discriminating means for discriminating whether or not both the read commands are read commands and the previous read command. A second discriminating means for discriminating a continuous relationship between the end address of the output command and the head address of the present read command is provided, and based on the discrimination results of the first discriminating means and the second discriminating means, The host control means comprises means for estimating the read command to be sent next and the start address of the read command.

[0006]

According to the above-mentioned means, when predicting the next command to be supplied to the disk device from the host control means, the disk device performs the determination by the plurality of determination means separately and then Since the prediction is performed, highly precise data processing, specifically, whether it is a continuous data read request or a non-continuous (random / repeated) data read request, can be accurately determined. It becomes possible to remarkably improve the probability (hit rate) that the data requested to be read from the host control means to the disk device is stored in the storage means.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings.

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

In FIG. 1, reference numeral 1 is a magnetic disk device (H
DA), 2 is a cache memory (storage unit) that serves as a data buffer, 3 is a control unit (control unit), 4 is a read operation control unit (reading unit), and 5 is a host-buffer data transfer control unit. (Transfer means), 6 is a command analysis section (estimation means), 6a is a read (read) command discrimination section (first discrimination means), 6b is an address continuity relation discrimination section (second discrimination means), and 7 is a cache. The data retrieval unit, 7a is a segment discrimination unit (third discrimination unit),
7b is an all-data presence / absence determining unit (fourth determining unit), 7c is a size determining unit (fifth determining unit), 8 is a cache data management table, 9 is a prefetch data determining unit, and 10 is a host control system (host control). Means). The pre-reading is performed by the read operation control unit 4 based on the data determined by the pre-reading data determination unit 9 under the control of the control unit 3. That is, the prefetching unit is composed of the control unit 3, the read operation control unit 4, and the prefetch data determination unit 9.

The magnetic disk device 1 includes a magnetic disk (not shown) on which data is recorded and reproduced.
Under the control of the read operation control unit 4, predetermined recording data is read from the magnetic disk. As will be described later in detail, the cache memory 2 is divided into a plurality of segments, the number of divided segments and the size (capacity) of each segment can be changed, and the data read from the magnetic disk is The temporary storage data is temporarily stored, and the temporary storage data is stored under the control of the host-buffer data transfer control unit 5.
Transferred to.

Further, the command analysis section 6 is provided with a read command discrimination section 6a and an address continuity relation discrimination section 6b, and has a function of analyzing a command from the host control system 10 and a read command discrimination section 6a and an address continuation section. It has a function of estimating that the next read command will be received subsequent to the read command previously received from the host control system 10, based on the determination result of the relationship determining unit 6b. Here, the read command determination unit 6a determines whether the previously received command and the command received this time are both read commands, and the address continuity determination unit 6b determines the end address of the previous read command. It is determined whether or not there is a continuous relationship with the start address of this read command. Cache data search unit 7
Is a segment discriminator 7a, an all data presence / absence discriminator 7b
The segment discriminating unit 7a is provided with a size discriminating unit 7c, and in cooperation with the cache data management table 8, the segment discriminating unit 7a discriminates which segment of the cache memory 2 the head address of the read command of this time is. The data presence / absence discriminating unit 7b discriminates whether or not the data of all the sectors requested to be read by the present read command can be covered by the data accumulated in the segment, and the size discriminating unit 7c makes the segment discriminator Regardless of the determination result of the determination unit 7a, the size of the data of the total number of sectors requested to be read by the current read command and the divided segment size are compared and determined.

Further, the prefetch data determination unit 9 prefetches the data stored in the magnetic disk via the read operation control unit 4 based on the estimation of the arrival of the read command performed by the command analysis unit 6. Then, the prefetched data is transferred to the cache memory 2. Control unit 3
Is responsible for the overall operation control of this disk device.

Next, FIG. 2 is an explanatory diagram showing an example of a segment division state in the cache memory 2.
(A) shows an example of segment division, and (b) shows an example of segment number / size change.

As shown in FIG. 2A, the cache memory 2 is segment-managed, for example, with a buffer size of 256 KB (kilobytes), each of which has a size of 64.
KB four segments SEG. 0 to SEG. 3 and each segment SEG. 0 to SEG. Each 3 is assigned as a unit to the read cache / look ahead. When the disk device receives a Set Features command from the host control system 10, it is controlled by the control unit 3 and, as shown in FIG. The number / segment size (capacity) is changed appropriately, but from the host control system 10 to FIG.
When a number other than the number of segments shown in is specified,
The cache memory 2 rejects the request for changing the number of segments and keeps the current number of segments.

It should be noted that the segment division state is naturally sent to the cache data management table 8 as well.

The cache memory 2 is operated according to some rules as described below. First,
When the number of data blocks specified by the read command from the host control system 10 (data size of all requested sectors) exceeds the segment size of the cache memory 2, two or more segments are used. Rather, one segment is used in the ring buffer format. Secondly, the contents of each segment are always discarded in chronological order (LRU method). Third, the conditions under which the read cache / lookahead are discarded are:
At reset ,. When the number of segments is changed by the command ,. When look ahead is prohibited by a command, etc.,. When executing Read Long, Write Long, Format, Format, Write Buffer command ,. Diagnos
tec) command execution ,. When the specified address of the write command overlaps with the data address in the buffer. In the above cases, all segments are discarded, and in the above case, only the corresponding segment is discarded.

Next, FIGS. 3 and 4 are flow charts showing various discriminating functions executed in the disk apparatus of this embodiment, FIG. 3 is a part thereof, and FIG. 4 is the remaining one. It is a department. Note that this flowchart is shown in FIG.
4 is divided into two drawings as shown in FIG. 4, but in reality, the portions indicated by arrows and double arrows in FIG.
Are respectively connected to the portions shown by arrows and double arrows.

The overall operation of the disk device of this embodiment will now be described with reference to the flow charts shown in FIGS. 3 and 4. Here, the disk device causes the host control system 10 to issue the previous command. The disk device will be described as assuming that it has received the command and then received the command again from the host control system 10 after performing the data processing corresponding to the command.

First, in step S1, the command analysis section 6 determines whether or not the command received this time is a read command. When it is determined that the command is a read command (Y), the process proceeds to the next step S2. If it is determined that the command is a command other than the read command (N), data processing according to the content of the command is performed.

Next, in step S2, the read command determination unit 6a determines whether the previously received command is the same read command as the command received this time, and determines that it is a read command (Y). Sometimes
When it is determined that the command is a command other than the read command (N), the process proceeds to the next step S3, and another step S3 is executed.
Move to 7.

In the next step S3, the address continuity relation judging unit 6b judges whether or not the end address of the read command received last time and the start address of the read command received this time have a continuous (sequential) relationship. If it is determined that there is a continuous relationship (Y), the process proceeds to the next step S4, and if it is determined that there is no continuous relationship (N), the process proceeds to another step S7.

Next, in step S4, the segment discriminating unit 7a determines whether or not the start address of the read command received this time hits any of the cache segments in the cache memory 2, and the start address is determined. When it is determined that the cache segment is hit (Y), the process proceeds to the next step S5, and when it is determined that the start address does not hit any of the cache segments (N),
The following fourth data processing is performed.

Subsequently, in step S5, the all-data presence / absence discriminating unit 7b determines whether the data of all the sectors requested to be read by the read command received this time can be covered by the data already stored in the cache segment. If it is determined that the data of all the sectors can be covered by the data stored in one segment (Y), the process proceeds to the next step S6, and the data stored in the segment cannot be covered. If it is determined (N), the following third data processing is performed.

In the next step S6, the size discriminator 7c determines whether or not the size of the data of the total number of sectors requested to be read by the read command received this time is smaller than the divided segment size, and the total size of all sectors is determined. When it is determined that the number data size is smaller than the segment size (Y), the first data processing described below is performed, and when it is determined that the data size of all sectors is equal to the segment size (N), The second data processing described in (1) is performed.

In step S7, the segment discriminator 7a determines whether or not the start address of the read command received this time hits any one of the cache segments in the cache memory 2, and the start address is determined. When it is determined that the address hits any cache segment (Y), the process proceeds to the next step S8, and when it is determined that the leading address does not hit any cache segment (N), The process moves to another step S11.

Then, in step S8, the all data presence / absence determining unit 7b determines whether or not the data of all the sectors requested by the read command received this time can be covered by the data already stored in the cache segment. If it is determined that the data of all the sectors can be covered by the data stored in one segment (Y), the process proceeds to the next step S9, and the data stored in the segment cannot be covered. Judged (N)
At times, the process proceeds to another step S10.

In the next step S9, the size discriminating unit 7c determines whether or not the size of the data of the total number of sectors requested by the read command received this time is smaller than the segment size of the divided sectors, and the total number of sectors is determined. When it is determined that the data size of is smaller than the segment size (Y), the following fifth data processing is performed, and when it is determined that the data size of all the sectors is equal to the segment size (N), the following data processing is performed. The sixth data processing to be described is performed.

Further, in the other step S10, the size discriminating unit 7c compares the data size of the total number of sectors requested by the read command received this time with the divided segment size, and the result of the comparison. When the data size of all sectors is smaller than the segment size, the following seventh data processing is performed, and when the data size of all sectors is equal to the segment size, the eighth data processing described below is performed. If the data size of the total number of sectors is larger than the segment size, the ninth data processing described below is performed.

Finally, in the other step S11, the size discriminator 7c compares the size of the data of the total number of sectors requested by the read command received this time with the segment size, and the comparison is made. As a result, when the data size of all sectors is smaller than the segment size, the following tenth data processing is performed, and when the data size of all sectors is equal to the segment size, the eleventh data processing described below is performed. If the data size of the total number of sectors is larger than the segment size, the twelfth data process described below is performed.

Here, each of the first to twelfth data processing will be described.

First, in the first data processing, the command received this time is a read command like the command received last time, and the end address of the read command received last time and the start address of the read command received this time are set. There is a continuous (sequential) relationship between them, and the start address of the read command received this time is the cache memory 2
One of the caches in the cache has been hit, and the data for all sectors requested to be read by the read command received this time can be covered by the data already stored in the segment, and read by the read command received this time. When the data size of the requested total number of sectors is smaller than the segment size divided,
In this case, the read command starts from the start address to the end address, resulting in a complete hit state, and the host-buffer data transfer control unit 5 reads this hit data from the cache memory 2 and transfers it to the host control system 10 side. To do. At this time, since the read commands are in a continuous relationship, while transferring data to the host control system 10 side based on the read command, from the cache end position of the stored (LAH) end in the segment to the command end position (read request end). Additional look ahead (LAH) up to the position).

FIG. 6 shows an example of the state of the stored data in the segment in the first data processing. Figure 6
(A) and (b) are the states before and after data transfer, respectively, in the figure, S indicates the start pointer of the cache, and R indicates the area where the data requested by the read command this time is stored. Show. In this example, starting from S, 1
Cache data is stored over the entire segment, the start pointer S and the start address requested by the read command match at position A, and position B is the end address (command end position). The host-buffer data transfer control unit 5 adds LAH from the position A to the position B indicating the cache end position while transferring the hit data from the position A to the position B to the host control system 10. At this time, the start pointer S is at the position A
To position B from. The position of the start pointer S is managed by the cache data management table 8.

Next, in the second data processing, the command received this time is a read command like the command received last time, and between the end address of the read command received last time and the start address of the read command received this time. , The head address of the read command received this time hits one of the caches in the cache memory 2, and the data of all sectors requested to be read by the read command received this time is already in the segment. When the data size of the total number of sectors requested to be read by the read command received this time is equal to the divided segment size, even in this case, the start address of the read command can be covered. Complete hit state from the end address to the end address Will, between the host buffer data transfer controller 5, the cache memory 2 of this hit data
Read out and transferred to the host control system 10 side.
At this time, since the read commands have a continuous relationship,
Host control system 1 for data based on read command
While transferring to the 0 side, additional look-ahead (LAH) is performed for the segment size from the cache end position to the command end position in the segment. At this time, since the data size of the requested total number of sectors is equal to the segment size, the position of the start pointer S does not change.

In the subsequent third data processing, the command received this time is a read command as well as the command received last time, and between the end address of the read command received last time and the start address of the read command received this time. The read command received this time has a start address hitting one of the caches in the cache memory 2, but the data size of all sectors requested to be read by the read command received this time is already a segment. In the case where the data stored in the area cannot be covered, in this case, the first half hit state in which the read command is hit from the start to the middle becomes a hit state, and the host-buffer data transfer control unit 5 makes this hit. Read the data in the cache memory from the cache memory 2 and The read operation control unit 4 individually reads the remaining unhit data from the magnetic disk device 1 while transferring the data to the system 10 side, and adds the data in the segment in the ring buffer format to the host control system 10 side. Forward. At this time, after the data is transferred to the host control system 10, one segment of the data starting from the next head block (sector) is look ahead (LAH) starting from the command end position.

Next, in the fourth data processing, the command received this time is a read command like the command received last time, and between the end address of the read command received last time and the start address of the read command received this time. However, if the start address of the read command received this time does not hit any of the caches in the cache memory 2, and in this case, the middle part of the read command is hit. Even if there is a hit, it is determined that there is no hit from the start of the command, the segment in which the oldest data is stored (data in the segment) is discarded by the LRU method, and the read operation control unit 4 has made a hit request. Read the missing data from the magnetic disk unit 1 into the discarded segment of the data. While watching, the host control system 10
Transfer to the side. At this time, after the data is transferred to the host control system 10, the entire area of one segment is look ahead (LAH) from the next head block.

Next, in the fifth data processing, the command received last time is not the same read command as the command received this time, or the end address of the read command received last time and the start address of the read command received this time are set. Is not continuous, that is, has a random or repeat relationship, and the start address of the read command received this time hits any cache in the cache memory 2, and a read request is issued by the read command received this time. The data of all sectors can be covered by the data already stored in the segment, and the data size of all sectors requested to be read by the read command received this time is smaller than the divided segment size. And in this case the start of the read command Become fully hit status until et termination is hit, between the host buffer the data transfer control unit 5
Reads this hit data from the cache memory 2 and transfers it to the host control system 10 side. However, at this time, leave the transferred data in the segment,
Do not look ahead (LAH). In order to evaluate the performance of the disk device, there is a high possibility that the host control system 10 performs a test called a core test in which the host control system 10 sends a read command with the same start address and end address next time. This is because.

Next, in the sixth data processing, the command received last time is not the same read command as the command received this time, or the end address of the read command received last time and the start address of the read command received this time are set. Is not continuous, that is, has a random or repeat relationship, and the start address of the read command received this time hits any cache in the cache memory 2, and a read request is issued by the read command received this time. When the data of all sectors can be covered by the data already stored in the segment and the data size of all sectors requested to be read by the read command received this time is equal to the divided segment size. So, in this case, with the fifth data processing As, become fully hit state until completion of hits from the start of the read command, between the host buffer the data transfer control unit 5 reads the hit data from the cache memory 2, the host control system 1
Transfer to 0 side. However, at this time as well, there is a high possibility of a core test, so the transferred data is left in the segment and not look ahead (LAH).

In the following seventh data processing, the command received last time is not the same read command as the command received this time, or between the end address of the read command received last time and the start address of the read command received this time. The read command is not continuous, that is, has a random or repeat relationship, and the start address of the read command received this time hits any cache in the cache memory 2, but a read request is issued by the read command received this time. The data of all sectors cannot be covered by the data already stored in the segment,
Moreover, when the data size of the total number of sectors requested to be read by the read command received this time is smaller than the divided segment size, in this case, the first half hit state in which the read command is hit from the start to the middle is hit. Data transfer control unit 5 between host and buffer
Reads the hit data from the cache memory 2 and transfers it to the host control system 10 side, and the read operation control unit 4 individually reads the remaining non-hit data from the magnetic disk device 1. reading,
The data is transferred to the host control system 10 side while being added to the data in the segment in the ring buffer format. At this time, all of the command size (all data requested by the read command) is cached and look ahead (LAH) to another area in the segment. That is, the first half hit portion and the data added for transfer to the host control system 10 side are left in the segment as a cache, and the amount of LAH is set to the remaining size in the segment.

Next, in the eighth data processing, the command received last time is not the same read command as the command received this time or the end address of the read command received last time and the start address of the read command received this time are set. However, the read command received this time receives a read request because the start address of the read command received this time hits one of the caches in the cache memory 2. The data of the total number of sectors created cannot be covered by the data already stored in the segment,
Moreover, in the case where the data size of the total number of sectors requested to be read by the read command received this time is equal to the segment size, and in this case as well, the first half hit state in which the read command is hit halfway up to the middle Data transfer control unit 5 between host and buffer
Reads the hit data from the cache memory 2 and transfers it to the host control system 10 side, and the read operation control unit 4 individually reads the remaining non-hit data from the magnetic disk device 1. reading,
The data is transferred to the host control system 10 side while being added to the data in the segment in the ring buffer format. However, at this time, look ahead (LAH) is not performed. This is because if the first half hit portion and the data added for transfer to the host control system 10 side are left in the segment as a cache, there will be no free area in the segment.

Next, in the ninth data processing, the command received last time is not the same read command as the command received this time, or between the end address of the read command received last time and the start address of the read command received this time. The read command is not continuous, that is, has a random or repeat relationship, and the start address of the read command received this time hits any cache in the cache memory 2, but a read request is issued by the read command received this time. If the data of all sectors cannot be covered by the data already stored in the segment and the data size of all sectors requested to be read by the read command received this time is larger than the segment size of the divided segments. Yes, in this case, the third day In the same manner as the processing, becomes the first half hit state that up to the middle from the start of the read command has been hit,
The host-buffer data transfer control unit 5 reads the hit data from the cache memory 2 and transfers it to the host control system 10 side, while the read operation control unit 4 reads the remaining unhit data. Are individually read from the magnetic disk device 1 and transferred to the host control system 10 side while being added to the data in the segment in the ring buffer format. At this time, after the data is transferred to the host control system 10, the position of the segment corresponding to the command end position is used as the starting point, and one segment from the next head block (sector) is look ahead (LAH).

In the following tenth data processing, the command received last time is not the same read command as the command received this time, or between the end address of the read command received last time and the start address of the read command received this time. It does not have a continuous relationship, that is, it has a random or repeat relationship, the start address of the read command received this time does not hit any cache in the cache memory 2, and furthermore, it is read by the read command received this time. In the case where the data size of the requested total number of sectors is smaller than the divided segment size, in this case, even if the middle part of the read command is hit, a no-hit state where no hit occurs from the start of the command, Segment is the oldest data by LRU method It is discarded from being stored,
The read operation control unit 4 reads the requested non-hit data from the magnetic disk device 1 into the discarded segment of the data while the host control system 1
Transfer to 0 side. At this time, the data transferred to the host control system 10 side in order to cache all command sizes is left in the segment, and another area in the segment is look ahead (LAH).

Next, in the eleventh data processing, the command received last time is not the same read command as the command received this time, or between the end address of the read command received last time and the start address of the read command received this time. Is not continuous, that is, has a random or repeat relationship, the start address of the read command received this time does not hit any cache in the cache memory 2, and the read command received this time In the case where the data size of the total number of read-requested sectors is equal to the divided segment size, in this case, as in the tenth data process, there is no hit from the start of the command, and the LRU Depending on the method, the segment stores the oldest data. Is discarded from the ones, the read operation control unit 4, while data that is not hit has been requested from the magnetic disk device 1 reads in the discarded segments of the data are transferred to the host control system 10 side. However, at this time, since the data transferred to the host control system 10 side is left in the segment as a cache, look ahead (LAH) is not performed.

Further, in the twelfth data processing, the command received last time is not the same read command as the command received this time, or between the end address of the read command received last time and the start address of the read command received this time. It does not have a continuous relationship, that is, it has a random or repeat relationship, the start address of the read command received this time does not hit any cache in the cache memory 2, and furthermore, it is read by the read command received this time. In the case where the data size of the requested total number of sectors is larger than the divided segment size, in this case as well, as in the case of the fourth data processing, there is no hit from the start of the command, and the LRU Depending on the method, the segment stores the oldest data Discarded from what are, read operation control unit 4, while data that is not hit has been requested from the magnetic disk device 1 reads in the discarded segments of the data are transferred to the host control system 10 side. At this time, after the data is transferred to the host control system 10, the entire area of one segment is look ahead (LAH) from the next head block. In the seventh, eighth, tenth, and eleventh data processing, the reason why the transferred data is left as a cache in the segment after the requested data is transferred is the fifth and sixth.
This is to prepare for the core test, as in the data processing of.

FIG. 5 is an explanatory diagram showing a cache / look ahead operation matrix in which the above first to twelfth data processes are summarized in a list.

In FIG. 5, the numbers shown in circles respectively indicate the first to twelfth data processing.

In the first to twelfth data processing, not only all of the read commands do not hit among the no hits, but also the intermediate hit state in which the read command is hit midway, the read command It also includes the latter half of the hit state where the game is hit from the middle to the end. By handling no hits in this way, it is only necessary to determine the hit address by the start address, so that efficient processing is possible and, as a result, the processing time can be shortened.

As described above, according to the present embodiment, when predicting the next command to be supplied from the host control system 1 to the disk device, the disk device will make a plurality of different judgments, and then perform the judgment. Since the command is predicted, it is a read request for data that has a continuous relationship with the data based on the previous read command, or a non-continuous (random / repeated) data such as a core test. It is possible to accurately determine whether it is a read request, and to significantly improve the probability (hit rate) that the data requested to be read from the host control system 1 to the disk device is stored in the cache memory 2. You can

[0048]

As described above, according to the present invention,
When predicting the next command to be supplied from the host control means to the disk device, the disk device will be
The determination means, preferably the first to fifth determination means, makes the respective determinations, and then the command is predicted, so that the data based on the previous read command has a continuous relationship. It is possible to increase the accuracy of determination as to whether it is a read request for certain data or a read request for non-continuous (random / repeated) data, and the data requested to be read from the disk drive from the host control means is stored. Probability of being stored in the means (hit rate)
There is an effect that it can be remarkably improved.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram showing an example of a segment division state in a cache memory 2.

FIG. 3 is a part of a flowchart showing various determination functions executed in the disk device of this embodiment.

FIG. 4 is the remaining part of the flowchart showing various determination functions executed in the disk device of this embodiment.

FIG. 5 is an explanatory diagram showing a cache / look-ahead operation matrix in which each data processing of the present embodiment is summarized in a list.

FIG. 6 is an explanatory diagram showing an example of stored data in a segment in the first data processing of the present embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 magnetic disk device 2 cache memory serving as data buffer (storage unit) 3 control unit (control unit) 4 read (read) operation control unit (reading unit) 5 host-buffer data transfer control unit (transfer unit) 6 command Analysis unit (estimation unit) 6a Read (read) command determination unit (first determination unit) 6b Address continuity relationship determination unit (second determination unit) 7 Cache data search unit 7a Segment determination unit (third determination unit) 7b All Data Existence Discrimination Unit (Fourth Discrimination Means) 7c Size Discrimination Unit (Fifth Discrimination Means) 8 Cache Data Management Table 9 Prefetch Data Determination Unit 10 Host Control System (Host Control Means)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Koji Amita, No.7 Yukiya Otsuka-cho, Ota-ku, Tokyo Alps Electric Co., Ltd. No. Alps Electric Co., Ltd.

Claims (5)

[Claims] 1. A read unit for reading data recorded on a disk based on a read command supplied from the host control unit, a transfer unit for transferring the read data to the host control unit, and the host. Estimating means for estimating the read command to be sent next by the control means, pre-reading means for pre-reading the information recorded on the disk by the output of this estimating means, and read-out means by the pre-reading means and / or the reading means. In a disk device having a storage unit for recording data and a control unit for controlling the entire device, the estimating unit determines whether the command received last time and the command received this time are both read commands. Determination means, the end address of the previous read command and this read command And a second discriminating means for discriminating a continuous relationship with the leading address of the first discriminating means. Based on the discriminating results of the first discriminating means and the second discriminating means, the host control means sends the read data next. A disk device which estimates the start address of a command and its read command. 2. The disk device according to claim 1, wherein the storage means is composed of at least one or more divided segments. 3. The disk device includes a third discriminating means for discriminating whether or not the start address of the read command this time hits any one of the divided segments of the storage means. The disk device according to claim 2. 4. The fourth discriminating means for discriminating whether or not the data of all the sectors requested to be read by the present read command can be covered by the data accumulated in the divided segment. The disk device according to claim 3, further comprising: 5. The disk device includes a fifth discriminating means for discriminating and discriminating between the size of the data of the total number of sectors requested to be read by the present read command and the divided segment size. The disk device described.