JP5220053B2 - Storage medium control device, control program, and control method - Google Patents

Description

  The present invention relates to a control device, a control program, and a control method for a storage medium storing data.

  A file system that manages data stored in a storage medium stores the data by setting the usage status of a sector of the storage medium that has not yet stored data such as a file to “Free” (unused). The sector usage status is managed by setting the sector usage status to “Use” (in use) and the sector usage status from which data has been deleted as “Del” (deleted). When the number of sectors whose usage status is “Del” increases, the free space of the storage medium decreases, so the usage status of the sector is returned from “Del” to “Free” to increase the free space of the storage medium. A technique for performing garbage collection is known (see, for example, Patent Document 1).

JP 2006-99797 A

  However, in the technique disclosed in Patent Document 1, for example, even if processing such as data input / output processing for a storage medium is executed, if the free space of the storage medium decreases, garbage collection (hereinafter simply referred to as GC) Is executed). For this reason, there is a problem that the execution speed of the processing on the storage medium is greatly reduced by executing GC when the load on the CPU (Central Processing Unit) is not low enough to perform garbage collection. .

  The present invention has been made in view of the above circumstances, and an object of the present invention is to provide a storage medium control device, a control program, and a storage medium capable of performing GC while suppressing a decrease in the execution speed of processing on the storage medium. And providing a control method.

In order to achieve the above object, a storage medium control device according to a first aspect of the present invention provides:
Load measuring means for measuring a load of an arithmetic device that executes processing on the storage medium;
The curve representing the load measured by the measurement means load, and peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak width calculating means for calculating the time from when the load measured by the load measuring means exceeds the predetermined threshold until it falls below the predetermined threshold as the peak width of the load peak detected by the peak detecting means When,
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak width calculated by the peak width calculating means. A load predicting means for predicting based on the peak width of the immediately preceding peak that is the load peak that appears last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
It is characterized by providing.
In order to achieve the above object, a storage medium control device according to a second aspect of the present invention provides:
Load measuring means for measuring a load of an arithmetic device that executes processing on the storage medium;
The curve representing the load measured by the measurement means load, and peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak height calculating means for calculating the peak height of the load peak detected by the peak detecting means;
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated by the peak height calculating means. A load predicting means for predicting based on the peak height of the immediately preceding peak that is the load peak that appears last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
It is characterized by providing.

In order to achieve the above object, a storage medium control program according to a third aspect of the present invention provides:
Computer
Load measuring means for measuring a load of an arithmetic device that executes processing on a storage medium;
The curve representing the load measured by the load measuring means, a peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak width calculating means for calculating the time from when the load measured by the load measuring means exceeds the predetermined threshold until it falls below the predetermined threshold as the peak width of the load peak detected by the peak detecting means ,
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak width calculated by the peak width calculating means. A load predicting means for predicting based on the peak width of the immediately preceding peak which is the load peak appearing last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
It is made to function as.
In order to achieve the above object, a storage medium control program according to a fourth aspect of the present invention provides:
Computer
Load measuring means for measuring a load of an arithmetic device that executes processing on a storage medium;
The curve representing the load measured by the load measuring means, a peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak height calculating means for calculating the peak height of the load peak detected by the peak detecting means;
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated by the peak height calculating means. A load predicting means for predicting based on the peak height of the immediately preceding peak which is the load peak last appearing before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
It is made to function as.

In order to achieve the above object, a storage medium control method according to a fifth aspect of the present invention provides:
A load measuring step for measuring a load of an arithmetic device that executes processing for the storage medium;
The curve representing the load measuring load measured in step, a peak detection step of detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
A peak width calculating step for calculating a time from when the load measured at the load measuring step exceeds the predetermined threshold to when the load falls below the predetermined threshold as a peak width of the load peak detected at the peak detecting step. When,
Whether or not the load of the arithmetic device is lower than a predetermined threshold in the required period required for the arithmetic device to execute garbage collection on the storage medium is determined by the peak width calculated in the peak width calculating step. A load prediction step for predicting based on the peak width of the immediately preceding peak that is the load peak that appeared last before the current measurement by the load measuring step;
A garbage collection step for performing garbage collection on the storage medium when the load is predicted to be low in the required period in the load prediction step;
It is characterized by having.
In order to achieve the above object, a storage medium control method according to a sixth aspect of the present invention includes:
A load measuring step for measuring a load of an arithmetic device that executes processing for the storage medium;
The curve representing the load measuring load measured in step, a peak detection step of detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
A peak height calculating step for calculating a peak height of the load peak detected in the peak detecting step;
Whether or not the load on the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated in the peak height calculating step. A load prediction step for predicting based on the peak height of the immediately preceding peak that is the load peak that appeared last before the current measurement by the load measuring step;
A garbage collection step for performing garbage collection on the storage medium when the load is predicted to be low in the required period in the load prediction step;
It is characterized by having.

  According to the storage medium control device, the control program, and the control method according to the present invention, it is possible to perform GC while suppressing a decrease in the execution speed of processing for the storage medium.

FIG. 1A is a hardware configuration diagram illustrating an example of a storage medium control device according to the first embodiment of the present invention. FIG. 1B is a functional block diagram illustrating an example of functions of the storage medium control device according to the first embodiment of the present invention. FIG. 2A is a diagram illustrating a configuration example of the flash memory. FIG. 2B is a diagram illustrating a configuration example of sectors included in the flash memory. FIG. 3A is a flowchart illustrating an example of information calculation processing executed by the information calculation unit. FIG. 3B is a flowchart illustrating an example of load statistical information calculation processing executed by the load statistical information calculation unit. FIG. 3C is a flowchart illustrating an example of the start of the load related information calculation process executed by the load related information calculation unit. FIG. 4A is a diagram illustrating a configuration example of the file system control unit. FIG. 4B is a diagram illustrating a configuration example of the load statistical information calculation unit according to the first embodiment. FIG. 4C is a diagram illustrating a configuration example of the load related information calculation unit according to the first embodiment. It is a graph showing an example of the time transition of the load measured in 1st Embodiment. FIG. 6A is a flowchart illustrating an example of a file system control process (FS control process) executed by the file system control unit (FS control unit). FIG. 6B is a flowchart illustrating an example of a load prediction process executed by the load prediction unit according to the first embodiment. It is a flowchart showing an example of the execution condition determination process which an execution condition determination part performs. FIG. 8A is a diagram illustrating a configuration example of the load statistical information calculation unit according to the second embodiment. FIG. 8B is a diagram illustrating a configuration example of the load related information calculation unit according to the second embodiment. FIG. 8C is a flowchart illustrating an example of the correlation coefficient calculation process executed by the correlation coefficient calculation unit. FIG. 9A is a graph showing an example of the time transition of the moving average value calculated for the load measured in the second embodiment. FIG. 9B is a graph showing the correlation between the peak width and the peak interval. It is a flowchart showing an example of the load prediction process which the load prediction part of 2nd Embodiment performs. FIG. 11A is a diagram illustrating a configuration example of the load statistics information calculation unit of the third embodiment. FIG. 11B is a diagram illustrating a configuration example of the load related information calculation unit according to the third embodiment. FIG. 11C is a flowchart illustrating an example of a load related information calculation process executed by the load related information calculation unit of the third embodiment. It is a flowchart showing an example of the load prediction process which the load prediction part of 3rd Embodiment performs. It is a graph showing an example of the time transition of the moving average value calculated with respect to the load measured in 3rd Embodiment.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described with reference to the accompanying drawings.

<First Embodiment>
The control device 10 according to the first embodiment of the present invention executes various applications, thereby transferring various data to a flash memory (hereinafter simply referred to as a memory) 20 that is a storage medium as illustrated in FIG. Save. Further, when the free capacity of the memory 20 decreases, the control device 10 performs GC (garbage collection) on the memory 20 in order to increase the free capacity. Note that the control device 10 may store data in a hard disk instead of the memory 20.

  As shown in FIG. 1A, the control device 10 includes a CPU (Central Processing Unit) 10a, a ROM (Read Only Memory) 10b, a RAM (Random Access Memory) 10c, a hard disk 10d, a media controller 10e, a USB (Universal Serial). Bus 10f, LAN card (Local Area Network) 10g, video card 10h, LCD (Liquid Crystal Display) 10i, keyboard 10j, and pointing device (hereinafter simply referred to as mouse) 10k.

  The CPU 10a is an arithmetic device that performs overall control of the control device 10 by executing software processing in accordance with a program stored in the ROM 10b or the hard disk 10d. The RAM 10c temporarily stores data to be processed when the CPU 10a executes the program.

  The hard disk 10d stores various application programs (hereinafter simply referred to as applications) and various data used by the applications. The media controller 10 e inputs and outputs various data to and from the memory 20. The LAN card 10g transmits and receives data to and from devices connected via a computer communication network such as the Internet, for example.

  The video card 10h outputs an image signal based on the digital signal output from the CPU 10a. The LCD 10i displays an image according to the image signal output from the video card 10h. The control device 10 may include a PDP (Plasma Display Panel) or an EL (Electro Luminescence) display instead of the LCD 10i.

  The keyboard 10j and the mouse k input signals according to user operations. The web control device 10 may include a touch panel instead of the keyboard 10j and the mouse k.

  The CPU 10a included in the control device 10 functions as the application execution unit (hereinafter referred to as a unit, application execution unit) 11 in FIG. 1B by executing the application stored in the hard disk 10d. Further, the CPU 10a executes a process for controlling a file system (hereinafter simply referred to as FS) used for managing data such as an electronic file (hereinafter simply referred to as a file) stored in the memory 20 to thereby execute a file. It functions as a system control unit (hereinafter simply referred to as FS control unit) 100.

  The application execution unit 11 illustrated in FIG. 1B performs opening / closing, writing, overwriting, and deletion of a file stored in the memory 20 via the FS control unit 100.

Next, before describing the FS control unit 100, the memory 20 controlled by the FS control unit 100 will be described once.
The memory 20 has a plurality of blocks including blocks 21 and 22 as shown in FIG. These blocks have a plurality of sectors like the block 21, for example. As shown in FIG. 2B, the sector sc1 included in the block 21 shown in FIG. 2 stores the usage status of the sector sc1, a logical number for identifying the sector sc1, and data.

Here, a case where the FS control unit 100 writes a file in the memory 20 will be described.
First, according to the file size, the FS control unit 100 secures a plurality of sectors that have not yet stored data (that is, the usage status is “Free” (unused)) and a file for the secured sectors. Write data. Thereafter, the control device 10 changes the usage status of the sector in which the data is written to “Use” (in use).

Next, a case where the FS control unit 100 overwrites a file in the memory 20 will be described.
The FS control unit 100 transfers the updated file data to a sector whose usage status is “Free”, which is different from the sector in which the file data before the update is written (that is, the usage status is “Use”). Write. Next, the FS control unit 100 associates the sector that stores the updated file data with the sector that stores the pre-update file data, and then determines the usage status of the sector that stores the pre-update file data as “Del ”(Deleted). This is because the file data before update can be recovered as necessary.

  On the other hand, when deleting data from the memory 20, the FS control unit 100 does not set the usage status to “Free” for each sector, but sets the sector usage status to “Free” for each block. In other words, the data stored in the memory 20 is deleted by setting the usage status of all sectors in a certain block to “Free”.

  For this reason, when the data stored in the memory 20 is rewritten repeatedly by the FS control unit 100, the sector whose usage status is “Del” increases and the sector whose usage status is “Free” is depleted. Therefore, the FS control unit 100 executes GC to return the sector usage status “Del” to “Free” at an appropriate time.

  For this reason, the Fs control unit 100 calculates information used for determining whether or not a GC execution condition established at an appropriate time for executing the GC is satisfied, as shown in FIG. 3A. Execute.

  The FS control unit 100 includes a load measurement unit 110, a load information database (hereinafter referred to as load information DB) 120, an information calculation unit 130, and a statistical information database (hereinafter referred to as statistical information DB) as shown in FIG. ) 140, for example, the information calculation process of FIG. 3A is executed once every few days.

  When the execution of the information calculation process of FIG. 3A is started, the information calculation unit 130 of FIG. 4A reads the load information stored in the load information DB 120 (step S01). The load information stored in the load information DB 120 is information representing the load on the CPU 10a (hereinafter referred to as CPU load) shown in FIG. The load information is stored in the load information DB 120 in association with measurement time information indicating the time when the load measurement unit 110 measures the load represented by the load information.

  Here, since the control device 10 is a device used for daily work, for example, the control device 10 performs predetermined processing in units of one day by executing the application. For this reason, since the CPU 10a in FIG. 1A periodically processes the data stored in the memory 20, the CPU 10a periodically performs processing for the FS (file system) that manages the files stored in the memory 20. To perform input / output processing on the memory 20 periodically. Therefore, the CPU load repeats the high load state and the low load state almost regularly many times during the day.

  Therefore, the FS control unit 100 reads the load information indicating the load measured at the time from the execution time of step S01 to several days ago. Note that the load may be represented by, for example, a CPU usage rate measured by an OS (Operating System).

  After step S01 is executed, the information calculation unit 130 reads information representing the GC execution history (hereinafter referred to as execution history information) from the execution history DB 190 (step S02). The execution history information includes a plurality of information indicating the time when GC is started and information indicating the time when GC is ended.

  Thereafter, the information calculation unit 130 corrects the value of the load information by subtracting the load generated by the execution of the GC from the load measured at the execution time of the GC represented by the execution history information (step S03). Note that the load generated by the execution of GC may be a value set in advance based on the calculation processing capability of the CPU 10a and the calculation amount required for the GC shown in FIG.

  Next, the load statistical information calculation unit 131 included in the information calculation unit 130 executes a load statistical information calculation process as illustrated in FIG. 3B for calculating the load statistical information (step S04). The load statistical information refers to information calculated by performing various statistical processes on the load information representing the load measured at each measurement time. In this embodiment, the load statistical information calculation unit 131 calculates load statistical information that represents an arithmetic average value (that is, arithmetic average value) of the load represented by the load information and load statistical information that represents a weighted average value of the load. To do. The load statistical information calculation unit 131 includes an average value calculation unit 131a and a moving average value calculation unit 131b as illustrated in FIG.

  When the execution of the load statistical information illustrated in FIG. 3B is started, the average value calculation unit 131a illustrated in FIG. 4B performs the load represented by the load information corrected in step S03 illustrated in FIG. Is calculated (step S11). Next, the load statistical information calculation unit 131 sets the calculated average value of the loads as a threshold Th used by the load related information calculation unit 132 (step S12). The threshold Th will be described later.

  Next, the average value calculation unit 131b in FIG. 4B performs a predetermined time from the measurement time at which the load represented by the load information is measured for each of the load information corrected in step S03 in FIG. Calculate a moving average value of the load information and a plurality of other pieces of load information representing loads measured in a section from the previous time to a time after the measurement time to a predetermined time (hereinafter referred to as a moving average section). (Step S13). This is because a high-frequency component generated in the short term is represented in a curve P representing a transition of a load represented in a coordinate space having an L-axis representing a load and a T-axis representing a load measurement time as shown in FIG. This is because the curve MA is smoothed and a low-frequency component that is a main component is extracted.

  Next, the load statistical information calculation unit 131 stores the load statistical information calculated in steps S11 and S13 in the statistical information DB 140 (step S14). Thereafter, the load statistical information calculation unit 131 ends the execution of the load statistical information calculation process.

  After step S04 in FIG. 3 (a) is completed, the load related information calculation unit 132 included in the information calculation unit 130 in FIG. 4 (a) loads the load related information as shown in FIG. 3 (c). Execute related information calculation processing. The load related information is information related to a load peak detected using load statistical information. In the present embodiment, the load related information calculation unit 132 calculates load related information representing a load peak width (hereinafter referred to as a peak width).

  Here, the load peak means a portion PK in which the smoothed load (that is, the moving average value of the load) is larger than a predetermined threshold Th in the curve MA as shown in FIG. Not that. The peak width refers to the time from time t12 when the smoothed load exceeds the threshold Th to time t14 when the load falls below the threshold Th. Note that the threshold value Th indicates that the execution speed of the process on the memory 20 decreases when the GC is executed in a state where the CPU load is increased beyond this value.

  When the execution of the load related information calculation process in FIG. 3C is started, the peak detection unit 132a included in the load related information calculation unit 132 in FIG. 4C is calculated in step S13 in FIG. A portion where the curve MA representing the transition of the moving average value of the load exceeds the threshold Th defined in step S12 is detected as a load peak (step S21). Next, the peak width calculation unit 132b calculates the peak width of the detected load peak (step S22). Next, the load related information calculation unit 132 stores the load related information calculated in step S22 in the statistical information DB 140 (step S23). Thereafter, the load related information calculation unit 132 ends the execution of the load related information calculation process.

  Next, the FS control unit 100 in FIG. 4A uses the calculated load statistical information and load related information (hereinafter simply referred to as statistical information) to control the file system, as shown in FIG. FS control processing is executed.

  The FS control unit 100 includes not only the load measurement unit 110, the load information DB 120, the information calculation unit 130, and the statistical information DB 140, but also the used sector information acquisition unit 150, the load prediction unit 170, FIG. For example, once every few minutes using the execution condition determination unit 180, the GC schedule unit 185, the garbage collection unit (hereinafter referred to as GC unit) 189, and the execution history database (hereinafter referred to as execution history DB) 190, FIG. The FS control process (a) is executed.

  When the execution of the FS control process in FIG. 6A is started, the load prediction unit 170 in FIG. 4A reads out statistical information including the threshold value Th stored in the statistical information DB 140 (step S31). Next, the load measuring unit 110 measures the load of the CPU 10a at the present time (that is, the execution time of step S32), and stores the load information representing the measured load in the load information DB 120 (step S32).

  Thereafter, the load statistical information calculation unit 131 calculates the average value of the peak widths (of peaks that appeared in the past several days) calculated in step S22 of FIG. (Referred to as “average peak width”). Next, the load statistical information calculation unit 131 calculates the moving average value of the load measured from the current time to the moving average section (step S33). According to these configurations, for example, even when any one or more of the memory 20 and the user of the memory 20 changes and the magnitude of the peak width of the load peak changes, a peak with a width larger than the average peak width ( In other words, the low frequency component) can be emphasized and detected.

  Next, the load related information calculation unit 132 detects a load peak that exceeds the threshold Th read in step S31 from the curve representing the time transition of the moving average value calculated in step S33 (step S34). Thereafter, the load prediction unit 170 performs a load prediction process as shown in FIG. 6B, in which a load at a time later than the current time is predicted.

Here, the time t11 shown in FIG. 5 is taken as an example, and the load prediction process executed at this time will be described.
When the execution of the load prediction process in FIG. 6B is started, the load prediction unit 170 determines whether or not the moving average value of the load measured at time t11 forms a load peak (that is, greater than the threshold Th). Is determined (step S41), and it is determined that a load peak is not formed (step S41; No).

  Next, at time t11, the load prediction unit 170 determines that the moving average value of the load measured at time t11 is greater than the moving average value of the load measured last time (that is, time t10 before time t11). Since it is large, it is determined that the moving average value of the load is not decreasing (that is, increasing) (step S42; No). Next, the load predicting unit 170 predicts that the load of the CPU 10a that executes GC is a high load from the current time t11 until the time required for execution of GC elapses (step S45). This is because the moving average value of the load is currently increasing. Thereafter, the load prediction unit 170 ends the execution of the load prediction process.

  Note that the load of the CPU 10a is high (low load) means that when the CPU 10a executes GC on the memory 20, the execution speed of the input / output processing on the memory 20 by the CPU 10a decreases below a predetermined value. This means that the load on the CPU 10a is high (low).

Next, taking time t13 in FIG. 5 as an example, load prediction processing executed at this time will be described.
When the execution of the load prediction process of FIG. 6B is started, the load prediction unit 170 determines that the moving average value of the load measured at time t13 forms a load peak (step S41; Yes). Next, the load prediction unit 170 predicts that the CPU load is high until the required time elapses from the current time t13 (step S45), and ends the execution of the load prediction process.

Next, taking time t15 in FIG. 5 as an example, load prediction processing executed at this time will be described.
When the execution of the load prediction process in FIG. 6B is started, after executing the process of step S41, the load prediction unit 170 determines that the moving average value of the load is decreasing (step S42; Yes). ). Next, the load prediction unit 170 determines that the moving average value of the load is not always below the threshold Th from the current time t15 to a time before the time Tc (step S43; No). Thereafter, the load prediction unit 170 predicts that the CPU load is high until the required time elapses from the current time t15 (step S45), and ends the execution of the load prediction process. This is because it is uncertain whether or not the decrease in CPU load is temporary.

Next, taking time t16 in FIG. 5 as an example, load prediction processing executed at this time will be described.
When the execution of the load prediction process of FIG. 6B is started, the load prediction unit 170 moves the load from the current time t15 to a time before the time Tc after executing the processes of step S41 and step S42. It is determined that the average value is always below the threshold Th (Step S43; Yes). Thereafter, the load prediction unit 170 predicts that the CPU load is low until the required time elapses from the current time t16 (step S44), and ends the execution of the load prediction process. This is because the CPU load has fallen below the threshold value Th over time Tc, and therefore it is determined that there is a high probability that the load is not temporarily reduced.

  According to these configurations, it is determined that the CPU load is low when the moving average value of the load falls below a predetermined threshold Th, and therefore it is possible to accurately predict whether the CPU load is low during the required GC period. .

  In addition, according to these configurations, when the moving average value of the load that has been lower than the predetermined threshold Th for a predetermined number of times (that is, the predetermined time Tc) further decreases, it is determined that the CPU load is low. Therefore, it can be predicted more accurately whether the CPU load is low in the required period of GC.

  After the execution of the load prediction process is completed in step S35 in FIG. 6A, the used sector information acquisition unit 150 in FIG. 4A obtains the used sector information indicating the usage status of the sectors included in the memory 20 as described above. Obtained from the file system (FS) (step S36). Further, the used sector information acquisition unit 150 has the number of sectors whose usage status is “Free” (hereinafter simply referred to as the number of free sectors) “NumFreeSec” and the block of the memory 20 based on the acquired used sector information. Get the number of sectors “NumEBSec”.

  Next, the execution condition determination unit 180 executes an execution condition determination process as shown in FIG. 7 for determining whether or not the GC execution condition is satisfied (step S37).

  First, an execution condition determination process that is executed when there are sufficient free sectors in the memory 20 will be described.

  When the execution of the execution condition determination process shown in FIG. 7 is started, the execution condition determination unit 180 determines whether or not a first condition represented by the following expression (1) is satisfied (step S51).

NumEBSec * C>NumFreeSec> NumEBSec (1)
Where C = 2

  Here, the usage status of the sectors of the memory 20 is returned to “Free” (unused) in units of blocks. For this reason, as shown in FIG. 2A, in the execution of GC, after searching for the block “EBDelMax” having the largest number of sectors whose usage status is “Del” (deleted), the searched block “EBDelMax” Before returning the sector to “Free”, the data stored in the sector whose usage status is “Use” (used) in the block “EBDelMax” is temporarily changed to “Free”. It is necessary to save (copy) to the sector. For this reason, when the number of empty sectors “NumFreeSec” is less than the number of sectors “NumEBSec” of the block, the execution of GC becomes impossible. The usage status “Mov” (moving) indicates that data is being written to the sector, and “Err” (error) indicates that the sector has become unusable.

  Therefore, the first condition is satisfied because the number of empty sectors “NumFreeSec” in the memory 20 is larger than the minimum number “NumEBSec” necessary for the execution of GC, and therefore it is not necessary to immediately execute the GC. "Indicates that it is necessary to execute GC before the number of empty sectors is reduced. In the present embodiment, the case where the variable C is the value “2” has been described. However, the present invention is not limited to this, and a person skilled in the art can determine an optimum value.

  In step S51, the execution condition determining unit 180 determines that the first condition is not satisfied because there are sufficient free sectors in the memory 20 (step S51; No). Next, the execution condition determination unit 180 determines whether or not the second condition represented by the following expression (2) is satisfied (step S52).

  NumFreeSec = NumEBSec (2)

  Satisfaction of the second condition indicates that the number of empty sectors “NumFreeSec” in the memory 20 is equal to the minimum number “NumEBSec” necessary for execution of GC, and therefore it is necessary to immediately execute GC.

  In step S52, the execution condition determination unit 180 determines that the second condition is not satisfied (step S52; No), and determines that the GC execution condition is not satisfied (step S53). This is because there are so many empty sectors in the memory 20 that it is not necessary to execute GC. Thereafter, the execution condition determination unit 180 ends the execution of the execution condition determination process.

  Next, an execution condition determination process executed when the file 20 is repeatedly overwritten on the memory 20 to reduce the number of empty sectors will be described.

  The execution condition determination unit 180 determines that the first condition is satisfied due to the decrease in empty sectors (step S51; Yes). Next, the execution condition determination unit 180 determines whether or not the CPU load predicted in step S35 of FIG. 6A is a low load (step S54).

  In step S54, when the execution condition determining unit 180 determines that the predicted load is not low (step S54; No), the execution condition determining unit 180 determines that the GC execution condition is not satisfied (step S55). This is because it is not necessary to immediately execute the GC, and if the GC is executed, the execution speed of the input / output processing with respect to the memory 20 may be significantly reduced (that is, lower than a predetermined value). Thereafter, the execution condition determination unit 180 ends the execution of the execution condition determination process.

  Conversely, when the execution condition determination unit 180 determines in step S54 that the predicted load is low (step S54; Yes), it determines that the GC execution condition is satisfied (step S56). Before the “NumEBSec” free sectors are reduced, it is not only necessary to execute the GC, but even if the GC is executed, the execution speed of the input / output processing for the memory 20 is remarkably high (that is, a predetermined value). This is because the possibility of lowering is low. Thereafter, the execution condition determination unit 180 ends the execution of the execution condition determination process.

  Next, a description will be given of an execution condition determination process that is executed when the file 20 is repeatedly overwritten on the memory 20 and empty sectors are exhausted.

  The execution condition determination unit 180 determines that the number of empty sectors “NumFreeSec” continues to decrease and reaches the same number as the number of sectors “NumEBSec” of one block. The first condition is not satisfied but the second condition is satisfied Then, it judges (Step S51; No and Step S52; Yes). Next, the execution condition determination unit 180 determines that the GC execution condition is satisfied without determining the predicted level of CPU load (step S56). This is because it is necessary to execute the GC before the GC cannot be executed whether or not the CPU load is high. Thereafter, the execution condition determination unit 180 ends the execution of the execution condition determination process.

  When execution of the execution condition determination process ends in step S37 of FIG. 6A, the GC schedule unit 185 determines whether it is determined that the GC execution condition is satisfied (step S38).

  If it is determined that the GC execution condition is not satisfied (step S38; No), the FS control unit 100 returns to step S32 and repeats the above processing.

  If it is determined that the GC execution condition is satisfied (step S38; Yes), the GC scheduling unit 185 schedules the GC unit 189 to perform GC immediately (step S39). After that, the FS control unit 100 returns to step S32 and repeats the above process. Also, the GC unit 189 in FIG. 4A immediately executes the GC according to the schedule changed by the GC schedule unit 185, and stores the GC execution history information in the execution history DB 190.

  According to these configurations, since the GC of the memory 20 is performed at the time when the CPU load is predicted to be low, the execution speed of the process executed by the CPU 10a of FIG. it can.

<Modification 1>
In this embodiment, as shown in FIG. 6B, the load prediction unit 170 in FIG. 4A is not a load peak at the measurement time (step S41; No), and the load is decreasing (step S42). Yes), when the moving average value of the CPU load falls below the threshold Th from the measurement time to the predetermined time Tc (step S43; Yes), the CPU load is predicted to be low (step S44). explained.

  However, in a modified example, if the load prediction unit 170 is not a load peak at the measurement time and the load is decreasing, the CPU load is low without determining the CPU load before the predetermined time Tc from the measurement time. Predict that

  According to this configuration, it is possible to predict whether the CPU load is sufficiently low in a required time required for GC in a short amount of calculation and in a short time.

<Modification 2>
In the present embodiment, the control apparatus 10 has been described as executing GC on the memory 20, but the present invention is not limited to this. For example, as in the hard disk 10d in FIG. You may perform GC with respect to the storage medium which has.

Second Embodiment
Next, the second embodiment will be described.
The control device according to the second embodiment predicts the peak interval that occurs after the load measurement time (hereinafter simply referred to as “after measurement”) based on the correlation between the peak width and the peak interval, and the predicted peak interval is the GC. When the required time is longer than the required time, the CPU load is predicted to be low until the required time elapses from the measurement time. Note that detailed description of portions common to the first embodiment is omitted.

  The control device according to the second embodiment has substantially the same configuration as the control device 10 according to the first embodiment, but the load information calculation unit 231 illustrated in FIG. 8A and the load-related information illustrated in FIG. The configuration of the information calculation unit 232 is different.

Before describing the load statistical information calculation unit 231, the load related information calculation unit 232 will be described.
The load related information calculation unit 232 includes a peak interval calculation unit 232c in addition to the peak detection unit 232a and the peak width calculation unit 232b in FIG. In the present embodiment, the load related information calculation unit 232 calculates load related information representing a peak interval in addition to load related information representing a peak width.

  Here, the peak interval means an interval WL between the peak PK1 and the peak PK2 as shown in FIG. 9, and the load moving average value is less than the threshold Th from time t22 when the moving average value of the load is less than the threshold Th. This is the next time until time t23.

  The load statistical information calculation unit 231 in FIG. 8A includes a correlation coefficient calculation unit 231c in addition to the average value calculation unit 231a and the moving average value calculation unit 231b.

  After the load statistical information calculation unit 231 executes the load statistical information calculation process in FIG. 3B, the load related information calculation unit 232 executes the load related information calculation process for calculating the load related information. Execution of the correlation coefficient calculation process as shown in FIG. 8C, which calculates a correlation coefficient representing the correlation between the width and the peak interval, is started.

  When the execution of the correlation coefficient calculation process shown in FIG. 8C is started, the correlation coefficient calculation unit 231c, as shown in FIG. 9B, the peak width calculated by the load related information calculation unit 232, A regression line RL with the peak interval is calculated (step S61). FIG. 9B is a diagram in which the peak width and the peak interval are plotted for each measured load peak with respect to a coordinate system in which the horizontal axis represents the peak width and the vertical axis represents the peak interval.

  Next, the correlation coefficient calculation unit 231c determines whether or not the peak width and the peak interval have a positive correlation based on the value of the correlation coefficient used for the regression line RL (step S62). As a specific example, the correlation coefficient calculation unit 231c determines that there is a positive correlation when the correlation coefficient satisfies the following expression (3).

  0.5 <correlation coefficient <1.0 (3)

  In step S62, when the correlation coefficient calculation unit 231c determines that there is a positive correlation (step S62; Yes), for example, the correlation coefficient calculation unit 231c reads the GC from the execution history DB included in the control device. While acquiring execution history information, the average value of the execution time of GC which execution history information represents is calculated, and let the calculated average value be the required time Gt required for execution of GC.

  Thereafter, as shown in FIG. 9B, the correlation coefficient calculation unit 231c calculates the peak width corresponding to the peak interval having a length twice the GC required time Gt using the regression line RL. At the same time, the calculated peak width is set as a peak width threshold Tw (step S62). Since the peak interval and the peak width have a positive correlation, when the peak width of the load peak that appears last before the load measurement (hereinafter referred to as the immediately preceding peak) exceeds the threshold Tw, the immediately preceding peak and the first after the measurement This is because the peak interval between the load peak that appears (hereinafter referred to as the peak immediately after) is likely to be longer (about twice) than the time required for GC.

  In the present embodiment, the peak width corresponding to the peak interval having a length twice as long as the GC required time Gt has been described as the threshold value Tw. However, the present invention is not limited to this value “2”. Further, the peak width corresponding to the peak interval having the length of GC required time Gt (that is, a length of 1 time) is not set as the threshold value Tw, but is twice as long as the GC required time Gt. By setting the peak width corresponding to the peak interval as the threshold value Tw, it is possible to prevent a state in which the CPU load is high such that the next peak occurs immediately after the execution of GC.

  In step S62, if the correlation coefficient calculation unit 231c determines that there is no positive correlation (step S62; No), the peak width threshold Tw is set to a predetermined value (step S64).

  After executing Steps S63 and S64, the correlation coefficient calculation unit 231c stores the calculated or set peak width threshold Tw as statistical information in the statistical information DB of the control device (Step S65). Thereafter, the correlation coefficient calculation unit 231c ends the execution of the correlation coefficient calculation process.

The control device according to the present embodiment executes the FS control process of FIG. 6A using the statistical information stored in step S65 and the like. In step S35, the load prediction as shown in FIG. It differs from the first embodiment in that processing is executed.
Therefore, the load prediction process executed at this time will be described below by taking the time t22 shown in FIG. 9A as an example.

  When the execution of the load prediction process of FIG. 10 is started, the load prediction unit included in the control device executes the same process as steps S41 and S42 of FIG. 6B (steps S71 and S72). Next, after determining that the moving average value of the load is decreasing at the current time t22 (step S72; Yes), the load prediction unit determines that the peak width W1 of the immediately preceding peak PK1 at the current time t22 is as shown in FIG. It is determined whether it is larger than the threshold value Tw calculated in step S63 of c) or set in step S64 (step S73).

  In step S73, when the load prediction unit determines that the peak width W1 of the load peak PK1 is larger than the threshold value Tw (step S73; Yes), the peak interval WL between the immediately preceding peak PK1 and the immediately following peak PK2 is the time required for GC. I guess it is longer. Thereafter, the load prediction unit predicts that the CPU load is low until the time required for GC elapses after the load is measured (step S74), and ends the execution of the load prediction process.

  In step S73, when the load prediction unit determines that the peak width W1 of the load peak PK1 is equal to or less than the threshold Tw (step S73; No), the peak interval WL between the immediately preceding peak PK1 and the immediately following peak PK2 is the time required for GC. I guess it is shorter. Thereafter, the load prediction unit predicts that the CPU load is high until the time required for GC elapses after the load is measured (step S75), and ends the load prediction process.

  Usually, after a peak with a long peak width ends, a long peak interval tends to come. Therefore, according to these configurations, the CPU load is reduced in the required period of GC after measurement based on the peak width of the immediately preceding peak. Whether the load is low or not can be accurately predicted.

  In addition, according to these configurations, when the peak width and the peak interval have a correlation, it is predicted whether the CPU load is low based on the peak width of the immediately preceding peak. Can be predicted.

  According to these configurations, when the peak interval estimated based on the peak width and the correlation coefficient is longer than the required time required for the GC, the CPU load is predicted to be low, so the load is more accurately detected. Can be predicted.

<Third Embodiment>
Next, the third embodiment will be described.
The control device according to the third embodiment predicts whether the CPU load is sufficiently low based on the peak height of the immediately preceding peak. Note that detailed description of portions common to the first embodiment is omitted.

  The control device according to the third embodiment has substantially the same configuration as the control device 10 according to the first embodiment, but the load information calculation unit 331 illustrated in FIG. 11A and the load-related information illustrated in FIG. The configuration of the information calculation unit 332 is different.

  The load information calculation unit 331 in FIG. 11A includes a non-peak average value calculation unit 331c in addition to the average value calculation unit 331a and the moving average value calculation unit 331b. The non-peak average value calculation unit 331c does not represent the load peak detected by the load related information calculation unit 332 in FIG. 11B, but calculates the arithmetic average value for the moving average value of the load representing the peak interval. Then, the calculated average value is set to a threshold value Tl described later. This threshold value Tl is lower than or equal to the threshold value Th that is the average value of the load.

  The load related information calculation unit 332 in FIG. 11B includes a peak classification unit 332c and a low load transition time calculation unit 332d in addition to the peak detection unit 332a and the peak height calculation unit 332b in FIG. In the present embodiment, the load related information calculation unit 332 represents the low load transition time in addition to the load related information representing the height of the load peak (hereinafter referred to as peak height) as shown in FIG. A load related information calculation process for calculating the load related information is executed.

  Here, the low load transition time is a threshold Tl lower than the threshold Th from a time t31 when the moving average value of the load is lower than the threshold Th (that is, a time when the load peak ends) as shown in FIG. The time until the time t32 which falls first is said. The threshold value Tl represents a value measured when the CPU load is stabilized at a sufficiently low value, and is set to a value calculated by the non-peak average value calculation unit 331c as described above. The peak height is represented by the maximum value of the moving average value of the load at the peak.

  When the load related information calculation process of FIG. 11C is started, the peak detection unit 332a of FIG. 11B executes the same process as step S21 of FIG. 3C (step S81). Next, the peak height calculation unit 332b calculates the peak height of the load peak detected in step S81 (step S82).

  Next, the peak classification unit 332c calculates the distribution of peak heights of the load peaks detected in step S81 (that is, which appeared in the past few days) (step S83). Next, the peak classification unit 332c classifies the calculated distribution with the lower 50% as the low level peak height and the upper 50% as the high level peak height (step S84). Thereafter, the peak classification unit 332c sets the classification to which the load peak having a low level peak height belongs to the first classification, and sets the classification to which the load peak having a high level peak height belongs to the second classification (step S85).

  Thereafter, the low load transition time calculation unit 332d calculates the low load transition time of the load peak belonging to the second category (step S86). Next, the low load transition time calculation unit 332d executes the same process as step S23 of FIG. 3C (step S87). Thereafter, the execution of the load related information calculation process is terminated.

The control device according to the present embodiment executes the FS control process of FIG. 6A using the load related information stored in the above step S87. In step S35, the load as shown in FIG. It differs from the first embodiment in that a prediction process is executed.
Therefore, the load prediction process executed at this time will be described below by taking time t30 shown in FIG. 13 as an example.

  When the execution of the load prediction process of FIG. 12 is started, the load prediction unit included in the control device executes the same process as steps S41 and S42 of FIG. 6B (steps S91 and S92). When the load prediction unit determines that the moving average value of the load is decreasing at time t30 in FIG. 13 (step S92; Yes), the peak height h1 of the peak PK1 immediately before the current time t30 is shown in FIG. It is determined whether or not the high level of the top 50% classified in step S94 of c) (step S93).

  The load prediction unit determines that the peak height h1 of the immediately preceding peak PK1 shown in FIG. 13 is not a high level (that is, a low level) (step S93; No), and in the required period of GC after time t30, the CPU load Is determined to have a low load (step S95). This is because, usually, when the low-level peak ends, the CPU load often becomes stable and low in a time shorter than the time required to start GC. Thereafter, the load prediction unit ends the execution of the load prediction process.

Next, taking time t31 shown in FIG. 13 as an example, load prediction processing executed at this time will be described.
When the execution of the load prediction process of FIG. 12 is started, the load prediction unit included in the control device executes the processes of steps S91 and S92, and then the peak height h1 of the peak PK1 immediately before the current time t31 is the top 50%. It is judged that it is the high level (step S93; Yes).

  Thereafter, the load prediction unit calculates the load peak belonging to the second classification (that is, the load peak having a high peak height) calculated in step S86 of FIG. 11C from the time t31 when the immediately preceding peak PK1 ends. It is determined whether or not the average time Tt of the low load transition time has elapsed (step S94).

  Since the current time is t31, the load predicting unit determines that the above average time Tt has not elapsed (step S94; No), and the CPU load is high during the required period of GC after the current time t30. (Step S96). This is because, normally, when the high level peak ends, the CPU load becomes stable and becomes a low load, and therefore it takes longer than the time required to start GC. Thereafter, the load prediction unit ends the execution of the load prediction process.

Next, taking time t32 shown in FIG. 13 as an example, load prediction processing executed at this time will be described.
When the execution of the load prediction process of FIG. 12 is started, the load prediction unit included in the control device executes the processes of steps S91 to S93, and then the current time t32 is a low load from the time t31 when the immediately preceding peak PK1 ends. It is determined that the average time Tt of the transition time has elapsed (step S94; Yes).

  Next, the load predicting unit determines that the CPU load is low in the required period of GC after the current time t32 (step S95). This is because the CPU load becomes stable and low load when the average time Tt of the low load transition time elapses after the high level peak ends. Thereafter, the load prediction unit ends the execution of the load prediction process.

  According to these configurations, after a load peak with a high peak height ends, a long peak interval tends to come. Therefore, it can be accurately predicted whether the CPU load is low based on the peak height.

  In addition, according to these configurations, in order to determine the level of the peak height of the immediately preceding peak based on the peak height distribution, for example, any one or more of the memory 20 and the user of the memory 20 change and the load peak is Even if the tendency of the peak height is changed, it can be accurately predicted whether or not the CPU load is low based on the peak height.

  Usually, it takes time until the CPU load is stabilized at a low value after the load peak ends. Therefore, according to these configurations, it is determined that the CPU load is low after the low load transition time has elapsed after the load peak ends, and therefore the load level can be accurately predicted.

  In addition, it is possible to provide a storage medium control apparatus having a configuration for realizing the functions according to the present invention in advance, and to make an existing control apparatus function as a storage medium control apparatus according to the present invention by applying a program. You can also. That is, according to the present invention, a control program for realizing each functional configuration by the control device 10 exemplified in the above embodiment is applied so that a computer (such as a CPU) that controls the existing control device can be executed. It can function as the control device 10.

  Such a program distribution method is arbitrary. For example, the program can be distributed by being stored in a storage medium such as a memory card, a CD-ROM, or a DVD-ROM, or via a communication medium such as the Internet. .

  Various embodiments and modifications of the present invention are possible without departing from the broad spirit and scope of the present invention. Further, the above-described embodiment is for explaining the present invention, and does not limit the scope of the present invention. That is, the scope of the present invention is shown not by the embodiments but by the claims. Various modifications within the scope of the claims and within the scope of the equivalent invention are considered to be within the scope of the present invention.

  The present invention is suitable for a control apparatus that rewrites an electronic file to a storage medium and that controls a file system that manages the electronic file stored in the storage medium.

10 control device 10a CPU
10b ROM
10c RAM
10d Hard disk 10e Media controller 10f USB bus 10g LAN card 10h Video card 10i LCD
10j Keyboard 10k Mouse 11 Application execution unit 20 Flash memory 100 File system control unit 110 Load measurement unit 120 Load information DB
130 Information calculation unit 131, 231, 331 Load statistical information calculation unit 131a, 231a, 331a Average value calculation unit 131b, 231b, 331b Moving average value calculation unit 132, 232, 332 Load related information calculation unit 132a, 232a, 332a Peak detection Part 132b, 232b peak width calculation part 140 statistical information DB
150 Utilization Sector Information Acquisition Unit 170 Load Prediction Unit 180 Execution Condition Determination Unit 185 GC Schedule Unit 189 GC Unit 190 Execution History DB
231c correlation coefficient calculation unit 232c peak interval calculation unit 331c non-peak average value calculation unit 333b peak height calculation unit 333c peak classification unit 333c low load transition time calculation unit PK, PK1, PK2 load peak

Claims (11)

Load measuring means for measuring a load of an arithmetic device that executes processing on the storage medium;
The curve representing the load measured by the measurement means load, and peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak width calculating means for calculating the time from when the load measured by the load measuring means exceeds the predetermined threshold until it falls below the predetermined threshold as the peak width of the load peak detected by the peak detecting means When,
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak width calculated by the peak width calculating means. A load predicting means for predicting based on the peak width of the immediately preceding peak that is the load peak that appears last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
A storage medium control device comprising: The predetermined threshold represents a magnitude of a load measured when an execution speed of processing on the storage medium decreases when the arithmetic device executes garbage collection.
The storage medium control device according to claim 1. A peak interval calculating means for calculating a peak interval from when the load measured by the load measuring means falls below the predetermined threshold until it exceeds the predetermined threshold;
The load predicting means, based on the peak width of the immediately preceding peak, when the peak interval calculated by the peak interval calculating means and the peak width of the load peak separated by the peak interval have a correlation. Predicting whether the load on the computing device in the required period after the previous peak ends is low,
The storage medium control device according to claim 1, wherein the storage medium control device is a storage medium control device. Based on the correlation between the peak width and the peak interval, and the peak width of the immediately preceding peak, the load predicting means determines the immediately following peak that is the first load peak that appears after the current measurement by the load measuring means. Estimating the peak interval, and predicting that the load of the computing device in the required period is low when the estimated peak interval of the immediately following peak is longer than the required period.
The storage medium control device according to claim 3. Load measuring means for measuring a load of an arithmetic device that executes processing on the storage medium;
The curve representing the load measured by the measurement means load, and peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak height calculating means for calculating the peak height of the load peak detected by the peak detecting means;
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated by the peak height calculating means. A load predicting means for predicting based on the peak height of the immediately preceding peak that is the load peak that appears last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
A storage medium control device comprising: Based on the distribution of peak heights calculated by the peak height calculation means, a first class to which a load peak having a low peak height belongs and a second class to which a load peak having a peak height higher than the first class belongs. And a peak classification means for classifying the immediately preceding peak,
The load predicting means predicts that the load on the arithmetic device in the required period is low when the immediately preceding peak is classified into the second classification by the peak classification means.
The storage medium control device according to claim 5. After the load peak classified into the second classification by the peak classifying unit ends, the load measured by the load measuring unit is low until the load shifts to a value lower than another threshold value lower than the predetermined threshold value. A low load transition time calculating means for calculating the load transition time;
The load predicting means, after the load measured by the load measuring means falls below the predetermined threshold, after the average time of the low load transition time calculated by the low load transition time calculating means has elapsed, the average Predicting that the load on the computing device is low in the required period after a lapse of time;
The storage medium control device according to claim 6. Computer
Load measuring means for measuring a load of an arithmetic device that executes processing on a storage medium;
The curve representing the load measured by the load measuring means, a peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak width calculating means for calculating the time from when the load measured by the load measuring means exceeds the predetermined threshold until it falls below the predetermined threshold as the peak width of the load peak detected by the peak detecting means ,
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak width calculated by the peak width calculating means. A load predicting means for predicting based on the peak width of the immediately preceding peak which is the load peak appearing last before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
A storage medium control program that functions as a storage medium. Computer
Load measuring means for measuring a load of an arithmetic device that executes processing on a storage medium;
The curve representing the load measured by the load measuring means, a peak detecting means for detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
Peak height calculating means for calculating the peak height of the load peak detected by the peak detecting means;
Whether or not the load of the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated by the peak height calculating means. A load predicting means for predicting based on the peak height of the immediately preceding peak which is the load peak last appearing before the current measurement by the load measuring means;
Garbage collection means for performing garbage collection on the storage medium when the load prediction means predicts that the load is low in the required period;
A storage medium control program that functions as a storage medium. A load measuring step for measuring a load of an arithmetic device that executes processing for the storage medium;
The curve representing the load measuring load measured in step, a peak detection step of detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
A peak width calculating step for calculating a time from when the load measured at the load measuring step exceeds the predetermined threshold to when the load falls below the predetermined threshold as a peak width of the load peak detected at the peak detecting step. When,
Whether or not the load of the arithmetic device is lower than a predetermined threshold in the required period required for the arithmetic device to execute garbage collection on the storage medium is determined by the peak width calculated in the peak width calculating step. A load prediction step for predicting based on the peak width of the immediately preceding peak that is the load peak that appeared last before the current measurement by the load measuring step;
A garbage collection step for performing garbage collection on the storage medium when the load is predicted to be low in the required period in the load prediction step;
A method for controlling a storage medium, comprising: A load measuring step for measuring a load of an arithmetic device that executes processing for the storage medium;
The curve representing the load measuring load measured in step, a peak detection step of detecting a portion exceeding the threshold value of Jo Tokoro as the load peak,
A peak height calculating step for calculating a peak height of the load peak detected in the peak detecting step;
Whether or not the load on the computing device is lower than a predetermined threshold in the required period required for the computing device to execute garbage collection on the storage medium is determined by the peak height calculated in the peak height calculating step. A load prediction step for predicting based on the peak height of the immediately preceding peak that is the load peak that appeared last before the current measurement by the load measuring step;
A garbage collection step for performing garbage collection on the storage medium when the load is predicted to be low in the required period in the load prediction step;
A method for controlling a storage medium, comprising:

Images