WO2015181923A1

WO2015181923A1 - Information processing device, control method, and control program

Info

Publication number: WO2015181923A1
Application number: PCT/JP2014/064219
Authority: WO
Inventors: 山中　英樹
Original assignee: 富士通株式会社
Priority date: 2014-05-29
Filing date: 2014-05-29
Publication date: 2015-12-03
Also published as: JP6222354B2; JPWO2015181923A1

Abstract

This information processing device is provided with: a storage unit which is provided with a plurality of partitions, and which stores the actual write frequency for each of the plurality of partitions; a calculation unit which calculates, for each of the plurality of partitions, from the actual write frequencies stored in the storage unit, a first prediction value, i.e. a prediction value of the write frequency in cases when data replacement is not performed, and a second prediction value, i.e. a prediction value of the write frequency in cases when data replacement is performed with respect to partition pairs satisfying prescribed conditions; and a replacement unit which, in cases when the highest value among the first prediction values is higher than the highest value among the second prediction values, performs data replacement with respect to the partition pairs satisfying the prescribed conditions.

Description

Information processing apparatus, control method, and control program

Related to writing control technology for storage media.

* Various data are stored in the storage medium. The number of times data is written to the storage medium varies greatly depending on the type of data.

For example, consider a case where a storage medium is used as a main storage of an embedded information device having a CPU (Central Processing Unit). A log is frequently written during execution of the microprogram in an area in which a log output by the CPU during execution of the microprogram is stored. In this case, it is estimated that writing is performed a plurality of times within one day, and hundreds to thousands of writings are performed annually. On the other hand, no writing occurs in the area where the microprogram itself is stored unless the microprogram is updated. Therefore, for example, it is estimated that writing is performed at a frequency of several times a month or several times a year. Such a deviation in the number of times of writing becomes a cause of shortening the life of a storage medium (for example, a flash memory) in which the number of times of writing is limited.

In addition, when the capacity of the storage medium is reduced for cost reduction, the ratio of the area where writing is not frequently performed (for example, an area where a microprogram or the like is stored) relatively increases. In addition, when embedded information devices have limited opportunities to connect to the network, microprogram updates are rarely performed. Due to such factors, the deviation in the number of writing becomes more remarkable.

Regarding a bias in the number of writings, a document discloses the following technology. Specifically, when the difference between the maximum erase count and the minimum erase count of the block erase count is equal to or greater than the reference count, the central controller determines that the data of the block with the minimum erase count and the data are Replace the data in the spare block that is not stored. However, when replacement is performed in this way, the number of erasures may increase after replacement.

JP-A-10-320984

Therefore, in one aspect, an object of the present invention is to equalize the number of times of writing in a storage medium in which the number of times of writing is limited.

An information processing apparatus according to the present invention includes a storage unit having a plurality of partitions and storing an actual number of times of writing for each of the plurality of partitions, and an actual writing stored in the storage unit for each of the plurality of partitions. From the number of times, a first predicted value that is a predicted value of the number of times of writing when data replacement is not executed, and a predicted value of the number of times of writing when data replacement is executed for a pair of partitions that satisfy a predetermined condition When the maximum value of the first predicted values is greater than the maximum value of the second predicted values, the calculation unit that calculates the second predicted value, and a pair of partitions that satisfy a predetermined condition And a replacement unit for performing data replacement.

In one aspect, the number of writes can be leveled on a storage medium with a limited number of writes.

FIG. 1 is a diagram illustrating a hardware configuration of the information processing apparatus. FIG. 2 is a functional block diagram of the information processing apparatus. FIG. 3 is a processing flow for explaining an outline of processing executed by the information processing apparatus. FIG. 4 is a diagram illustrating a processing flow of processing executed by the trigger setting unit of the information processing apparatus. FIG. 5 is a diagram illustrating a processing flow of processing executed by the pace setting unit of the information processing apparatus. FIG. 6 is a diagram illustrating a processing flow of processing executed by the data setting unit of the information processing apparatus. FIG. 7 is a diagram illustrating a processing flow of processing executed by the partition setting unit of the information processing apparatus. FIG. 8 is a diagram illustrating an example of data stored in the set value storage unit. FIG. 9 is a diagram showing a main processing flow. FIG. 10 is a diagram illustrating a processing flow of determination processing. FIG. 11 is a diagram illustrating a processing flow of prediction processing. FIG. 12 is a diagram illustrating an example of data stored in the partition data storage unit. FIG. 13 is a diagram illustrating a list structure according to the present embodiment. FIG. 14 is a diagram for explaining generation of the list L6. FIG. 15 is a diagram illustrating a processing flow of pace management processing. FIG. 16 is a diagram illustrating a processing flow of frequency update processing. FIG. 17 is a diagram illustrating an example of data stored in the main memory in the frequency update process. FIG. 18 is a diagram illustrating a processing flow of list update processing. FIG. 19 is a diagram showing a comparative example of the state of the flash memory. FIG. 20 is a diagram showing a specific example of the state of the flash memory.

FIG. 1 shows a hardware configuration of the information processing apparatus 1 in the present embodiment. The information processing apparatus 1 includes a flash memory 11 having a plurality of partitions, a processor 13 that is a CPU, for example, and a main memory 15 that is a dynamic random access memory (DRAM), for example. The processor 13 is connected to the flash memory 11 and the main memory 15 via a bus or the like. A program for executing the processing of the present embodiment is stored in the flash memory 11 and is read from the flash memory 11 to the main memory 15 when executed by the processor 13. The program for executing the processing of the present embodiment is included in, for example, an OS (Operating System) program or firmware.

FIG. 2 shows a functional block diagram of the information processing apparatus 1 in the present embodiment. The information processing apparatus 1 includes a trigger setting unit 101, a pace setting unit 102, a data setting unit 103, a section setting unit 104, a set value storage unit 105, a trigger detection unit 106, a determination unit 107, and a prediction unit. 108, a partition data storage unit 109, an addition unit 110, a first list update unit 111, a frequency update unit 112, a second list update unit 113, a pace management unit 114, and a replacement unit 115. The set value storage unit 105 and the partition data storage unit 109 are provided in the flash memory 11, for example.

The trigger setting unit 101 receives an input of a trigger setting value and stores it in the setting value storage unit 105. The pace setting unit 102 receives an input of a pace setting value and stores it in the setting value storage unit 105. The data setting unit 103 receives an input of a data set value and stores it in the set value storage unit 105. The partition setting unit 104 receives an input of a set value of the partition and stores it in the set value storage unit 105. The trigger detection unit 106 detects a trigger using the set value of the trigger stored in the set value storage unit 105 and calls the determination unit 107. The determination unit 107 calls the prediction unit 108. In response to this, the prediction unit 108 uses the data stored in the partition data storage unit 109 as a predicted value of the number of writings when the replacement is not executed (hereinafter referred to as a first predicted value of the number of writings). Then, a predicted value of the number of writings when the replacement is executed (hereinafter referred to as a second predicted value of the number of writings) is calculated and notified to the determination unit 107. The determination unit 107 determines whether to perform replacement using the data received from the prediction unit 108 and the data stored in the setting value storage unit 105. When performing replacement, the determination unit 107 calls the pace management unit 114. In response to this, the pace management unit 114 uses the data stored in the set value storage unit 105 to control the pace at which the replacement unit 115 executes replacement. The replacement unit 115 performs replacement in accordance with the control of the pace management unit 114. When the replacement is completed, the pace management unit 114 calls the frequency update unit 112. In response to this, the frequency update unit 112 changes the writing frequency for the section that has been replaced. Further, the frequency update unit 112 calls the second list update unit 113. In response to this, the second list updating unit 113 sets a pointer in the ascending order of the writing frequency stored in the partition data storage unit 109.

When the writing to the flash memory 11 is performed, the addition unit 110 updates the number of times of writing stored in the partition data storage unit 109. In addition, the addition unit 110 calls the first list update unit 111. In response to this, the first list updating unit 111 sets a pointer in the ascending order list of the number of writings stored in the partition data storage unit 109. In addition, the addition unit 110 calls the frequency update unit 112. In response to this, the frequency updating unit 112 updates the writing frequency for the partition in which writing has been performed. Further, the frequency update unit 112 calls the second list update unit 113. In response to this, the second list updating unit 113 sets a pointer in the ascending order of the writing frequency stored in the partition data storage unit 109.

Next, processing executed by the information processing apparatus 1 will be described with reference to FIGS. 3 to 20. First, an outline of processing executed by the information processing apparatus 1 will be described with reference to FIG.

First, the adding unit 110 detects that writing to the flash memory 11 has been performed (FIG. 3: step S1001). Then, the adding unit 110 updates the number of times of writing stored in the partition data storage unit 109 for the partition in which writing has been performed. Also, the frequency update unit 112 updates the writing frequency stored in the partition data storage unit 109 for the partition in which writing has been performed (step S1003).

The trigger detection unit 106 determines whether a trigger is detected using the trigger setting value stored in the setting value storage unit 105 (step S1005). When the trigger is not detected (step S1005: No route), it is not the timing for calculating the predicted value, so the process returns to step S1001.

When the trigger is detected (step S1005: Yes route), the prediction unit 108 uses the data stored in the partition data storage unit 109 to calculate the first predicted value of the number of writes (that is, when no replacement is performed). A predicted value of the number of times of writing) is calculated for each partition (step S1007).

The prediction unit 108 identifies the partition where the replacement is performed using the data stored in the partition data storage unit 109 (step S1009). Then, the prediction unit 108 calculates a second predicted value of the number of writes (that is, a predicted value of the number of writes when replacement is performed) for each partition (step S1011).

The determination unit 107 determines whether to perform replacement using the first prediction value and the second prediction value calculated by the prediction unit 108 (step S1013). When the replacement is not executed (step S1013: No route), the process returns to step S1001.

When executing replacement (step S1013: Yes route), the replacement unit 115 performs replacement at the pace set by the pace management unit 114 (step S1015). Then, the frequency updating unit 112 replaces the writing frequency stored in the partition data storage unit 109 with respect to the replaced partition (step S1017). Then, the process ends.

If the processing as described above is executed, the replacement can be performed after confirming that the replacement is effective, so that the number of writes can be effectively leveled.

Next, the processing executed by the information processing apparatus 1 will be described in more detail with reference to FIGS. First, a trigger setting process executed by the trigger setting unit 101 of the information processing apparatus 1 will be described with reference to FIG.

The trigger setting unit 101 determines whether an input of a set value of a trigger (in this embodiment, a trigger for the determination unit 107 to start processing) has been detected (FIG. 4: step S1). The set value of the trigger is input by, for example, an administrator. When the input of the trigger setting value is not detected (step S1: No route), the process returns to step S1.

When the input of the trigger setting value is detected (step S1: Yes route), the trigger setting unit 101 determines whether the trigger setting value includes the time setting value (step S3). When the trigger setting value does not include the time setting value (step S3: No route), the process proceeds to step S7.

When the trigger setting value includes the time setting value (step S3: Yes route), the trigger setting unit 101 stores the time setting value in the setting value storage unit 105 (step S5).

The trigger setting unit 101 determines whether the trigger setting value includes a setting value for an event (for example, a decrease in CPU load) (step S7). If the trigger setting value does not include the event setting value (step S7: No route), the process returns to step S1. On the other hand, when the trigger setting value includes the event setting value (step S7: Yes route), the trigger setting unit 101 stores the event setting value in the setting value storage unit 105 (step S9). Then, the process returns to step S1.

In this way, the trigger setting value can be used in the processing described later.

Next, the pace setting process executed by the pace setting unit 102 of the information processing apparatus 1 will be described with reference to FIG. First, the pace setting unit 102 determines whether or not an input of a set value of a pace (in this embodiment, the amount of data moved by one replacement) is detected (FIG. 5: step S11). The set value of the pace is input by, for example, an administrator. When the input of the set value of pace is not detected (step S11: No route), the process returns to step S11. On the other hand, when the input of the pace setting value is detected (step S11: Yes route), the pace setting unit 102 stores the pace setting value in the setting value storage unit 105 (step S13). Then, the process returns to step S11.

In this way, the pace setting value can be used in the processing described later.

Next, a data setting process executed by the data setting unit 103 of the information processing apparatus will be described with reference to FIG. First, the data setting unit 103 determines whether an input of a data setting value (in this embodiment, a setting value indicating whether or not replacement can be performed on data) is detected (FIG. 6: Step S21). The data set value is input by, for example, an administrator. When the input of the data set value is not detected (step S21: No route), the process returns to step S21. On the other hand, when the input of the data set value is detected (step S21: Yes route), the data setting unit 103 stores the data set value in the set value storage unit 105 (step S23). Then, the process returns to step S21.

By doing so, the data set value can be used in the processing described later.

Next, the partition setting process executed by the partition setting unit 104 of the information processing apparatus will be described with reference to FIG. First, has the partition setting unit 104 detected an input of a partition setting value (in this embodiment, a setting value indicating whether or not replacement of data stored in the partition can be performed)? Judgment is made (FIG. 7: step S31). The set value of the section is input by, for example, an administrator. When the input of the set value of the section is not detected (step S31: No route), the process returns to step S31. On the other hand, when the input of the set value of the partition is detected (step S31: Yes route), the partition setting unit 104 stores the set value of the partition in the set value storage unit 105 (step S33). Then, the process returns to step S31.

By doing so, it becomes possible to use the set value of the section in the processing described later.

FIG. 8 shows an example of data stored in the set value storage unit 105. In the example of FIG. 8, a time setting value, an event setting value, a pace setting value, a data setting value, and a section setting value are stored. When the value of the time flag in the time setting value is 1, the time setting value is valid. When the value of the CPU load flag in the set value of the event is 1, the determination unit 107 starts processing when the value of the CPU load is equal to or less than a predetermined reference value. When the value of the IO (Input / Output) load flag in the set value of the event is 1, the determination unit 107 starts processing when the value of the IO load becomes equal to or less than a predetermined reference value. When the value of the replacement flag in the data setting value is 1, replacement can be performed on the data. When the value of the replacement flag in the section setting value is 1, replacement can be performed on the data stored in the section.

Next, a process for exchanging data will be described with reference to FIGS. 9 to 20. First, the trigger detection unit 106 acquires the current time from the OS (FIG. 9: Step S41).

The trigger detection unit 106 determines whether or not the current time acquired in step S41 is later than the set time represented by the set value of the time stored in the set value storage unit 105 (step S43). If the current time is after the set time (step S43: Yes route), the process proceeds to step S55.

When the current time is not later than the set time (step S43: Yes route), the trigger detection unit 106 determines whether the value of the CPU load flag stored in the set value storage unit 105 is 1 (step S45). . When the value of the CPU load flag is not 1 (step S45: No route), the process proceeds to step S51.

When the value of the CPU load flag is 1 (step S45: Yes route), the trigger detection unit 106 acquires the value of the CPU load from the OS (step S47). The trigger detection unit 106 determines whether the CPU load value acquired in step S47 is equal to or less than a predetermined reference value (step S49). When the CPU load value is equal to or less than the predetermined reference value (step S49: Yes route), the process proceeds to step S55.

When the value of the CPU load is larger than the predetermined reference value (step S49: No route), the trigger detection unit 106 indicates that the trigger detection unit 106 indicates that the IO load flag value stored in the set value storage unit 105 is 1. It is determined whether or not there is (step S50). If the value of the IO load flag is not 1 (step S50: No route), the process returns to step S41.

When the value of the IO load flag is 1 (step S50: Yes route), the trigger detection unit 106 acquires the value of the IO load from the OS (step S51).

The trigger detection unit 106 determines whether the IO load value acquired in step S51 is equal to or less than a predetermined reference value (step S53). If the IO load value is not less than or equal to the predetermined reference value (step S53: No route), the process returns to step S41.

When the IO load value is equal to or smaller than the predetermined reference value (step S53: Yes route), the trigger detection unit 106 calls the determination unit 107. And the determination part 107 performs a determination process (step S55). The determination process will be described with reference to FIG.

First, the determination unit 107 specifies “no replacement” as an argument and calls the prediction unit 108. In response to this, the prediction unit 108 executes a prediction process (FIG. 10: step S61). The prediction process will be described with reference to FIGS. 11 to 13.

First, the prediction unit 108 determines whether “replacement is present” is designated as an argument (FIG. 11: step S81). When “with replacement” is not specified as an argument (step S81: No route), the prediction unit 108 calculates a first predicted value of the number of writes after a predetermined time has elapsed for each partition in the flash memory 11 ( In step S83, the data is stored in the partition data storage unit 109. In step S83, for example, the first predicted value of the number of write times = the number of write times + (write frequency * predetermined time) is calculated.

FIG. 12 shows an example of data stored in the partition data storage unit 109. In the example of FIG. 12, the number of partitions is n (n is a natural number of 2 or more), and partition data is stored for each of the n partitions. The partition data includes a partition identifier, a write count, a pointer in the list L1 that is a list of the write count, a first predicted value of the write count, and a list L2 that is a list of the first predicted value of the write count. , The second predicted value of the write count, the list L3 that is a list of the second predicted value of the write count, the write frequency, the pointer in the list L4 that is the list of the write frequency, and the write when the replacement is executed It includes a pointer in a list L5 that is a list of frequencies (hereinafter referred to as post-replacement writing frequency) and a pointer in a list L6 that is a list of partitions on which replacement is performed. The partition data storage unit 109 stores the pointers of the anchors in the lists L1 to L6.

FIG. 13 shows the structure of the list in the present embodiment. A pointer that points to the previous element (hereinafter referred to as the previous element pointer) and a pointer that points to the next element (hereinafter referred to as the next element pointer) are added to each element. The previous element pointer added to the anchor is “NULL”, and the next element pointer points to the first element (element [x] in the example of FIG. 13). The lists L1 to L5 in the present embodiment are lists arranged so that element values are in ascending order. Therefore, the value of the element indicated by the previous element pointer of the anchor is the smallest and the value of the last element is the largest.

Returning to the description of FIG. 11, the prediction unit 108 sets the pointer in the list L2 (that is, the list of the first predicted value of the number of writes) in ascending order of the first predicted value of the number of writes according to the processing result of step S83. (Step S85). Then, the process returns to the calling process. Note that step S87 and subsequent steps in the determination process (FIG. 11) will be described later.

Returning to the description of FIG. 10, the determination unit 107 specifies the maximum value M1 among the first predicted values of the number of writes from the list L2 (that is, the list of the first predicted values of the number of times of writing) (Step S63). ).

The determination unit 107 generates a list L7 of partitions that are not replacement targets by using the data setting values and the partition setting values stored in the setting value storage unit 105 (step S65). The list L7 includes identifiers of sections that are not replacement targets. For example, when the replacement flag for data A is not 1, the identifier of the partition in which data A is stored is included in the list L7. For example, when the replacement flag for the section Sa is not 1, the identifier of the section Sa is included in the list L7.

The determination unit 107 calls the prediction unit 108 by designating “with replacement” and “list L7” as arguments. In response to this, the prediction unit 108 executes a prediction process (step S67). The prediction process will be described with reference to FIGS. 11 and 14.

First, the prediction unit 108 determines whether “replacement is present” is designated as an argument (FIG. 11: step S81). When “with replacement” is specified as the argument (step S81: Yes route), the prediction unit 108 uses the number of writes stored in the partition data storage unit 109 to execute the list L6 (that is, replacement is executed). A list of sections to be generated) (step S87).

The generation of the list L6 will be described with reference to FIG. In step S87, first, a difference in the number of writing times is calculated for a pair of partitions selected from the list L1 (that is, a list of the number of writing times), and it is determined whether the calculated difference is equal to or greater than a predetermined value. Here, the pair of sections is selected as shown in FIG. That is, the pair including the partition with the largest number of writes and the partition with the smallest number of writes is selected, then the partition with the second largest number of writes and the partition with the second smallest number of writes are selected, and the next In addition, the partition with the third largest number of writes and the partition with the third smallest number of writes are selected. When the difference calculated for the selected pair is equal to or greater than a predetermined value, and none of the sections included in the pair is included in the list L7 (that is, the list of sections not to be replaced) , The partitions included in the pair are added to the list L6.

Returning to the description of FIG. 11, for each partition, the prediction unit 108 copies the pointer in the list L4 stored in the partition data storage unit 109 to the pointer in the list L5 (step S89). Further, the prediction unit 108 copies the writing frequency stored in the partition data storage unit 109 to the writing frequency after replacement for each partition (step S91).

The prediction unit 108 identifies a pair of partitions using the list L6 (that is, the list of partitions to be replaced) and the list L5 (that is, the list of write frequencies after replacement), and writes after replacement for the specified pair. Frequency replacement is executed (step S93).

The prediction unit 108 calculates a second predicted value of the number of times of writing after a predetermined time has elapsed for each partition (step S95) and stores it in the partition data storage unit 109. In step S95, for example, the second predicted value of the number of write times = the number of write times + (write frequency after replacement * predetermined time).

The prediction unit 108 sets the pointers in the list L3 (that is, the list of second predicted values of the number of writes) according to the processing result of step S95 so that the second predicted value of the number of writes is in ascending order (step S97). Then, the process returns to the calling process.

As described above, the first predicted value and the second predicted value of the number of writes can be easily calculated by using the write frequency.

Returning to the description of FIG. 10, the determination unit 107 identifies the maximum value M2 of the second predicted values of the number of writes from the list L3 (that is, the list of second predicted values of the write frequency) (step S69). ).

The determination unit 107 determines whether M1> M2 is satisfied (step S71). When M1> M2 is not satisfied (step S71: No route), the process returns to the caller process.

When M1> M2 is satisfied (step S73: Yes route), the determination unit 107 calls the pace management unit 114 by designating “list L6” as an argument. In response to this, the pace management unit 114 executes pace management processing (step S73). The pace management process will be described with reference to FIG.

First, the pace management unit 114 determines whether it is the first execution (FIG. 15: step S101). In the case of the first execution (step S101: Yes route), the pace management unit 114 displays the pace setting values stored in the setting value storage unit 105 and the list L6 (that is, a list of sections on which replacement is executed). The number of executions, the amount of data for one execution, and the execution interval are determined (step S103) and stored in the main memory 15. For example, it is assumed that the number of sections included in the list L6 is 100, the capacity of one section is 1 megabyte, and the set value of the pace is 10 megabytes / second. In this case, for example, if the number of executions is 10, the data amount for one execution is 10 megabytes and the execution interval is 1 second. For example, if the number of executions is 20, the amount of data for one execution is 5 megabytes, and the execution interval is 0.5 seconds.

The pace management unit 114 stores data indicating that replacement is being performed in the main memory 15 (step S105).

On the other hand, when it is not the first execution (step S101: No route), the pace management unit 114 acquires the current time from the OS (step S107).

The pace management unit 114 determines whether the time corresponding to the execution interval has elapsed since the previous execution time (step S109). In step S109, it is determined whether the time at which the execution interval has elapsed from the previous execution time is later than the current time. The previous execution time is stored in the main memory 15 in step S115 described later. The execution interval is determined in step S103.

If the time corresponding to the execution interval has not elapsed since the previous execution time (step S109: No route), the processing returns to step S101. When the time corresponding to the execution interval has elapsed from the previous execution time (step S109: Yes route), the pace management unit 114 calls the replacement unit 115. In response to this, the replacement unit 115 performs data replacement by the amount of data determined in step S103 (step S111). In step S111, for example, the data in the section to be replaced is moved to the main memory 15, and then moved to the replacement destination section.

The pace manager 114 determines whether the actual number of executions has reached the number of executions determined in step S103 (step S113). If not reached (step S113: No route), the pace management unit 114 stores the time when the replacement is executed in the main memory 15 as the previous execution time (step S115). Then, the process returns to step S101.

When the actual number of executions reaches the number of executions determined in step S103 (step S113: Yes route), the pace management unit 114 indicates that the replacement stored in the main memory 15 in step S105 is being executed. The indicated data is deleted (step S117).

The pace management unit 114 calls the frequency update unit 112 by designating “replacement” as an argument. In response to this, the frequency update unit 112 executes a frequency update process (step S119). The frequency update process will be described with reference to FIG.

First, the frequency update unit 112 determines whether “replacement” is designated as an argument (step S121). When “replacement” is designated as an argument (step S121: Yes route), the call is a call from the pace management unit 114. Therefore, the frequency update unit 112 performs replacement of the writing frequency stored in the partition data storage unit 109 for the partition that has been replaced (step S123). In step S123, the same replacement as in step S93 is performed.

The frequency update unit 112 calls the second list update unit 113 with “replacement” specified as an argument. In response to this, the second list update unit 113 executes a list update process (step S125). The list update process will be described later.

On the other hand, when it is determined in step S121 that “replacement” is not specified as an argument (step S121: No route), the call is a call from the addition unit 110. Therefore, the frequency update unit 112 acquires the current time from the OS (step S127).

The frequency update unit 112 calculates the update interval by update interval = current time−previous time (step S129), and updates the update interval stored in the main memory 15. FIG. 17 shows an example of data stored in the main memory 15. In the example of FIG. 17, the update interval, the old measurement time, the measurement time, and the previous time are stored. The old measurement time is the time from the start time to the previous time. The measurement time is the time from the start time to the current time. The previous time is the time when the previous update was performed.

The frequency update unit 112 sets the old measurement time (step S131). In step S131, the old measurement time is updated with the measurement time stored in the main memory 15. Further, the frequency updating unit 112 calculates the measurement time by measurement time = measurement time + update interval (step S133), and updates the measurement time stored in the main memory 15.

The frequency update unit 112 calculates write frequency = ((write frequency * old measurement time) +1) / measurement time for the partition in which writing has been performed, and updates the write frequency stored in the partition data storage unit 109. (Step S135).

The frequency update unit 112 sets the previous time (step S137). In step S137, the previous time stored in the main memory 15 is updated with the current time.

The frequency update unit 112 calls the second list update unit 113 by designating “update” as an argument. In response to this, the second list update unit 113 executes list update processing (step S139). The list update process will be described with reference to FIG.

First, the second list updating unit 113 determines whether “replacement” is designated as an argument (FIG. 18: step S141). When “replacement” is specified as an argument (step S141: Yes route), the second list update unit 113 sets a pointer in the list L4 (that is, a list of write frequencies) based on the write frequency after replacement. (Step S145). Here, the setting is performed so that the writing frequency after replacement is in ascending order.

On the other hand, when “replacement” is not specified as an argument (step S141: No route), the second list updating unit 113 sets a pointer in the list L4 based on the updated writing frequency (step S143). Here, the setting is performed so that the updated writing frequency is in ascending order. Then, the process returns to step S41.

If the processing as described above is executed, it is ensured that the replacement is effective by comparing the predicted values, so that the number of writing can be effectively leveled. Further, since the pair of partitions is selected as shown in FIG. 14, the pair that does not contribute to the leveling of the number of times of writing is eliminated, and the leveling can be performed effectively. In addition, it is possible to calculate a predicted value that reflects an increase in the number of writes due to replacement.

19 and FIG. 20, a specific example of leveling the number of writes will be shown. Here, the flash memory has eight partitions (partitions s0 to s7), and the upper limit of the number of times of writing in the partition is 100. Data a consumes 4 sections, and the update frequency of data a is lower than that of data b. Data b consumes 3 sections, and the update frequency of data b is higher than data a. Time zones t0 to t3 are times having a certain length.

First, the state of the flash memory when the present embodiment is not used will be described with reference to FIG. Here, four sections are assigned to the data b, but one section out of the four sections is made empty, and the empty section is circulated. In time zone t0, there is no write request and no write is performed. In the time zone t1, there are 53 write requests for data b. Since the section s7 is in an empty state in the time zone t1, the number of writes in the sections s4 to s6 is incremented by 53. In the time zone t2, there are 37 write requests for data b. Since the partition s6 is in an empty state in the time zone t2, the number of times of writing in the partitions s4, s5, and s7 is incremented by 37. There are 8 write requests for data b in time zone t3. Since the partition s5 is in an empty state in the time zone t3, the number of writes in the partitions s4, s6, and s7 is incremented by 8, and the number of writes in the partition s4 reaches the upper limit at this point. The number of write requests processed is 98. However, almost no writing is performed in the partitions s0 to s3 at this point, and it cannot be said that the flash memory is being used efficiently.

Next, the state of the flash memory 11 in the present embodiment will be described with reference to FIG. Here, any of the sections s0 to s7 is set to an empty state, and the empty sections are circulated. In time zone t0, there is no write request and no write is performed. In the time zone t1, there are 53 write requests for data b. Since the section s7 is in an empty state in the time zone t1, the number of writes in the sections s4 to s6 is incremented by 53.

In time zone t2, replacement is performed and there is a write request of 37 for data b. Since the data b is stored in the partitions s0, s1, and s3 in the time zone t2, the number of times of writing in the partitions s0, s1, and s3 is incremented by 37. In addition, since the data is replaced, the number of times of writing in all the partitions is incremented by one.

In time zone t3, replacement is performed and there is an 8 write request for data b. Since the data b is stored in the partitions s4, s6, and s7 in the time zone t3, the number of writes in the partitions s4, s6, and s7 is incremented by 8. In addition, since the data is replaced, the number of times of writing in all the partitions is incremented by one.

In time zone t4, replacement is performed, and there are 43 write requests for data b. Since the data b is stored in the partitions s1, s3, and s4 in the time zone t4, the number of writes in the partitions s1, s3, and s4 is incremented by 43. In addition, since the data is replaced, the number of times of writing in all the partitions is incremented by one.

In time zone t5, replacement is performed, and there are 32 write requests for data b. Since the data b is stored in the partitions s4, s6, and s7 in the time zone t5, the number of writes in the partitions s4, s6, and s7 is incremented by 32. In addition, since the data is replaced, the number of times of writing in all the partitions is incremented by one. At this time, the number of times of writing in the section s6 reaches the upper limit. The number of write requests processed is 173. Compared with the example of FIG. 19, a write request of 173 / 98≈1.8 times can be processed.

Although one embodiment of the present invention has been described above, the present invention is not limited to this. For example, the functional block configuration of the information processing apparatus 1 described above may not match the actual program module configuration.

Further, the configuration of each data holding form described above is an example, and the configuration as described above does not have to be. Further, regarding the processing flow, as long as the processing result does not change, the processing order may be changed or a plurality of steps may be executed in parallel.

It should be noted that a pair may be selected more flexibly by dynamically changing the predetermined value used in step S87.

The format of the lists L1 to L6 may be an array format or a red-black tree format.

Note that, in the lists L1 to L5, element values may be arranged in descending order instead of ascending order.

Note that the information processing apparatus 1 may be a personal computer or a portable terminal (for example, a smartphone) equipped with an SSD (Solid State Drive).

The embodiments of the present invention described above are summarized as follows.

The information processing apparatus according to the first aspect of the present embodiment includes (A) a storage unit that has a plurality of sections and stores the actual number of times of writing for each of the plurality of sections, and (B) a plurality of sections For each of the data, the first predicted value, which is the predicted value of the number of writes in the case where data replacement is not performed, and the pair of partitions satisfying the predetermined condition, from the actual number of writes stored in the storage unit. A calculation unit that calculates a second predicted value that is a predicted value of the number of times of writing when the replacement is performed; and (C) a maximum value among the first predicted values is a maximum value among the second predicted values. When the value is larger than the value, a replacement unit that performs data replacement for a pair of partitions satisfying a predetermined condition is provided.

In this way, since the replacement is guaranteed to be effective, the number of writes can be effectively leveled.

Further, the calculation unit described above uses (b1) the actual number of times of writing in the first section of the pair of sections and the first writing frequency calculated from the actual number of writes in the first section. The first predicted value of the first partition is calculated, and (b2) the second calculated from the actual number of writes of the first partition and the actual number of writes of the second partition in the pair of partitions. May be used to calculate the second predicted value of the first partition. In this way, the predicted value can be easily obtained.

In addition, the predetermined condition described above is that a difference between the first predicted value of the first partition of the pair of partitions and the first predicted value of the second partition of the pair of partitions is equal to or greater than a predetermined value. It may include a condition that In this way, it is possible to perform replacement for a pair of sections that contribute to leveling of the number of writes.

Also, the first partition and the second partition may be selected so that the difference between the first predicted value of the first partition and the first predicted value of the second partition is maximized. In this way, the most effective pair can be selected.

Further, the information processing apparatus may further include (D) a determination unit that determines a partition that is not a replacement target based on the second setting value acquired in advance. The predetermined condition described above may further include a condition that the first section and the second section are sections other than the sections that are not to be replaced. In this way, it is possible to prevent replacement of data or sections that should not be replaced.

The information processing apparatus may further include (E) a control unit that causes the calculation unit and the replacement unit to start processing when a predetermined time has elapsed or when a predetermined event has occurred. In this way, processing can be performed at an appropriate timing, so that waste of resources of the information processing apparatus can be suppressed.

In addition, the information processing apparatus calculates (F) the interval for executing the replacement, the number of times of replacement, and the amount of data moved by one replacement based on the first setting value acquired in advance, and based on the calculation result You may further have a replacement management part which makes a replacement part perform replacement. If it does in this way, it will come to be able to replace appropriately according to the number or size of a division, for example.

Further, the storage unit described above may be (a1) a storage medium having an upper limit on the number of times of writing.

The control method according to the second aspect of the present embodiment includes (G) a storage unit for each of a plurality of partitions in a storage unit that has a plurality of partitions and stores the actual number of times of writing for each of the plurality of partitions. When data exchange is executed for a pair of partitions that satisfy a predetermined condition and a first predicted value that is a predicted value of the number of write times when data exchange is not executed from the actual number of writes stored in A second predicted value that is a predicted value of the number of times of writing in (H), and (H) when the maximum value among the first predicted values is greater than the maximum value among the second predicted values, a predetermined condition Including a process of exchanging data for a pair of partitions satisfying the condition.

It is possible to create a program for causing a computer or processor to perform the processing according to the above method, and the program is a computer-readable storage such as a flexible disk, a CD-ROM, a magneto-optical disk, a semiconductor memory, and a hard disk. Stored in a medium or storage device. The intermediate processing result is temporarily stored in a storage device such as a main memory.

Claims

A storage unit having a plurality of partitions and storing the actual number of times of writing for each of the plurality of partitions;
For each of the plurality of partitions, from the actual number of writes stored in the storage unit, a partition that satisfies a first predicted value that is a predicted value of the number of writes when data replacement is not performed and a predetermined condition A calculation unit that calculates a second predicted value that is a predicted value of the number of times of writing when the data replacement is performed on the pair of
A replacement unit that performs replacement of the data for a pair of partitions satisfying the predetermined condition when a maximum value of the first predicted values is greater than a maximum value of the second predicted values; ,
An information processing apparatus.
The calculation unit includes:
Using the actual number of writes in the first partition of the pair of partitions and the first write frequency calculated from the actual number of writes in the first partition, the first partition Calculating a first predicted value;
Using the actual number of writes of the first partition and the second write frequency calculated from the actual number of writes of the second partition of the pair of partitions, the first partition of the first partition The information processing apparatus according to claim 1, wherein the second predicted value is calculated.
The predetermined condition is that a difference between the first predicted value of the first section of the pair of sections and the first predicted value of the second section of the pair of sections is a predetermined value or more. The information processing apparatus according to claim 1, further comprising:
The first partition and the second partition are selected such that the difference between the first predicted value of the first partition and the first predicted value of the second partition is maximized. The information processing apparatus according to claim 3.
A determination unit that determines a section that is not the target of replacement based on the second setting value acquired in advance;
The information processing apparatus according to claim 3, wherein the predetermined condition further includes a condition that the first section and the second section are sections other than the section that is not the replacement target.
The information processing apparatus according to claim 1, further comprising: a control unit that causes the calculation unit and the replacement unit to start processing when a predetermined time has elapsed or when a predetermined event has occurred.
Based on the first set value acquired in advance, the interval for performing the replacement, the number of times of replacement, and the amount of data to be moved by one replacement are calculated, and the replacement is performed in the replacement unit based on the calculation result. The information processing apparatus according to any one of claims 1 to 6, further comprising a replacement management unit.
The information processing apparatus according to claim 1, wherein the storage unit is a storage medium having an upper limit on the number of times of writing.
For each of the plurality of partitions in the storage unit that has a plurality of partitions and stores the actual number of writes for each of the plurality of partitions, the data is replaced from the actual number of writes stored in the storage unit. A first predicted value that is a predicted value of the number of times of writing when the data is not executed, and a second predicted value that is a predicted value of the number of times of writing when the data replacement is executed for a pair of partitions satisfying a predetermined condition And
When the maximum value of the first predicted values is greater than the maximum value of the second predicted values, the computer executes a process of exchanging the data for a pair of partitions satisfying the predetermined condition A control program to be executed.
For each of the plurality of partitions in the storage unit that has a plurality of partitions and stores the actual number of writes for each of the plurality of partitions, the data is replaced from the actual number of writes stored in the storage unit. A first predicted value that is a predicted value of the number of times of writing when the data is not executed, and a second predicted value that is a predicted value of the number of times of writing when the data replacement is executed for a pair of partitions satisfying a predetermined condition And
When the maximum value of the first predicted values is greater than the maximum value of the second predicted values, the computer executes a process of exchanging the data for a pair of partitions satisfying the predetermined condition Control method executed by.