JP6007628B2 - Cache control device and cache control method - Google Patents

Description

  The present invention relates to a cache control device and a cache control method having a hierarchical storage configuration of two or more stages, and more particularly to a cache control device and a cache control method capable of reducing processing time for creating a free area in a cache area. .

  Currently, when an application program (hereinafter also simply referred to as an “application”) that runs on a computer acquires necessary data, if the data exists in a storage device with a low access speed, the data is stored at the low speed. A cache function is known in which, after being acquired from a storage device, the acquired data is temporarily stored in a storage device with a high access speed in preparation for subsequent use.

  Needless to say, the cache function is effective when the data requested to be accessed is already cached in a high-speed storage device (this is called “cache hit”).

  Here, in order to increase the effect of the cache function, a method of increasing the storage capacity of a storage device for storing cache data (hereinafter referred to as “cache memory”) and a cache hit with a cache memory having a limited storage capacity It is conceivable to increase the probability of.

  First, it is a method of increasing the storage capacity of the cache memory. Generally, a high-speed storage device used as a cache memory is more expensive than a low-speed storage device, and the cache memory is simply added in terms of cost. There are limits to what you can do.

  Therefore, there is a method in which an expensive but high-speed primary cache memory and an inexpensive but low-speed secondary cache memory are prepared and the secondary cache memory is used as an auxiliary storage device in a hierarchy below the primary cache memory.

  Next, as a method for increasing the probability of a cache hit, when new data to be cached occurs, if the data is stored in the memory until the storage capacity of the cache memory is full, a part of the stored data is cached. There has been considered a method of securing a storage area for storing the new data by discarding the hit probability so as not to decrease as much as possible.

  As an example of such a technique, Patent Document 1 describes a content storage system in which a high-speed storage device having a limited storage capacity and a large storage capacity but a low-speed storage device are connected to each other to access data. A technique for selecting a storage destination based on frequency is disclosed.

  In this technique, after the content stored in the high-speed storage device is transferred to the low-speed storage device, the empty area of the high-speed storage device is secured by erasing the content from the high-speed storage device.

  Also, with this technology, by selecting the content to be erased from the high-speed storage device based on the access frequency, content with few download requests is deleted from the high-speed storage device, and content with many download requests is processed at high speed. It is configured to remain in the storage device.

  As a result, content with many download requests is permanently stored in the high-speed storage device, so that it is possible to provide a high-speed and large-scale content storage system with suppressed cost increase.

  That is, the present technology is an example in which the above-described method for increasing the storage capacity of the cache memory and the method for increasing the probability of cache hit are used in combination. The primary cache memory described above corresponds to the high-speed storage device of the present technology, and the secondary cache memory described above corresponds to the low-speed storage device of the present technology.

  As another method for increasing the storage capacity of the cache memory, another method for increasing the amount of data stored in a limited storage capacity is known.

  As an example of such a technique, in Patent Document 2, in a hierarchical storage device system, data is compressed when data is written from the host device to the cache memory, and data is read from the cache memory to the host device. A technique for decompressing data is disclosed. This technique increases the data capacity that can be stored in the cache memory by compressing the data, and intends to obtain the same effect as that of providing a larger cache memory. In other words, it is possible to increase the utilization efficiency of the limited capacity cache memory, increase the cache hit rate, and increase the calculation speed.

Japanese Patent No. 4464656 JP 05-07413 A

  Here, considering the processing time for registering new data in the cache, in the technique disclosed in Patent Document 1, if there is no data that can be immediately erased in the high-speed storage device, the content is stored at low speed from the high-speed storage device. After transferring and writing to the device, it takes time to erase the content and create a free space.

  On the other hand, in the technique disclosed in Patent Document 2, if it is considered that the storage capacity for registering new data is secured by compressing the cache data so far, the new data (compressed) is compressed. Data compression time is required to obtain a free space.

  In the technique disclosed in Patent Document 1, when data that has been cached in a free area is used again, it takes time to read and transfer from the low-speed storage device.

  On the other hand, in the technique disclosed in Patent Document 2, it takes time to decompress data in order to use cache data.

  Which of these two technologies has the shorter total processing time depends on the storage device used and the performance of the data compression algorithm, etc., but as shown in FIG. It is done. Note that the “write time” in FIG. 1 includes the time to transfer (send) and write (write) to the secondary cache.

  In FIG. 1, in the range A, the compression time is less than the write time, and the data compression process is faster. On the other hand, in the range of B, the compression time> the write time, and the process of writing to the secondary cache is faster.

  That is, depending on the size of the target data, there are a case where it is faster to move from the primary cache to the secondary cache, and a case where it is faster to compress the data in the primary cache to create a free area. Even if it is adopted, it cannot always be said that optimum processing can be performed.

  The cache control apparatus according to the present invention includes a primary storage means for storing data, a secondary storage means for reading data at a lower speed than the primary storage means, and compressing and storing the data in the primary storage means When data is stored in the primary storage means and there is no free space for storing the new data in the primary storage means, it is designated based on a predetermined condition. Selection means for selecting data in the primary storage means to be processed as data to be processed, each size of the data, and compression that is a predicted value of the compression time required to compress the data of each size by the data compression means The relationship between the method prediction value and the movement method prediction value that is the predicted value of the movement time required to move and write data of each size from the primary storage means to the secondary storage means is stored in advance. In the case of the size of the processing target data, in the determination unit that determines which one of the compression method predicted value and the movement method predicted value is smaller, and in the determination unit, the compression method predicted value is more If it is determined to be small, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit. On the other hand, if it is determined that the movement method predicted value is smaller, And processing means for moving and writing the processing object data to the second storage means.

  The cache control apparatus according to the present invention includes a primary storage means for storing data, a secondary storage means for reading data at a lower speed than the primary storage means, and compressing and storing the data in the primary storage means When data is stored in the primary storage means and there is no free space for storing the new data in the primary storage means, it is designated based on a predetermined condition. Selection means for selecting the data in the primary storage means to be processed data, a compression method predicted value that is a predicted value of the compression time required to compress the processing target data by the data compression means, A movement method predicted value, which is a predicted value of the movement time required to move and write data to be processed from the primary storage unit to the secondary storage unit, and calculates the compression method predicted value and When the determination unit determines which of the movement method predicted values is smaller and the determination unit determines that the compression method predicted value is smaller, the compression unit stores the processing target data. Processing means for storing compressed compressed data in the primary storage means, and on the other hand, if it is determined that the movement method predicted value is smaller, the processing target data is moved to the second storage means and written. With.

  The cache control method of the present invention includes a primary storage means for storing data, a secondary storage means for reading data at a lower speed than the primary storage means, and compressing and storing the data in the primary storage means Each of the data sizes, a compression method prediction value that is a prediction value of the compression time required to compress the data of each size by the data compression means, and each size A predicted value storage step for storing in advance in the storage means a relationship with a movement method predicted value that is a predicted value of the movement time required to move and write data from the primary storage means to the secondary storage means; In the case where new data is stored in the means, and there is no free space for storing the new data in the primary storage means, the primary data specified based on a predetermined condition. A selection step of selecting data in the means as processing target data, and a determination step of determining which one of the compression method prediction value and the movement method prediction value is smaller in the case of the size of the processing target data; In the determination step, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit, while the movement When it is determined that the method predicted value is smaller, the processing step includes moving and writing the processing target data to the second storage unit.

  The cache control method of the present invention includes a primary storage means for storing data, a secondary storage means for reading data at a lower speed than the primary storage means, and compressing and storing the data in the primary storage means A method of using data compression means that can store new data in the primary storage means, and there is no free space in the primary storage means for storing the new data, A selection step of selecting data in the primary storage means designated based on a predetermined condition as processing target data, and a predicted value of a compression time required to compress the processing target data by the data compression means Calculate a compression method predicted value and a movement method predicted value that is a predicted value of the movement time required to move and write the processing target data from the primary storage unit to the secondary storage unit In the determination step of determining which one of the compression method prediction value and the movement method prediction value is smaller, and in the determination step, if it is determined that the compression method prediction value is smaller, the processing The compressed data obtained by compressing the target data by the compression unit is stored in the primary storage unit. On the other hand, if it is determined that the movement method predicted value is smaller, the processing target data is stored in the second storage unit. Processing steps to move and write to the means.

  The present invention has an effect that, in a cache control device having a hierarchical memory structure of two or more stages, it is possible to achieve an improvement in cache registration processing performance in a situation where there is no free area in the primary cache.

It is a graph which shows an example of the processing time by data compression, and the processing time by the write to a secondary cache. It is a block diagram which shows the structure of the 1st Embodiment of this invention. It is a block diagram which shows the structure of a primary storage device and a secondary storage device. It is a flowchart which shows the data registration operation | movement to the primary cache in the 1st thru | or 3rd embodiment of this invention. It is a flowchart which shows the reference operation | movement to the cached data in the 1st thru | or 3rd embodiment of this invention. It is a block diagram which shows the structure of the 2nd Embodiment of this invention.

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

<First Embodiment>
FIG. 2 is a block diagram showing the first embodiment of the present invention.

  Referring to FIG. 2, the present embodiment includes a computer 100 and an external information source 200.

  The computer 100 includes an arithmetic processing device 10, a primary storage device 20, a secondary storage device 30, a processing data selection unit 40, a method determination unit 50, a data compression unit 60, and a data expansion unit 70.

  The arithmetic processing device 10 controls the operation of each unit.

  The primary storage device 20 is a main storage device connected to the arithmetic processing device 10, and is a storage device having a high data read / write speed (hereinafter referred to as “access speed”).

  The primary storage device 20 performs an operation as a primary cache memory, and also performs an operation of holding an OS program and application programs themselves and data used by those programs. For this purpose, an area for holding cache data and an OS It is divided into areas for holding data used by programs and application programs.

  In the following description, programs such as OS programs and application programs are collectively referred to simply as “applications”.

  Next, a detailed configuration of the primary storage device 20 will be described with reference to FIG.

  The storage area of the primary storage device 20 is divided into a primary cache area 21 and an application area 25.

  The primary cache area 21 is a storage area for temporarily storing data acquired from the external information source 3 by the application in preparation for the next access opportunity. The next storage device 20 is assigned with a predetermined storage capacity.

  The primary cache area 21 further includes a decompressed information storage area 22, a compressed information storage area 23, and a free area 24. The free area 24 indicates an unused area in the primary cache area 21 and may not exist.

  The development information storage area 22 is an area in which the cache data to be stored in the primary cache area 21 is stored as it is without being compressed. Note that cache data that has not been compressed may be referred to as “uncompressed data”.

  The compressed information storage area 23 is an area for storing the cache data to be stored in the primary cache area 21 in a compressed state.

  The sizes of the decompressed information storage area 22 and the compressed information storage area 23 vary depending on the state of cached data. Further, the region 22 and the region 23 may be discontinuous.

  The application area 25 is an area used by the application, and the data read from the external information source 3 is stored in this area and delivered to the application that requested the data.

  The secondary storage device 30 is a storage device that has a lower access speed than the primary storage device 20 but a higher access speed than the external information source 3.

  Specifically, for example, the primary storage device 20 is a semiconductor memory such as a RAM (Random Access Memory), the secondary storage device 30 is an SSD (Solid State Drive, a semiconductor disk), and the external information source 3 is a magnetic disk ( HDD).

  Next, the detailed configuration of the secondary storage device 30 will be described with reference to FIG.

  In the present embodiment, the secondary storage device 30 includes only a secondary cache area 31, and the secondary cache area 31 includes a decompressed information storage area 32, a compressed information storage area 33, and a free area 34. The free area 34 indicates an unused area in the secondary cache area 31 and may not exist.

  The development information storage area 32 is an area in which the cache data to be stored in the secondary cache area 31 is stored without being compressed.

  The compressed information storage area 33 is an area for storing the cache data to be stored in the secondary cache area 31 in a compressed state.

  The sizes of the decompressed information storage area 32 and the compressed information storage area 33 vary depending on the state of cached data. Further, the region 32 and the region 33 may be discontinuous.

  The secondary cache area 31 is for temporarily storing cache data that cannot be stored in the primary cache area 21. When the application refers to the data again, the data is first read from the secondary cache area 31 and transferred to the primary cache area 21, and the application is thus transferred to the primary cache area. The data is used.

  The processing data selection unit 40 selects data to be moved or deleted from the data stored in the primary cache area 21 in order to increase the free area of the primary cache area 21. The selection criteria for data selection performed by the processing data selection unit 40 is basically based on selecting data that has a low re-access probability in consideration of, for example, the use frequency and the last access time.

  Therefore, for example, the processing data selection unit 40 collects information such as the access timing of each data from the application, and arranges it as information such as the usage frequency and the last access time.

  The data compression means 50 reads and compresses uncompressed data on the primary storage device 20 and writes the compressed data after compression onto the primary storage device 20.

  As the data compression algorithm, an algorithm that can predict the time required for compression based on the size of the data to be compressed and the expansion time of the compressed data is employed.

  For example, in a so-called lexicographic algorithm such as the well-known LZSS (Lempel-Ziv-Storer-Szymanski) method, compression is performed by repeating the same process for each element of the original data. The number of compression processes increases in proportion to the size of. Thereby, if the compression processing time per time is measured, the approximate compression time can be predicted from the data size. The same applies to development.

  Further, the arithmetic processing device 10 may be used for the compression and decompression arithmetic processing, or another device such as a GPU (Graphics Processing Unit) may be used.

  The data expansion unit 60 reads out the data compressed by the data compression unit 50 from the primary storage device 20, expands the data, and writes the expanded data on the primary storage device 20. A data expansion algorithm corresponding to the compression algorithm of the data compression means 50 is employed.

  Based on the size of the processing target data selected by the processing data selection unit 40, the method determination unit 70 takes the time required for processing to secure the free space necessary for writing the data to the primary cache area 21. Determine the short processing method.

  As this processing method, in this embodiment, a method of compressing existing uncompressed data in the primary cache area 21 and freeing the reduced size (hereinafter referred to as “compression method”), and 1 There are two processing methods: a method of moving data from the next cache area 21 to the secondary cache area 31 (hereinafter referred to as “move method”).

  First, the processing time of the compression method will be described.

  In the present embodiment, the processing time of this method is the time required for compressing the uncompressed data in the primary cache area 21 by the data compressing means 50 and the data expanding means 60 for expanding the compressed data. It shall be expressed as the total time with time. As described above, the time required for these two processes can be estimated from the data size.

  In the present embodiment, the read / write time for the primary storage device 20 is sufficiently short compared with the time required for the above two processes and can be ignored, but is not taken into consideration, but the primary storage device 20 and the data compression means 50 are not taken into consideration. Depending on the performance of the data expansion means 60, it may be taken into consideration.

  Next, the processing time of the movement method will be described.

  In this embodiment, the processing time of this method is the time required to send data from the primary storage device 20 to the secondary storage device 30, the time required to write data to the secondary storage device 30, and the secondary storage device. It is represented by the total time of the time required for reading from 30 and the time required for sending to the primary storage device 20. These four times can be roughly estimated for each data size to be processed, based on measurement values obtained from tests after the device is configured and device specification information.

  Similarly, in this method, the reading / writing time for the primary storage device 20 is sufficiently short compared with the above two times and can be ignored, but is not taken into consideration. However, the primary storage device 20 and the secondary storage device 30 are not taken into consideration. Depending on the degree of performance difference between the two, it may be taken into account.

  As described above, the two methods can predict the processing time from the target data size. Therefore, in the present embodiment, determination information in which the relationship between the data size and the processing time in the two methods is predicted is stored in advance in the method determination unit 70, and thus the method determination unit 70 responds to the data size. Select the method with the shorter processing time.

  The determination information may be calculated by the method determination unit 70 during processing.

  Here, for example, if the relationship between the processing time and the data size in the two methods can be represented by a graph as shown in FIG. 1, the determination information is when the data size to be processed is d and 0 ≦ d <a. Indicates information such as a compression method, a movement method when a ≦ d <b, and a compression method when b ≦ d <x (x is not shown).

  In the above moving method, the time required for reading from the secondary storage device 30 and the time required for sending to the primary storage device 20 among the four times may be reduced from the processing time.

  The external information source 200 is a storage device that stores original data to be accessed by an application running on the computer 100. The external information source 200 is a storage means having a lower access speed than the primary storage device 20 and the secondary storage device 30, and is a magnetic disk device in this embodiment as described above.

  In the present embodiment, all the components other than the external information source 200 are mounted on one computer, but the present invention is not limited to this. For example, the secondary storage device 30 may be provided in another computer connected through a high-speed network line, and the external information source 200 may be provided in another computer connected through a lower-speed network line.

  Next, the operation of the present embodiment having the above-described configuration will be described in detail.

  The main operation of this embodiment is divided into a data registration operation to the primary cache area 21 and a reference operation to the cached data. These will be described in order below.

  First, the data registration operation in the primary cache area 21 will be described with reference to FIG.

  Assume that the state of the primary cache area 21 and the secondary cache area 31 when the computer 100 is started is as shown in FIG.

  First, when the application requests registration of the data D1 and the data D1 needs to be registered in the primary cache area 21, the arithmetic processing unit 10 determines the size of the data D1 to be registered in the primary cache area 21. Is checked to determine whether or not it can be stored in the free area 24 of the primary cache area 21 (step S100).

  The determination as to whether or not registration to the primary cache area 21 is necessary uses a general cache technology, and details are not described in this embodiment.

  If it is stored in the free area 24, the arithmetic processing unit 10 writes the data D1 in the free area 24 of the primary cache area 21 and defines that area as a part of the development information storage area 22 (step S130). ). In this case, the registration process of the data D1 in the primary cache area 21 is completed.

  On the other hand, if the size of the data D1 is larger than the size of the free area 24, the process proceeds to increase the size of the free area 24 in the primary cache area 21.

  The arithmetic processing unit 10 checks whether or not the compressed information storage area 23 exists in the primary cache area 21 (step S101).

  If there is no compressed information storage area 23 here, the process proceeds to step S102 described later.

  Since there is the compressed information storage area 23 now, the process proceeds to the process of moving the compressed data.

  The processing data selection unit 40 selects the compressed data D2 having the lowest usage in the compressed information storage area 23 from the access status so far (step S110).

  The arithmetic processing unit 10 writes the compressed data D2 to the free area 34 in the secondary cache area 31 of the apparatus 30, and defines the written area as the compressed information storage area 33. Then, the compressed data D2 is deleted from the compressed information storage area 23 of the primary cache area 21 (step S111). As a result, the free space 24 is increased by the amount of the data D2. As a result, when a free space 24 having a size necessary for registering the data D1 is secured, the data D1 is registered there (step 130).

  On the other hand, if the free area 24 of the size necessary for registering the data D1 cannot be secured yet (step 100), the processing of steps S110 and S111 can secure the free area 24 of a size that can accommodate the data D1. Repeat until. If these steps are repeated and all the data in the compressed information storage area 23 is deleted, but the empty area 24 of the size necessary for registering the data D1 cannot be secured, the processing data selection unit 40 From the access status up to that point, uncompressed data D3 having the lowest usage is selected from the expanded information storage area 22 of the primary cache area 21 (step S102).

  Next, the method determination unit 70 confirms the size of the data D3 and selects a method with a short processing time as follows (step S103).

  That is, when the size d3 of the data D3 is in the range B in FIG. 1 (a ≦ d3 <b), the method determination unit 70 selects the transfer method to the secondary cache area 31 and proceeds to step S120.

  The arithmetic processing unit 10 writes the uncompressed data D3 into the empty area 34 in the secondary cache area 31, and defines the written area as the development information storage area 32. Then, the uncompressed data D3 is deleted from the expanded information storage area 22 of the primary cache area 21 (step S120). As a result, the free space 24 increases by the amount of the data D3. Then, the process returns to step S100.

  On the other hand, in step S102, when the size d3 of the data D3 is in the range A in FIG. 1 (0 ≦ d3 <a or b ≦ d3 <x, x is not shown), the method determination unit 70 selects the compression method. And proceed to step S104.

  The data compression means 50 compresses the uncompressed data D3, creates the compressed information storage area 23 in the primary cache area 21 again, and writes the compressed data D3 therein (step S104).

  Next, the arithmetic processing unit 10 deletes the data D3 before compression from the expanded information storage area 22 of the primary cache area 21 (step S105). As a result, the free space 24 increases by the size reduced by the compression of the data D3. Then, the process returns to step S100.

  At this time, if the empty area 24 of the size necessary for registering the data D1 cannot be secured yet, and the movement method is selected for the data D3 in step 103, the steps S102 and S103 are performed again. The process proceeds to step S120 or steps S104 and S105. On the other hand, if the compression method is selected for the data D3 in step 103, the process proceeds to steps S110 and S111.

  For example, when there are two data a and data b in the development information storage area 22, first, data a having a low utilization rate is selected as a processing target first, and if it is faster to move as it is, it remains uncompressed. If it is moved to the secondary cache area 31 and the compression is faster than the movement, the area is compressed. If the free space 24 is still insufficient, the previously compressed data a is moved to the secondary cache area 31 before another data b in the development information storage area 22 is processed. That is, until the size of the free area 24 can be sufficiently secured, the data in the development information storage area 22 is moved to the secondary cache area 31 in the order in which the processing data selection unit 40 determines that the usage rate is low. The form of the data is a form of compression or non-compression according to the processing time.

  The size of the free area 24 is increased in accordance with the processing flow as described above. Finally, a free area 24 having a size sufficient to register the data D1 is secured (step S100), and this is defined as a development information storage area 23. By storing the data D1 (step S130), the registration process of the data D1 in the primary cache area 21 is completed.

  In this way, the existing data in the primary cache area 21 is moved to the secondary cache area 31 using means capable of moving the data fastest, or is compressed in the primary cache area 21. Thus, it is possible to reduce the size of the existing data and register the new data D1 in the primary cache area 21.

  Next, referring to FIG. 5, the reference operation for the cached data will be described.

  Now, it is assumed that some data has already been registered in the primary cache area 21 and the secondary cache area 31, and the application requests reference to the data D4.

  First, the arithmetic processing unit 10 checks whether or not the access target data D4 is stored in the primary cache area 21 (step S200).

  If it is stored in the expansion information storage area 22 or the compressed information storage area 23 of the primary cache area 21, the process proceeds to step S203.

  On the other hand, when the data is not stored in the primary cache area 21, the arithmetic processing unit 10 checks whether the access target data D4 is stored in the secondary cache area 31 (step S201).

  If it is not stored in the expanded information storage area 22 or the compressed information storage area 23 of the secondary cache area 21, it is processed as a cache miss. That is, the arithmetic processing device 10 acquires the data D4 from the external information source 200 (step S210), stores it in the application area 25 of the primary storage device 20, and delivers it to the application (step S206).

  Thereafter, when it is determined that the data D4 needs to be registered in the primary cache area 21, the above-described data registration operation in the primary cache area 21 may be performed.

  On the other hand, when the data D4 is stored in the secondary cache area 31 in step S201, the arithmetic processing unit 10 transfers the data D4 from the secondary cache area 31 to the primary cache area 21. If the transferred data D4 is non-compressed data, the arithmetic processing unit 10 stores the data D1 in the decompressed information storage area 22, and if the data D4 is compressed data, stores it in the compressed information storage area 23 (step S202). After the transfer, the arithmetic processing unit 10 deletes the data D4 in the secondary cache area 31.

  Next, the arithmetic processing unit 10 confirms whether the data D4 is compressed (step S203).

  If it is uncompressed data, the process proceeds to step S206.

  On the other hand, in the case of compressed data, the data expansion means 60 expands the compressed data D4 (step S204) and stores the expanded data D4 in the expanded information storage area 22 of the primary cache area 21 and the original compressed data. D4 is deleted from the compressed information storage area 23 (step S205).

  Finally, the arithmetic processing unit 10 stores the uncompressed data D4 in the decompressed information storage area 22 of the primary cache area 21 in the application area 25 of the primary storage apparatus 20 and delivers it to the application (step S206).

  In steps S202 and S205, when the free area 24 of the primary cache area 21 for storing the data D4 is insufficient, the data registration operation to the primary cache area 21 described above is performed.

  Here, the time required for referring to the cache data is considered.

  First, when it is stored in the secondary cache area 31, it takes time to read from the secondary storage device 30 and to transfer to the primary storage device 20.

  Next, when data is compressed, it takes time to decompress.

  However, since the method determining unit 70 adds the time required for the reference as described above to the determination information when the method determining unit 70 selects the processing method in the data registration operation in the primary cache area 21 (step S103). The time required for processing when using data later has already been taken into account in the method selection. That is, in the present embodiment, it can be said that the data compression is controlled only when the processing time is shortened even if the expansion processing for cache data reference is included. Similarly, it can be said that the data movement is controlled only when the processing time becomes short even when the reading from the secondary storage device 30 and the time to send to the primary storage device 20 are included.

  As described above, this embodiment has a cache data registration processing performance higher than that of a conventional cache control device having a hierarchical storage configuration and a cache control device that only stores compressed cache data. Has the effect of improving.

  The reason is that, in order to register new data, data is selected from the primary cache area based on a predetermined condition, and the time required for compressing the selected data and the selected data are determined as secondary data. This is because a process having a shorter time out of the time required for moving to the cache area and writing is selected, and the selected process is performed on the data.

  Further, the data stored in the secondary cache area has an effect of improving the performance of the cache reference process.

  The reason is based on a predicted value obtained by adding the time required for the processing for movement and expansion when reading data from the secondary cache area at the time of the process for securing the free capacity of the primary cache area, This is to determine the method with the shorter processing time. In other words, even if the time required for compression and decompression is added, the processing time required for reading the compressed and reduced data from the secondary storage device and transferring it to the primary storage device can be shortened. When the data size is small, the data is compressed and stored in the secondary cache area. Therefore, the reference performance is improved as compared with the conventional method in which data is always stored in the secondary cache in an uncompressed state. On the other hand, if the data size takes extra time by performing compression and decompression, it is moved to the secondary cache area without being compressed, so performance is always better than the conventional method of compressing cache data. Will improve.

<Second Embodiment>
Next, a second embodiment based on the above-described first embodiment will be described. In the following, the characteristic part according to the second embodiment will be mainly described, and the components of the second embodiment having the same configuration as the first embodiment are attached in the first embodiment. The same reference numerals as those of the reference numerals are attached, and the detailed description of the constituent elements is omitted.

  FIG. 6 is a block diagram showing the second embodiment.

  Referring to FIG. 6, the present embodiment is different from the first embodiment in that a computer 101 has a plurality of data compression means 50 and a plurality of data expansion means 60.

  Note that the number of data compression means 50 and data expansion means 60 need not be the same.

  Next, the operation of the present embodiment having the above-described configuration will be described in detail.

  The operation of the present embodiment is basically the same as that of the first embodiment, and only the following points are different.

  First, in the data registration operation in the primary cache area 21, the method determination unit 70 divides the data D3 into a plurality of partial data according to the number of the data compression means 50 and the data expansion means 60. The data size adopted when determining the time method is set to the largest data size among the divided data (step S103 in FIG. 4). That is, the method determination unit 70 divides the data D3 into a plurality of first partial data so as to have a one-to-one correspondence with the plurality of compression units 50, and causes each compression unit 50 to perform compression processing on each first partial data. The longest predicted processing time required and the compressed data obtained by compressing the data D3 are divided into a plurality of second partial data so as to have a one-to-one correspondence with the plurality of decompressing means 60, and each second data is divided. A value obtained by adding the longest predicted value of the processing time required when the corresponding data is expanded by the corresponding expansion means 60 is determined as a compression method predicted value.

  Here, the predicted value of the compression method is calculated in consideration of the predicted value of the decompression process time by the decompression means 60, but the compression process is performed without considering the predictive value of the decompression process time by the decompression means 60. Needless to say, the predicted value may be determined only from the processing time.

  Next, when the data D3 is actually compressed, the plurality of first partial data divided by the method determination unit 70 in step S103 is distributed to the corresponding plurality of data compression means 50, and the plurality of data The compression means 50 performs compression processing in parallel (step S104 in FIG. 4).

  In the reference operation to the data, when decompressing the compressed data D3, the data D3 is divided into a plurality of second partial data, and the plurality of second partial data thus divided are respectively The data is distributed to a plurality of corresponding data expansion means 60 and processed in parallel by the plurality of data expansion means 60 (step S203 in FIG. 5).

  As described above, the present embodiment has an effect that the time for the cache registration process and the reference process can be further shortened in addition to the same effect as the first embodiment described above.

  This is because the data compression and decompression processing time can be shortened by dividing the data and performing parallel processing.

  In addition, although part or all of this invention can be described also as the following additional remarks, it cannot be overemphasized that this invention is not limited to the following additional remarks.

(Appendix 1)
A primary storage unit that stores data, a secondary storage unit that reads data at a lower speed than the primary storage unit, and a data compression unit that compresses and stores the data in the primary storage unit are used. In the cache control method,
Each size of data, a compression method predicted value that is a predicted value of the compression time required to compress the data of each size by the data compression unit, and each size of data from the primary storage unit to the secondary storage unit A predicted value storage step for storing in advance in the storage means the relationship with the predicted value of the movement method, which is the predicted value of the travel time required to move and write,
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. A selection step of selecting data in the means as processing target data;
In the case of the size of the processing target data, a determination step of determining which one of the compression method predicted value and the movement method predicted value is smaller;
In the determination step, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit, while the movement method is stored. And a processing step of moving and writing the processing target data to the second storage means when it is determined that the predicted value is smaller.

(Appendix 2)
A primary storage unit that stores data, a secondary storage unit that reads data at a lower speed than the primary storage unit, and a data compression unit that compresses and stores the data in the primary storage unit are used. In the cache control method,
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. A selection step of selecting data in the means as processing target data;
A compression method predicted value, which is a predicted value of the compression time required for compressing the processing target data by the data compression means, and the processing target data required to move from the primary storage means to the secondary storage means for writing. A determination step of calculating a movement method prediction value that is a predicted value of a movement time, and determining which of the compression method prediction value and the movement method prediction value is smaller;
In the determination step, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit, while the movement method is stored. And a processing step of moving and writing the processing target data to the second storage means when it is determined that the predicted value is smaller.

(Appendix 3)
The cache control method according to appendix 1 or 2, wherein the movement method predicted value is a value obtained by adding a time required to read the data from the secondary storage unit and move it to the primary storage unit.

(Appendix 4)
A plurality of the data compression means are used,
The determination step is a process required when the processing target data is divided into a plurality of partial data so as to have a one-to-one correspondence with the plurality of compression units, and each partial data is compressed by the corresponding data compression unit. Use the longest prediction value of time as the compression method prediction value,
If it is determined in the determining step that the compression method predicted value is smaller, the plurality of partial data are compressed in parallel by the plurality of corresponding data compressing means. 3. The cache control method according to 3.

(Appendix 5)
A plurality of the data compression means and data expansion means for expanding compressed data into uncompressed data are used,
The determination step divides the processing target data into a plurality of first partial data so as to have a one-to-one correspondence with the plurality of compression units, and causes each data compression unit to perform compression processing on each first partial data. The longest predicted value of the processing time required at the time and the compressed data obtained by compressing the processing target data are divided into a plurality of second partial data so as to correspond one-to-one with the plurality of data expansion means. Using the value obtained by adding the longest of the predicted values of the processing time required when the second data is expanded by the corresponding data expansion means as the compression method predicted value,
When it is determined in the determining step that the compression method predicted value is smaller, the plurality of first partial data are compressed in parallel by the plurality of corresponding data compression means. The cache control method according to appendices 1 to 3.

(Appendix 6)
A system comprising primary storage means for storing data, secondary storage means for reading data at a lower speed than the primary storage means, and data compression means for compressing and storing data in the primary storage means In
Each size of data, a compression method predicted value that is a predicted value of the compression time required to compress the data of each size by the data compression unit, and each size of data from the primary storage unit to the secondary storage unit A predicted value storage process for storing in advance in a storage means a relationship with a predicted value of a movement method that is a predicted value of a moving time required to move and write;
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. Selection processing for selecting data in the means as processing target data;
In the case of the size of the processing target data, a determination process for determining which of the compression method predicted value and the movement method predicted value is smaller;
In the determination process, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit, while the movement method When it is determined that the predicted value is smaller, the cache control program causes the computer to execute a process of moving and writing the processing target data to the second storage unit.

(Appendix 7)
A system comprising primary storage means for storing data, secondary storage means for reading data at a lower speed than the primary storage means, and data compression means for compressing and storing data in the primary storage means In
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. Selection processing for selecting data in the means as processing target data;
A compression method predicted value, which is a predicted value of the compression time required for compressing the processing target data by the data compression means, and the processing target data required to move from the primary storage means to the secondary storage means for writing. A determination process for calculating a movement method prediction value that is a predicted value of a movement time and determining which one of the compression method prediction value and the movement method prediction value is smaller;
In the determination process, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the processing target data by the compression unit is stored in the primary storage unit, while the movement method When it is determined that the predicted value is smaller, the cache control program causes the computer to execute a process of moving and writing the processing target data to the second storage unit.

(Appendix 8)
8. The cache control program according to appendix 6 or 7, wherein the compression method predicted value is a value obtained by adding a time required for decompressing the compressed data to uncompressed data by a data decompressing unit.

(Appendix 9)
The cache control program according to any one of appendices 6 to 8, wherein the movement method predicted value is a value obtained by adding a time required for reading the data from the secondary storage unit and moving it to the primary storage unit.

(Appendix 10)
The system includes a plurality of the data compression means,
The determination process is a process required when the processing target data is divided into a plurality of partial data so as to have a one-to-one correspondence with the plurality of compression units, and each partial data is compressed by the corresponding data compression unit. Use the longest prediction value of time as the compression method prediction value,
Appendix 6, wherein when the determination process determines that the compression method predicted value is smaller, the plurality of partial data are compressed in parallel by the plurality of corresponding data compression means. The cache control program according to 7 or 9.

(Appendix 11)
The system includes a plurality of data compression means and a plurality of data expansion means for expanding compressed data into uncompressed data,
In the determination process, the data to be processed is divided into a plurality of first partial data so as to have a one-to-one correspondence with the plurality of compression units, and each of the first partial data is compressed by the corresponding data compression unit. The longest predicted value of the processing time required at the time and the compressed data obtained by compressing the processing target data are divided into a plurality of second partial data so as to correspond one-to-one with the plurality of data expansion means. Using the value obtained by adding the longest of the predicted values of the processing time required when the second data is expanded by the corresponding data expansion means as the compression method predicted value,
When it is determined that the compression method predicted value is smaller in the determination process, the plurality of first partial data are compressed in parallel by the plurality of corresponding data compression units. The cache control program according to appendix 6, 7 or 9.

DESCRIPTION OF SYMBOLS 10 Arithmetic processing unit 20 Primary storage device 21 Primary cache area 22 Expansion information storage area 23 Compression information storage area 24 Free area 25 Application area 30 Secondary storage device 31 Secondary cache area 32 Expansion information storage area 33 Compression Information storage area 34 Free area 40 Processing data selection section 50 Data compression means 60 Data expansion means 70 Method determination section 100 Computer 200 External information source 101 Computer

Claims (10)

Primary storage means for storing data;
Secondary storage means for reading data slower than the primary storage means;
Data compression means capable of compressing and storing data in the primary storage means;
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. Selecting means for selecting data in the means as processing target data;
And each size of the data, the compression method predicted value is a prediction value of the compression time required to compress the data for each size by the data compressing means, and said secondary storage means data for each size from the primary storage means Any one of the compression method prediction value and the movement method prediction value predicted from the size of the processing target data based on the relationship with the movement method prediction value that is a prediction value of the movement time required to move and write to Determining means for determining whether or not is small;
When the determination means determines that the compression method predicted value is smaller, the compressed data obtained by compressing the data to be processed by the data compression means is stored in the primary storage means, while the movement A cache control device comprising : a processing unit that moves and writes the processing target data to the secondary storage unit when it is determined that the method predicted value is smaller. Primary storage means for storing data;
Secondary storage means for reading data slower than the primary storage means;
Data compression means capable of compressing and storing data in the primary storage means;
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. Selecting means for selecting data in the means as processing target data;
A compression scheme predicted value is a prediction value of the compression time required to compress the processed data by said data compression means, for writing to move the processing target data from said primary storage means in said secondary storage means A determination unit that calculates a movement method predicted value that is a predicted value of the required movement time, and determines which of the compression method predicted value and the movement method predicted value is smaller;
When the determination means determines that the compression method predicted value is smaller, the compressed data obtained by compressing the data to be processed by the data compression means is stored in the primary storage means, while the movement A cache control device comprising : a processing unit that moves and writes the processing target data to the secondary storage unit when it is determined that the method predicted value is smaller. Data expansion means for expanding compressed data into uncompressed data;
The compression method predicted value, the cache control apparatus according to claim 1 or 2, wherein the compressed the data was a value obtained by adding the time required to deploy the data expansion means. A plurality of the data compression means are provided,
The determination unit is required when the processing target data is divided into a plurality of partial data so as to have a one-to-one correspondence with the plurality of data compression units, and each partial data is compressed by the corresponding data compression unit. Using the longest prediction value of processing time as the compression method prediction value,
2. When the determination unit determines that the compression method predicted value is smaller, the plurality of partial data is compressed in parallel by the plurality of corresponding data compression units. Or the cache control apparatus of 2 . A plurality of data compression means and a data expansion means for expanding compressed data into uncompressed data are provided,
The determination unit divides the processing target data into a plurality of first partial data so as to have a one-to-one correspondence with the plurality of data compression units, and compresses each first partial data into the corresponding data compression unit. And dividing the compressed data obtained by compressing the processing target data into a plurality of second partial data so as to have a one-to-one correspondence with the plurality of data expansion means. A value obtained by adding the longest of the predicted values of the processing time required when the second data is expanded by the corresponding data expansion means as the compression method predicted value,
When the determination unit determines that the compression method predicted value is smaller, the plurality of first partial data are compressed in parallel by the plurality of corresponding data compression units. The cache control device according to claim 1 or 2 . The moving method predicted value, any one of claims 1 to 5, characterized in that a value obtained by adding the time required to move the data from the secondary storage means to read said primary storage means the cache control apparatus according to. A primary storage unit that stores data, a secondary storage unit that reads data at a lower speed than the primary storage unit, and a data compression unit that compresses and stores the data in the primary storage unit are used. In the cache control method,
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. A selection step of selecting data in the means as processing target data;
Each size of data, a compression method predicted value that is a predicted value of the compression time required to compress the data of each size by the data compression unit, and data of each size are transferred from the primary storage unit to the secondary storage unit. Which one of the compression method prediction value and the movement method prediction value predicted from the size of the processing target data is based on the relationship with the movement method prediction value that is a predicted value of the movement time required to move and write. A determination step for determining whether the value is small;
In the determination step, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the data to be processed by the data compression unit is stored in the primary storage unit, while the movement And a processing step of moving and writing the processing target data to the secondary storage means when it is determined that the method predicted value is smaller. A primary storage unit that stores data, a secondary storage unit that reads data at a lower speed than the primary storage unit, and a data compression unit that compresses and stores the data in the primary storage unit are used. In the cache control method,
In the case where new data is stored in the primary storage means, and there is no free area for storing the new data in the primary storage means, the primary storage specified based on a predetermined condition. A selection step of selecting data in the means as processing target data;
A compression scheme predicted value is a prediction value of the compression time required to compress the processed data by said data compression means, for writing to move the processing target data from said primary storage means in said secondary storage means A determination step of calculating a movement method predicted value that is a predicted value of the required movement time, and determining which of the compression method predicted value and the movement method predicted value is smaller;
In the determination step, when it is determined that the compression method predicted value is smaller, the compressed data obtained by compressing the data to be processed by the data compression unit is stored in the primary storage unit, while the movement And a processing step of moving and writing the processing target data to the secondary storage means when it is determined that the method predicted value is smaller.   9. The cache control method according to claim 7, wherein the compression method predicted value is a value obtained by adding a time required to expand the compressed data into uncompressed data. The moving method predicted value, in any one of claims 7 to 9, characterized in that the data is a value obtained by adding the time required to move the read primary storage means from said secondary storage means The described cache control method.

Images