JPH10301847A - Data storage device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To assign necessary segment in accordance with access from a host device without futilely using any cache memory. SOLUTION: This storage device is provided with a storage medium 6 which stores data transmitted from the host device, cache memory 8 arranged side by side with the storage medium 6 which stores data written or data to be read by the host device as cache data with segment length, and controlling part 2 which stores the data in the cache memory 8 in response to the access of the writing or reading of the data to the storage medium 6 of the host device. Also, the controlling part 2 is provided with a segment assigning means 4 which assigns the segment length of the cache memory with the segment length preliminarily decided for each classification of access based on the classification of access from the host device.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a data storage device, and more particularly to a data storage device for storing data used by a computer. The data storage device refers to a rewritable direct access storage device such as a magnetic disk device and a magneto-optical disk device, and further includes a part of a non-rewritable direct access device such as an optical disk device. That is, the storage medium of the data storage device is a magnetic disk, a magneto-optical disk, an optical disk, or the like. In this case, it is preferable to use a semiconductor memory as the cache memory. Further, the present invention is applied to a semiconductor memory such as a PC card and a RAM attached to a CPU. Furthermore, it can be used in a disk array system using a plurality of magnetic disk devices.

[0002]

2. Description of the Related Art Conventionally, a data storage device such as a magnetic disk device uses a cache memory to speed up writing and reading. This is to use a medium that has a higher access speed than a magnetic disk as a cache memory, and when data is stored in the cache memory, read out the data from the cache memory to speed up the access.

For example, Japanese Patent Laying-Open No. 4-180140 discloses means for determining a segment capacity based on an average value of an access block length. The means in this publication determines the segment capacity using the average value of the access block length as a measure. In other words, the focus is on sequential access, and the cache hit rate is increased by assigning several times the average value of the access block length to the segment capacity.

[0004]

However, in the above-described conventional means for determining the segment capacity, when the higher-level device operates by an access method other than the sequential access, the segment determining method wastefully uses the capacity of the cache memory. Had the disadvantage of becoming Specifically, in the conventional method of changing the number of cache segments, the cache is not effectively used for a server where the number of clients changes frequently or for use on a network.

[0005] For example, an access method called reaccess is an operation of performing read / write to the same area many times.
In this case, since the segment length requires only the access block length, a cache memory larger than that is wasted.

Furthermore, in the case of read / write in an environment in which read-ahead and sweep-out operations are not performed, read / write processing must be completed within one command, so that a cache memory longer than the access block length of the command is wasted. Become.

[0007]

An object of the present invention is to improve the disadvantages of the prior art, and in particular to provide a data recording device which can be allocated to a necessary segment in response to an access from a host device without uselessly using a cache memory. To do so.

[0008]

Therefore, according to the present invention, there is provided a storage medium for storing data transmitted from a higher-level device, and data written or read by the higher-level device which is provided alongside the storage medium and which is stored in a predetermined segment. A cache memory that stores the cache data as long cache data; and a control unit that stores the data in the cache memory in response to data write or read access to the storage medium of the host device. In addition, the control unit is provided with a segment allocating means for allocating the segment length of the cache memory with a predetermined segment length for each type of access based on the type of access from the host device. This aims to achieve the above-mentioned object. In the present invention, the segment allocating means allocates the segment length of the cache memory based on the type of access from the higher-level device, using a segment length predetermined for each type of access. For this reason, re-access, access to a continuous area,
Type of access (attribute), such as one-time access
, A segment length of the cache memory corresponding to each access is allocated. Thus, an efficient cache is realized without wasting the capacity of the cache memory.

Further, in the present invention, the control unit includes: a cache control table for sequentially storing attributes of accesses from the host device; and a cache memory segment in accordance with the access attributes based on a change in the contents of the cache control table. And a table use segment allocating means for allocating a length. The attributes of the access from the host device are sequentially stored in the cache control table, and according to the change in the cache control table,
Change the segment configuration of the cache memory. For this reason, the cache automatically responding to a change in the configuration of the higher-level device or a change in the application used in the higher-level device is automatically realized.

Further, in the present invention, the control unit predicts the number of higher-level devices accessing the storage medium based on the type of access from the higher-level device,
It is possible to adopt a configuration in which a high-level device-specific segment allocating unit that allocates the segment length of the cache memory to each high-level device predicted by the high-level device number prediction unit is provided. The higher-level device number prediction means predicts the number of higher-level devices currently accessing the storage medium according to the type of access from the higher-level device. Then, a cache memory is allocated to each virtual host device based on the number of host devices (the number of virtual host devices). Then, even when connected to a network, the effect of the cache is evenly distributed to each higher-level device.

[0011]

Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a data storage device according to the present invention. As shown in FIG.
The data storage device includes a storage medium 6 for storing data transmitted from a higher-level device (host), and data written or read by the higher-level device provided alongside the storage medium 6 as cache data having a predetermined segment length. The cache memory 8 stores the data, and the control unit 2 stores the data in the cache memory 8 in response to data write or read access to the storage medium 6 of the host device. In addition, the control unit 2 includes a segment allocating unit 4 for allocating a segment length of the cache memory with a predetermined segment length for each type of access based on the type of access from the host device.

Since the segment allocating means 4 individually determines the segment length of the cache for each type of access, it is possible to further increase the hit rate without wasteful use of the cache memory. Is done faster.

FIG. 2 is a block diagram showing a detailed configuration of the segment allocating means 4. In one embodiment, the segment allocating unit 4 includes an access type classification function 10 for classifying an access type based on address information of the storage medium 6 to be accessed by the higher-level device and the data length of the accessed data; This access type classification function 10
And a segment length condition table 24 storing a segment length determined from the total capacity of the cache memory 8 and the data length of the accessed data for each access type classified according to.

The access type classification function 10 determines the type (attribute) of access from a higher-level device. FIG. 3 is an explanatory diagram showing an example of the type table. In the example shown in FIG.
The access type classification function 10 assigns an access for reading data (read attribute) to an Only attribute, an Area attribute, a Reac
cess attribute, Read sequential attribute, and Read large attribute. The Only (only) attribute is read access only once (read access), the Area (area) attribute is read access to a nearby area, the Reaccess (reaccess) attribute is access to the same area, and s
The equential (read sequential) attribute is a read access to a continuous area, and the Read large (read large) attribute is a read access to read data larger than a certain value.

Further, the access type classification function 10 defines an access (Write attribute) for writing data as Wri.
Classify into te attribute, Write sequential attribute and Write large attribute. Write attribute is write access, and Write sequential
Attribute is write access to the continuous area,
The ge (write large) attribute is a write access larger than a certain value. In addition, it has a Free attribute when the cache function is turned off.

The access type classification function 10 classifies various accesses by, for example, nine types of attributes shown in FIG.
The type of access is classified based on the address information of the storage medium 6 to be accessed by the host device and the data length of the accessed data. In this embodiment,
Based on the address information, it is determined whether the access is to the same area, to the continuous area, or to an adjacent area, and the type of access is determined based on the determination result.

Further, the access type classification function 10 has a function of monitoring the progress of a plurality of accesses from a host device and classifying the access type based on the progress information of the plurality of accesses. This allows, for example, On
When the ly (only) attribute (only one read access) occurs multiple times in a continuous area, it is determined to be the Read sequential attribute based on the progress information of the multiple accesses.

Specifically, when an access occurs, the following items are checked for attribute determination. First, a continuous area check and a reverse continuous area check are performed based on the address information to determine whether or not the data is sequential (or reverse sequential). Further, it is confirmed whether or not the access is to the same area based on the address information.
It is determined that the attribute. In addition, by performing a skip access check and a reverse skip access check based on the address information, it is determined whether or not the access is to an area slightly away from the front and back, and further, by performing an overlap access check. , It is determined whether or not the access is to an area where a part is overlapped. In these cases, the attribute is Area. The length of the data to be accessed is determined by performing a length check. If the data length is equal to or greater than the predetermined segment maximum value, the attribute becomes the Read large attribute or the Write large attribute.

When the access type classification function 10 determines the attribute of the access from the host device, the segment allocating means 4 allocates the segment of the cache memory 8 according to the access according to the designation of the segment length condition table. FIG. 4 is an explanatory diagram showing an example of the segment length condition table. As shown in FIG. 4, the segment length to be assigned for each attribute is determined by the relationship with the cache memory capacity and a predetermined margin.

For example, since the Reaccess attribute is for accessing the same area, an area longer than the command length (data length) is not required, so the length is set to a length (length) value. That is, the access type classification function 10 has a function of determining that the access type is a reaccess when the same address is accessed a plurality of times based on the progress information of the plurality of accesses. When the access type is reaccess, the condition table includes condition information in which the data length of the access is set as the segment length. This prevents useless use of the cache memory by allocating segments equal to or greater than the length value in the case of reaccess.

Since the Read / Write sequential attribute is infinitely desired considering the pre-reading / sweep-out portion, the attribute is set to the maximum value that can secure one segment in the cache memory 8. The Read / Write large attribute is set to the maximum value for which one segment can be secured because the command length is preferably the segment length but the cache capacity does not accompany it.
In the example shown in FIG. 4, a segment is assigned to this Read / Write large attribute only when the value of LRU (Least Recently Used, newness of use) is "0" and the access is the latest. The Write attribute is set to the maximum value because it is infinitely desirable considering the speed of sweeping from the host. Since the Free attribute is not required for the command length or more, it is set to the length (length) value. Only attribute and Are
For the a attribute, a margin is required because it is an attribute that will change in the future.

When the segment length is assigned to each access based on the segment length condition table shown in FIG. 4, a cache segment having a required data length can be assigned according to the attribute of the access. The access from the host can be speeded up by effectively using the capacity of the cache memory.

Access to the storage device changes one after another, such as a change in the access of the virtual host or an increase or decrease in the number of virtual hosts. Therefore, it is important to determine when to change the segment configuration. Here, the judgment is made based on the attribute, and the type and the number are used. Specifically, the segment allocating means 4 has a segment length changing function for changing the segment length of the cache memory 8 according to the access type when the access type changes based on the progress of a plurality of accesses from the host device. Prepare. With this segment length changing function, it is possible to quickly respond to access to the storage device that changes in various ways. In one embodiment, the segment length change function includes a segment configuration change condition table shown in FIG. 5 that stores a change condition of the segment configuration of the cache memory when the access type is continued a predetermined number of times. In the example shown in FIG. 5, when the number of attributes increases or decreases or the maximum access length of a certain attribute changes, the segment configuration is updated again based on the segment length condition table shown in FIG.

FIG. 6 is an explanatory diagram showing an example of the attribute assignment order table. When making segment assignments, care must be taken in the order of assignment. It requires a large cache area, but only needs to be read / write once.
This is because there are a ge attribute, a Read / Write sequential attribute that requires as much capacity as possible in consideration of subsequent commands, and the like. When allocating segments, it is necessary to consider the characteristics of these attributes to make an optimal segment configuration.

As shown in FIG. 6, first, a Read / Write large attribute that can be predicted to disappear from the segment configuration by one command is assigned. Next, Reaccess, Write, Free, Only with fixed capacity
Assign attributes. Finally, Read / Wri for unlimited capacity
Assign te sequential and Area attributes.

FIG. 7 is an explanatory diagram showing an example where the data storage device is connected to a plurality of higher-level devices. In the example illustrated in FIG. 7, the control unit 2 includes a cache control unit 2A that controls a cache and a CPU that performs overall control. The storage medium 6 is connected to the bus 5 via the cache memory 8.
4 is connected. The bus 54 includes a plurality of higher-level devices (host computers) 51, 52, and 53.

When the data storage device is connected to a plurality of higher-level devices, the segment allocating means 4 accesses the storage medium based on the access type classified by the access type classification function, as shown in FIG. A virtual host device prediction function 14 for predicting the number of host devices as the number of virtual host devices, and a segment of the cache memory is allocated to each virtual host device predicted by the virtual host device prediction function 14 with reference to the segment length condition table. And a higher-level device-specific segment allocation function 16. As described above, the cache memory is allocated by estimating the number of accessing higher-level devices, so that the effect of the cache can be equally provided to each higher-level device. To predict the number of devices,
More capacity can be allocated to each higher-level device than to allocate the segments evenly by the number of higher-level devices connected, and therefore, the capacity of the cache memory can be used as efficiently as possible.

In this embodiment, the access type classification function 10 monitors the progress of a plurality of accesses from a host device and classifies the access type based on the progress information of the plurality of accesses. A function of recording whether the access of the access type classified by the access type classification function has been used recently (LRU). In this case, the higher-level device-based segment allocation function 16 determines whether or not the access has been used recently (L
RU) to allocate a segment of the cache memory to a recently used access type. Then, optimal segment allocation can be dynamically performed according to a change in access from each higher-level device.

When a plurality of higher-level devices access the same data, it is possible to share a segment of the cache memory. That is, the segment allocating means 4 has a shared segment allocating function for sharing the segment with a plurality of virtual higher-level devices when the data of the segments allocated by the higher-level device-specific segment allocating function match. Thus, the same data can be allocated to another access without storing the same data in the cache memory.

Next, the operation of the configuration shown in FIG. 7 will be described. In the example shown in FIG. 7, a higher-level device (virtual host) that will be currently accessed is predicted. The segmentation of the cache memory 8 is made variable depending on the number of virtual hosts and the access method. For example, when three higher-level devices perform sequential access, three virtual hosts are generated in the cache control table 3 shown in FIG. Then, the cache control unit 2 recognizing it.
A divides the cache memory 8 into three equal-capacity segments and assigns them to respective virtual hosts.

The analysis information for determining a virtual host in the cache control table 3 is called a candidate. The contents of the candidate include an access history for determining that the access is a certain access. This includes access data indicating cumulative access, first access indicating first access, and back access indicating previous access.
When it is determined that the access is constant based on the access data, the attribute of the virtual host is changed. However, regardless of access, size of length, cache ON
Some attributes are determined by / OFF.

FIG. 8 shows analysis information (cache control table) for the types of attributes shown in FIG. The cache control table contains an access data item for each access,
It has a first access item, a back access item, an attribute (access type) item, an attribute change count item, an LRU data item, a use frequency item, and N items.

The LBA of the access data item (Logical
Block Address, logical block address) indicates the access start address, and Length indicates the number of blocks from the start LBA. First access indicates the content accessed first by this virtual host. The back access indicates the content of the previous access of the virtual host.

The attribute item indicates the type of access. As read attributes, the only attribute indicates read access before the change, the Areaa custom indicates read access to the neighboring area, the Reaccess attribute indicates read access to the same area, and the Read sequential attribute indicates read access to the continuous area. A write access is indicated, and a Read large attribute indicates a read access having a length larger than a certain value. As a write attribute, a Write attribute indicates a write access before a change, and a Wri
The te sequential attribute indicates a write access to a continuous area, and the Write large attribute indicates a write access of a length larger than a certain value. Free attribute is cached
Use when off or for other purposes.

The attribute change count item is a count up to a specified number at which the type of access (attribute) changes. Here, Rea (reaccess counter), Rseq (read sequential counter), and Wseq (write sequential counter) are provided.

The smaller the number of LRU (Least Recently Used) data items, the more recently they are used. When this number becomes a certain number, the cache control unit 2A does not update the segment length of the access. The usage frequency item indicates that the smaller the number is, the more frequently or the user has used it. The frequency of use is LR according to the following equation:
It is calculated when U = 0.

New usage frequency = (old usage frequency + old LRU) / 2

The N (Nomination) item is a candidate information valid bit.

When segment allocation is performed, it is inefficient to use all candidates in the cache control table. This is because an old candidate (garbage candidate) is included in the cache control table, and it can be predicted that the candidate virtual host will no longer access. Therefore, in order not to use the garbage candidates, the search is performed in ascending order of the candidate LRU, and up to a portion where the difference between the LRUs is large or up to the maximum number of allocated segments is set as a use candidate.

When allocating a segment to a virtual host, what is decisive is the attribute. Since the attribute indicates the access type of the virtual host, it is possible to determine how much cache capacity the virtual host requires. As shown in FIG. 4, since the Reaccess attribute accesses the same area, an area larger than the command length is unnecessary. I want the Read / Write sequential attribute to be infinite considering the pre-reading / sweep-out part. For the Read / Write large attribute, the command length is desired, but the cache capacity does not accompany it. I want the Write attribute to be infinite considering the sweeping speed. The Free attribute is unnecessary when the command length is longer than the command length, and the Only attribute and the Area attribute require a margin since they are attributes that will change from now.

The use of one segment by a plurality of virtual hosts is herein referred to as segment sharing. This is useful for data that is used only once and attributes that can be predicted. By doing so, the cache area can be effectively used, and the number of times of segment allocation execution can be reduced. However, in order to prevent the destruction of the pre-read data and the destruction of the continuous sweeping position, the access type is limited to the attribute before change whose indetermination is undetermined.

The function of predicting prefetch data for an empty cache segment is called prefetch range prediction. The pre-read data is a range for a plurality of read commands predicted to be issued to the storage device in the future. Read sequent is the attribute for which the prefetch range can be predicted for the candidate to which the segment is assigned
The ial attribute and the Area attribute.

The Read sequential attribute is an attribute for predicting access to a continuous area, and the Area attribute is an attribute for measuring access to a nearby area. Therefore, the read-ahead range can be predicted. The Read sequential attribute performs pre-reading measurement of all data after the accessed command LBA. Area attribute is
A front neighborhood range and a rear neighborhood range are calculated from the access data and first access of the other candidate Area attribute to obtain an average value. Then, the average value is used as a prefetch range in addition to the accessing command LBA. However, if there is only a candidate being accessed in the Area attribute in the cache control table, or in the case of pure pre-reading / post-reading, the pre-reading range is predicted only by the front neighboring range and the rear neighboring range of the accessing candidate. .

[0044]

Embodiment 1 Next, an embodiment under certain conditions will be described. In the first embodiment, both the read cache and the write cache are turned on. The attribute occurrence candidate use count (the specified value of the attribute change count) is the Read sequential attribute, Write
Both the sequential attribute and the Reaccess attribute are set to three times. The maximum index frequency is 7Fh, and the address of the cache memory is from address 0 to 3FCh blocks. Only
Attribute margin is 20h block, Write attribute current capacity is 100h block, large attribute occurrence start block number is 200h block, Area attribute occurrence average use frequency is 4
0h. Further, the maximum use candidate LRU is 0Fh, the maximum number of allocated segments is 16, and the number of bytes of one block is 2
02h.

FIG. 9 is an explanatory diagram showing an example of the cache control table in the first embodiment. The role of each item is the same as in the cache control table shown in FIG. The cache segment item shown in FIG.
Indicates the contents of the cache segment allocated by Cseg
Is the cache segment number, Bseg is the buffer segment number, Adr is the buffer segment relative address, Block
Is the number of blocks in the cache segment (segment length)
Is shown.

Further, the CS (Chache Segment) item is
A bit indicating that the candidate is a candidate (virtual host device or access) holding a cache segment.
S (Modify Segment) is a bit indicating a use candidate at the time of segment configuration change (modify). This MS entry does not include a shared segment.

FIG. 9 is an explanatory diagram showing an example of the sequential read. First, when the command 1 is read, the Only attribute is generated. At this time, the command length is 10h
Therefore, as indicated by reference numeral 3a, 30h obtained by adding a margin of 20h is the capacity of the segment.

When the command 2 is read, it is determined that the command is sequential in the continuous area check of the attribute determination check item. Therefore, the read sequential (Rseq) count of the attribute change count is incremented. Command 3
In the read, the attribute change count becomes the condition of the candidate use number 3 after determining that the attribute is sequential as in the case of the command 2, so that the increment is stopped and the attribute is changed (reference numeral 3b). Here, candidate 0 has the Read sequential attribute.

Then, the segment configuration is changed according to the condition for increasing or decreasing the number of Read sequential attributes in the segment configuration change condition table (reference numeral 3c). According to the segment length condition table shown in FIG. 4, the allocation is the remaining total capacity.

[0050]

Embodiment 2 FIG. 10 shows an embodiment of random read.
First, when the command 1 is read, an Only attribute is generated. At this time, since the length of the command is 10h, 2
30h obtained by adding the margin of 0h is the capacity of the segment.

When the command 2 is read, since it does not match any of the attribute determination check items, it is determined to be a new candidate. The attribute of the candidate is the Only attribute of the attribute before change. Therefore, since the segments can be shared, the candidate shares the segment of the command 1 (reference numeral 3d).

In the read of the command 3, since the length is 20h after judging that the command is also random, the candidate is 4
Create a 0h segment. Then, this segment is shared by candidate 0 and candidate 1 (reference numeral 3e).

[0053]

Third Embodiment FIGS. 11 to 16 show an embodiment in which various commands are used to access the neighborhood. In the third embodiment, the capacity calculation based on the segment length condition table shown in FIG. 4 is performed according to the flowchart shown in FIG. In the flowchart shown in FIG. 15, the capacity calculation array shown in FIG. 16 is used. Further, in the third embodiment, an attribute search process for determining the type of access is performed using the flowchart shown in FIG. In the flowchart shown in FIG. 17, the search sequence shown in FIG. 18 is used. Access is checked by checking whether a check item is applicable to the check attribute of the search sequence shown in FIG. If the check is successful, the attribute of the candidate may change to the mutable attribute of the item. Pre-change attribute (Onl
If the check for (y, Write) succeeds, the attribute change count of each candidate is incremented.
When the attribute change count becomes equal to or greater than a predetermined value, the increment operation is stopped and the attribute is changed.

In the case of the Area attribute, it does not exist in the prepared attribute change item, but occurs in place of the Only attribute when the neighborhood access is frequently performed. The candidates for the first access in this state have the Area attribute instead of the Only attribute.

As shown in FIG. 11, when the command 1 is read, an Only attribute is generated. At this time, since the length of the command is 10h, a margin of 20h is added to 3
0h is the capacity of the segment. In the write of the command 2, when the attribute determination check is calculated from the corresponding candidate search in the flowchart shown in FIG. 14 which shows the attribute determination check more specifically, the condition is satisfied in the unconditional match part of the last condition in the write command, and Judge as attribute. In the segment allocation, the candidate 0 is 30h from the modification, and the capacity of the Write attribute is the current value (100h) (reference numeral 3f).

In the sequence from command 3 to command 7 shown in FIGS. 12 and 13, the Readsequential attribute is 2
I can do it. When this segment allocation is calculated from the capacity calculation array of the modify, first, a Write attribute is allocated, and then two Read sequential attributes are allocated.
Each cache capacity is the remaining capacity / 2 (reference numeral 3g). Further, a Reaccess attribute is generated in the sequence from the command 8 to the command 12 shown in FIG.
Since the segment capacity of the Reaccess attribute is only the access length, it becomes one block (reference numeral 3h).

The command 13 shown in FIG. 14 has a read large attribute because the length is 1000 h blocks and the number of large attribute occurrence start blocks exceeds 200 h. At this time, the segment configuration is a segment including only the candidate 4 when calculated from the capacity calculation array of the modify. However, for the large type, segment allocation is performed only at the loop when LRU = 0, so the next command 14
Is issued, no segment is allocated (reference numeral 3j).

Command 1 to Command 1 shown in FIG.
Reference numeral 6 denotes a proximity access. At this time, since the average use frequency of all Only attributes and the Area attribute was the Area attribute occurrence use frequency, the attribute here is not the Only attribute.
The area attribute is set (reference numeral 3k). Then, near access from command 17 to command 31 shown in FIG. 16 continues. Since the LRU has become 10, no segment is allocated to the access from the candidates 0 to 4 (virtual host device) (reference numeral 3m). If there is no candidate that matches the condition in the candidate search for the next command 32 with the currently existing candidate, the next command 32 prefetches the front neighborhood range = 8 and the rear neighborhood range = 8. Make a pre-reading prediction.

Next, the flow of the process of assigning and changing the segment length will be described with reference to FIG. 17 and FIG. First, the items of the capacity calculation array (capacity array) shown in FIG. 18 are sequentially input to the ptr according to their priorities (S1).
1). When the attribute of the access is determined in the attribute search loop in step S11, next, a predetermined upper limit value (nomimax) of the number of candidates is sequentially searched (S12), and whether or not the access attribute becomes a candidate is determined. (S1
3). If it is a candidate, the capacity is calculated in accordance with the calculation method of the capacity calculation array (S14), and the result of this segment length is checked (S15) to add, share,
Branch to shortage. When adding, the address is calculated (S16), and the information of the tag segment is calculated (S1).
7). When sharing, segment sharing information is stored (S18). In the case of addition and sharing, the capacity calculation processing of the next candidate is performed. On the other hand, if there is a shortage, a capacity shortage status is set (S19) and the information is stored in an unused cache area (S20).

FIG. 19 is a flowchart showing an example of the access attribute determining process, and FIG. 20 is an explanatory diagram showing an example of a search sequence used in the flowchart. First, an attribute search (S1) and a segment candidate specific loop (S2) are performed. It is checked whether or not the candidate for the segment is the attribute under search (S3). If the attribute is under search, whether or not the access is applicable is checked (S4). If the access is applicable, candidate information is extracted (S5).

As described above, according to the present embodiment, since an appropriate segment capacity is selected for a virtual host, a useless area in the cache memory can be eliminated. As a result, the performance can be improved by increasing the memory use efficiency. Moreover, even when used in an environment such as a storage device on a bus connecting a plurality of hosts, the performance improvement effect of the cache can be maximized.

[0062]

Since the present invention is constructed and functions as described above, according to this, the segment allocating means determines the segment length predetermined for each type of access based on the type of access from the higher-level device. In order to allocate the segment length of the cache memory, the segment length of the cache memory corresponding to each access is determined based on the type (attribute) of the access, such as reaccess, access to a continuous area, and one-time access. Can be assigned,
An efficient cache can be realized without wasting the capacity of the cache memory, so that an excellent data storage device which can be accessed at a high speed from a host device can be provided.

[Brief description of the drawings]

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

FIG. 2 is a block diagram illustrating a detailed configuration of a segment allocating unit illustrated in FIG. 1;

FIG. 3 is an explanatory diagram illustrating an example of an attribute type table.

FIG. 4 is an explanatory diagram showing an example of a segment length condition table.

FIG. 5 is a block diagram illustrating an example of a segment configuration change condition table.

FIG. 6 is an explanatory diagram showing an example of an attribute assignment order table.

FIG. 7 is a block diagram showing a configuration of another embodiment.

FIG. 8 is an explanatory diagram showing details of a cache control table shown in FIG. 7;

FIG. 9 is an explanatory diagram illustrating an example of a cache control table according to the first embodiment;

FIG. 10 is an explanatory diagram illustrating an example of a cache control table according to the second embodiment;

FIG. 11 is an explanatory diagram illustrating an example of a cache control table executed from command 1 to command 2 according to the third embodiment.

FIG. 12 is an explanatory diagram illustrating an example of a cache control table executed from command 3 to command 4 according to the third embodiment.

FIG. 13 is an explanatory diagram illustrating an example of a cache control table executed from command 5 to command 12 according to the third embodiment.

FIG. 14 is an explanatory diagram illustrating an example of a cache control table in which a command 13 according to the third embodiment has been executed.

FIG. 15 is an explanatory diagram illustrating an example of a cache control table in which a command 14 according to the third embodiment has been executed.

FIG. 16 is an explanatory diagram illustrating an example of a cache control table executed from a command 15 to a command 16 according to the third embodiment.

FIG. 17 is a flowchart showing a flow of processing for executing segment assignment and change.

FIG. 18 is an explanatory diagram showing an example of a capacity calculation array used in the flowchart shown in FIG. 17;

FIG. 19 is a flowchart illustrating a flow of a process of determining an access type.

FIG. 20 is an explanatory diagram showing an example of a search sequence used in the flowchart shown in FIG. 19;

[Explanation of symbols]

 Reference Signs List 2 control unit 2A cache control unit 2B CPU 3 cache control table 4 segment allocating unit 6 storage medium 8 cache memory

Claims (13)

[Claims] 1. A storage medium for storing data transmitted from a higher-level device, and a cache memory provided in the storage medium and storing data written or read by the higher-level device as cache data having a predetermined segment length. And a control unit that stores the data in the cache memory in response to write or read access of data to the storage medium of the higher-level device, wherein the control unit is configured to receive the data from the higher-level device. A data storage device comprising: a segment allocating unit that allocates a segment length of the cache memory with a segment length determined in advance for each type of access based on the type of access. 2. An access type classification function for classifying an access type based on address information of the storage medium to be accessed by the higher-level device and a data length of the accessed data, A segment length condition table storing a segment length determined from a total capacity of the cache memory and a data length of the accessed data for each access type classified by the access type classification function. Item 2. The data storage device according to Item 1. 3. The method according to claim 2, wherein the access type classification function has a function of monitoring the progress of a plurality of accesses from the host device and classifying the access type based on the progress information of the plurality of accesses. Claim 2
A data storage device as described. 4. The access type classification function has a function of determining that the access type is a reaccess when the same address is accessed a plurality of times based on the progress information of the plurality of accesses, 4. The data storage device according to claim 3, wherein the segment length condition table includes condition information for setting the data length of the access to the segment length when the access type is reaccess. 5. A segment length changing function, wherein the segment allocating means changes a segment length of the cache memory according to an access type when the access type changes based on a plurality of accesses from the host device. The data storage device according to claim 3, further comprising: 6. The segment length change function includes a segment configuration change condition table storing a change condition of a segment configuration of the cache memory when the access type is continued for a predetermined number of times. The data storage device according to claim 5. 7. A virtual higher-level device prediction, wherein the segment allocating unit predicts the number of the higher-level devices accessing the storage medium as the number of virtual higher-level devices based on the access type classified by the access type classification function. And a higher-level device-based segment allocation function of allocating a segment of the cache memory with reference to the segment length condition table for each virtual higher-level device predicted by the virtual higher-level device prediction function. The data storage device according to claim 2. 8. The access type classification function monitors the progress of a plurality of accesses from the host device and classifies the access type based on the progress information of the plurality of accesses. A function of recording whether or not the access of the access type classified by the classification function has been recently used; and the upper-level device-based segment allocation function has recently used the access based on information on whether or not the access has been recently used. 8. The data storage device according to claim 7, further comprising a function of allocating a segment of said cache memory to said access type. 9. The access type classification function monitors the progress of a plurality of accesses from the host device and classifies the access type based on the progress information of the plurality of accesses. A function of recording the frequency of use of the access type classified by the classification function, wherein the higher-level device-based segment allocation function is configured to change the access type from the most frequently used access type to the less frequently used access type based on the use frequency of the access type. 8. The data storage device according to claim 7, further comprising a function of allocating segments of said cache memory in the order of: 10. The apparatus according to claim 1, wherein said segment allocating means has a shared segment allocating function for sharing said segment with said plurality of virtual higher-level devices when the data of the segments allocated by said higher-level device-specific segment allocating function match. The data storage device according to claim 7, characterized in that: 11. The segment allocating means, wherein the access type classification function has a function of determining that the access is of a continuous or close access type when an address of an access from the higher-level device is continuous or close, Has a read-ahead range assignment function of predicting a read-ahead range from an access determined as a continuous or close access type by the access type classification function and assigning the read-ahead range to the continuous or close access type. 3. The data storage device according to claim 2, wherein: 12. A storage medium for storing data transmitted from a higher-level device, and a cache memory provided in the storage medium and storing data written or read by the higher-level device as cache data having a predetermined segment length. And a control unit that stores the data in the cache memory in response to write or read access of data to the storage medium of the higher-level device, wherein the control unit is configured to receive the data from the higher-level device. A cache control table for sequentially storing access attributes; and a table use segment allocating unit for allocating a segment length of the cache memory according to the access attribute based on a change in the contents of the cache control table. Data storage device. 13. A storage medium for storing data transmitted from a higher-level device, a cache memory provided in parallel with the storage medium for storing the data as cache data having a predetermined segment length, and a storage medium for storing the data in the higher-level device. A control unit that stores the data in the cache memory in response to write or read access, wherein the control unit accesses the storage medium based on a type of access from the higher-level device. A high-level device number prediction means for predicting the number of high-level devices, and a high-level device-specific segment allocation means for allocating a segment length of the cache memory for each high-level device predicted by the high-level device number prediction means. Characteristic data storage device.