JP5412902B2 - Storage system - Google Patents

Description

  The present invention relates to a storage system, and particularly relates to a storage system equipped with a plurality of storage devices.

  In recent years, a storage system having a large number of disk devices has been developed in order to improve data storage capacity and reliability. For example, there are NAS (Network Attached Storage) devices and grid storage systems.

Here, it is assumed that files read by hundreds of clients in a specific process at startup are stored on the storage system. Then, since this file is accessed when hundreds of clients are activated at the same time, the load is concentrated on the file and it becomes a hot spot. However, the time at which such hot spots appear is short, and the daily average disk busy rate occupies few resources of the entire system. In addition, in a storage system, there is a general problem that performance degradation tends to occur when the capacity approaches to full. In order to cope with this, and to withstand sudden increase in capacity, it is usually operated with a margin of 70% or less. In addition, ILM (Information Lifecycle Management: Information Lifecycle
According to the concept of “Management”, it is said that most data is data with low access frequency (for example, once a day to less than once every several months).

  However, as described above, even in a case where the operation rate (load / capacity) of the entire storage system is low, in a storage system having a large number of disk devices, it is necessary to continuously drive the large number of disk devices. . As a result, there is a problem that the power consumption of the storage system is difficult to decrease.

  On the other hand, Patent Document 1 discloses a technique aimed at saving energy in a storage system such as a NAS device. Specifically, the load of each virtual storage that makes up the storage system is measured, the processing of the virtual storage with a high load is divided by other storage processing units, and the processing of the virtual storage with a low load is processed with other storage processing Integrate in department Thereby, the power supply to the storage processing unit that controls the virtual storage system is controlled.

JP 2003-296153 A

  However, in the technique disclosed in Patent Document 1 described above, the load is measured in units of virtual storage, and only the power supply to the storage processing unit that controls the virtual storage is controlled. The power supply to the device is not properly controlled. Therefore, there is a problem that the power consumption of the disk device cannot be reduced and the power consumption of the entire system is still difficult to decrease.

  Therefore, an object of the present invention is to provide a storage system that can solve the above-described problem that it is difficult to reduce power consumption.

In order to achieve such an object, a storage system according to one aspect of the present invention provides:
For each disk device, measure the operating status per unit time for the disk device, determine that the disk device that was operating is an active disk, determine that the disk device that is not operating is a non-operating disk, and A performance monitoring unit for measuring the load on the disk device and storing the measurement result;
A data rearrangement unit that stores data stored in the disk device determined to be an active disk based on a measurement result by the performance monitoring unit in another disk device having a higher load than the load in the disk device. When,
Is provided.

Moreover, the program which is the other form of this invention is:
On the computer,
For each disk device, measure the operating status per unit time for the disk device, determine that the disk device that was operating is an active disk, determine that the disk device that is not operating is a non-operating disk, and A performance monitoring unit for measuring the load on the disk device and storing the measurement result;
A data rearrangement unit that stores data stored in the disk device determined to be an active disk based on a measurement result by the performance monitoring unit in another disk device having a higher load than the load in the disk device. When,
It is a program for realizing.

In addition, a data storage method according to another aspect of the present invention includes:
For each disk device, measure the operating status per unit time for the disk device, determine that the disk device that was operating is an active disk, determine that the disk device that is not operating is a non-operating disk, and Measure the load on the above disk device, store the measurement results, perform performance monitoring,
Based on the measurement result, data stored in the disk device determined to be an active disk is stored in another disk device having a higher load than the load in the disk device, and data is rearranged. ,
The structure is taken.

  By configuring as described above, the present invention can achieve further power saving of the entire storage system.

1 is a functional block diagram illustrating a configuration of a storage system in Embodiment 1. FIG. FIG. 2 is a diagram illustrating an example of measurement of a load on a disk device in the storage system disclosed in FIG. 1. It is a figure which shows an example of the data memorize | stored in the rearrangement candidate list memory | storage part disclosed in FIG. It is a figure which shows an example of the data set to the storage system disclosed in FIG. 3 is a flowchart showing the operation of the storage system disclosed in FIG. 1. 3 is a flowchart showing the operation of the storage system disclosed in FIG. 1. It is a figure which shows the mode at the time of the data rearrangement in the storage system disclosed in FIG. 3 is a flowchart showing the operation of the storage system disclosed in FIG. 1. It is a figure which shows the mode at the time of the data rearrangement in the storage system disclosed in FIG. 3 is a flowchart showing the operation of the storage system disclosed in FIG. 1. It is a figure which shows the mode at the time of the data rearrangement in the storage system disclosed in FIG. 1 is a functional block diagram illustrating a configuration of a storage system in Embodiment 1. FIG.

<Embodiment 1>
A first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a functional block diagram showing the configuration of the storage system in this embodiment. FIG. 2 is a diagram illustrating an example of the load measurement of the disk device. 3 to 4 are diagrams illustrating an example of data stored in the storage system. 5 to 11 are diagrams showing the operation of the storage system.

[Constitution]
As shown in FIG. 1, the storage system 10 in the present embodiment includes a plurality of disk devices 6, 7,..., N, and is realized by, for example, a NAS device or a grid storage system. In FIG. 1, the storage system is illustrated as being integrated with the disk device 6 or the like, but the storage system in the present invention is separated from the disk device 6 or the like, and the disk device 6 or the like. It may be configured by a computer that performs data recording / reproduction control for the above.

  The storage system 10 includes a file management unit 1, a performance monitoring unit 2, a rearrangement candidate list storage unit 3, and a file rearrangement control unit 4 which are constructed by incorporating a program into the equipped arithmetic device. And a file group selection control unit 5. Note that a storage device such as a flash memory is also provided for storing data that needs to be stored along with the processing by each of the units 1 to 5. Hereinafter, each configuration will be described in detail.

  The file management unit 1 manages the storage location of the file data stored in the disk device 6 or the like, that is, in which partition of which disk. Note that the storage capacity of the disk device 6 or the like is divided into a plurality of sections as indicated by reference numerals 61, 71, n1, etc. in FIG. The file data 62, 72, n2, etc. are stored across the sections or between the sections. For example, the storage capacity of one disk device is divided into about 1000 partitions.

  Further, the performance monitoring unit 2 identifies each section described above by a unique section number. Then, the performance monitoring unit 2 observes an I / O (Input / Output) frequency distribution for each partition and a busy rate of the entire disk in a certain period (for example, 10 minutes). Specifically, the performance monitoring unit 2 calculates, as the disk busy rate, the operating rate per unit time of the corresponding disk device, that is, the ratio of the operating time per fixed period. Then, the disk busy rate is not “0”, that is, a disk device operating within a certain period (simply time) is determined as an active disk, and the disk busy rate is “0”, that is, a certain period (unit time). The disk device that has not been operated is determined as a non-working disk. Further, the disk device determined as a non-working disk is stopped by the performance monitoring unit 2 or other functions provided in the storage system 10. That is, power supply to the disk device that is a non-working disk is stopped. A non-working disk does not mean that it is unused. There is a possibility that data is not stored but only operating per unit time described above.

  As described above, the disk device to be measured is not limited to a single disk device, and may be a plurality of disk devices. In other words, the disk busy rate may be measured with a plurality of grouped disk devices as one disk device.

  Further, the performance monitoring unit 2 calculates the busy rate for each partition from the disk busy rate measured as described above and the I / O frequency distribution for each partition, and uses this value as the load of each partition. To do. At this time, based on the read count and the write count for each partition, the read busy rate at the time of read processing and the write busy rate at the time of write processing are calculated for each partition.

  Here, an example of the busy rate calculation in each section of the predetermined disk device 6 will be described with reference to FIG. First, as shown in FIG. 2A, it is assumed that the disk device 6 is divided into sections # 1 to # 7. The busy rate of the entire disk device 6 at this time is measured to be 80%, and the number of reads and the number of writes for each of the partitions # 1 to # 7 are the numbers shown in FIG. To do. That is, when the disk device 6 is divided into seven partitions # 1 to # 7, the I / O frequency in each partition is 4 from the partition # 1 to Read as shown in the upper diagram of FIG. 1 time, 8 times, 0 times, 0 times, 0 times, 4 times, and as shown in the lower diagram of FIG. 2B, Write from the partition # 1 is 0 times, 2 times, Assume 0 times, 0 times, 2 times, 0 times, and 0 times. In this case, the busy ratio of each section is calculated by proportionally distributing the disk busy ratio (80%) with the I / O frequency distribution of each section. In this case, since the total number of I / Os is 20 in total, one I / O contributes to 4% of the overall busy rate of 80%. Therefore, as shown in FIG. 2C, the read busy rate of each partition is 12%, 4%, 32%, 0%, 0%, 0%, 16%, respectively, and the write busy rate of each partition is , 0%, 8%, 0%, 0%, 8%, 0%, and 0%, respectively. In this way, the Read busy rate and Write busy rate of each partition are calculated and set as the load of each partition.

  Note that the calculation of the busy rate of each partition described above is performed for a disk device in which the disk busy rate is determined to be an active disk instead of “0”. If the calculated partition busy rate is not “0”, the performance monitoring unit 2 inputs and stores the calculated partition busy rate (Read busy rate, Write busy rate) in the rearrangement candidate list storage unit 3. .

  Then, the relocation candidate list storage unit 3 sorts the busy rate of each partition received from the performance monitoring unit 2 as described above, that is, the read busy rate and the write busy rate, in order from the highest busy rate to the lowest busy rate, respectively. And manage their priorities. Specifically, as shown in FIG. 3, in the two types of processing of Read and Write, a total of four types of priority lists, each of two types of high busy rate and low busy rate, are managed. Each priority list holds a preset number (for example, five) of all the partitions in the storage system in descending order of the value. The partition registered in this list is the number of files that are rearranged by the file rearrangement control unit 4 to be described later, or a higher or lower partition is registered in the list, and is set at the top of each list. It remains in the list until it is evicted from within. However, the same partition number does not appear twice in one list. For example, if the list has a heap structure, the busy rate of each area is periodically compared with the lowest value of the list, and if it is within the list, it is sorted after the replacement with the lowest value. A priority list can be realized.

  Here, a method of sorting the list of the read / write busy rates in each input partition in the rearrangement candidate list storage unit 3 will be described. First, if the number of partitions has already been stored exceeding the maximum number of lists, the busy rate of the newly input partition is compared with the busy rate of the lowest partition of the stored list. If the value is higher (when the sections are arranged in descending order) or lower (when the sections are arranged in lower order), the section number and the busy rate value are exchanged. That is, the partition number is added to the list. Thereafter, the list in which the exchange has occurred is sorted, and if a duplicated partition is already registered, the oldest registered partition is deleted.

  For example, if there is a high read busy rate list whose maximum number of lists is “3” and it is currently empty, the above-mentioned disk partition read busy rates are 12% (# 1), 4% (# 2), and 32% (# 3) Consider the case of inserting 16% (# 7). First, since the list is empty, # 1, # 2, and # 3 are registered in the list and sorted. The status of the list is in the order of # 3 → # 1 → # 2. Next, when # 7 (16%) is inserted, since the list is already the maximum number (3), the busy rates of the lowest partition # 2 (4%) and the additional partition # 7 (16%) are compared. Here, since the busy rate of # 7 is high, # 2 and # 7 are exchanged. Then, sorting is performed again to obtain a new list # 3 → # 7 → # 1. Since there is no other # 7 before and after # 7, the inserted section # 7 is not duplicated and there is no need to delete duplicates.

  Note that the partition busy rate list is not limited to the separate formation of Read and Write as shown in FIG. For example, a list corresponding to only Read or Write may be stored, or the performance monitoring unit 2 described above measures the busy rate for each partition and rearranges without distinguishing between Read / Write. You may manage with the candidate list memory | storage part 3. FIG.

  Furthermore, in the above, the busy rate for each section is used as a value representing the load of each section. However, other measured values and calculated values are used as the loads of each section, and based on this, FIG. 3 shows. A list of each section may be managed. For example, the performance monitoring unit 2 measures a turn around time (TAT) that is a time required from the start of input per unit time to the completion of output for the input in each section, and the accumulated time is It may be a load in each section. Then, the list for selecting the section where the file is rearranged may be rearranged in order of increasing or decreasing load value. As described above, when TAT is used as a value representing the load on a partition, the value of the TAT may be high in a disk device that seems to be out of order. Therefore, as will be described later, by selecting a partition in which a file is to be rearranged based on the value of TAT, it is possible to select a disk that seems to be out of order and use it for applications such as saving data from the disk. .

  The file group selection control unit 4 has a function of specifying a file included in a certain section. Specifically, by using the layout information of the file system, it is possible to select the file group and metadata group using the partition from the partition number. Such reverse map information normally exists as internal information in recent file systems.

  In addition, the file relocation control unit 5 periodically (for example, once a day) reallocates the files stored in each section based on the list stored in the relocation candidate list storage unit 3 described above. It has the function to plan and execute the arrangement. Specifically, the following three processes A, B, and C are executed roughly.

(Process A)
First, based on the rearrangement candidate list storage unit 3, the partition having the highest busy rate (the highest partition or the partition from the highest order to the predetermined order) is selected, and the load is distributed to the currently operating disks. Predict and calculate the effect of the case. As a specific load distribution method, for example, a part of a file group is moved to another disk, and a single file with a high write request is distributed to multiple disks by a distribution method such as RAID5 (Redundant Arrays of Inexpensive Disks 5). A relocation process such as creating a copy of a file with a high read request distributed on multiple disks can be considered. In addition, the prediction calculation is set in advance for each case where only files with high read request partitions are distributed, only files with high write request partitions are distributed, and files with both high read / write partitions are distributed. The effect is digitized by the evaluation function, and a suitable file relocation process is determined based on the numeric value.

  Then, when a value such as a predicted evaluation after data rearrangement (for example, an evaluation value by an evaluation function or a busy rate of each section) is within a preset range, a file rearrangement process is executed. . Thereafter, the information in the file management unit 1 is appropriately changed along with the rearrangement, the partition is deleted from the rearrangement candidate list storage unit 3, and the process is terminated.

(Process B)
If the load after the file rearrangement is not reduced in the process A, the process B is performed. In this process B, first, a partition with a high busy rate is selected from the rearrangement candidate list storage unit 3, and the effect when the load is distributed including the currently inactive disks is related to the busy rate and power consumption. Predictive calculation is performed using the evaluation function EV (N). Since N in this evaluation function EV (N) is the number of disks to be used including non-working disks, the current number of working disks is already the lower limit.

  The evaluation function (N) uses a function that can obtain an optimal number of operating disks that balances the effect of decreasing the busy rate with the increase in the number of disks and the effect of increasing power consumption. For example, an evaluation function EV (N) as shown in FIG. 4 is used, which adds a busy rate and a constant multiple of the number of disks, and the evaluation value based on this evaluation function is better when the value is lower. Therefore, in the example of FIG. 4, Nmax indicates the most efficient number of disks. When the performance and power consumption are outside the predetermined threshold values, the operation of the system can be controlled by setting EV (N) to 0.

  Then, as a result of the evaluation using the evaluation function (N) described above, if it is determined that the file should be relocated using one or more non-working disks, the file is relocated. Thereafter, the information in the file management unit 1 is appropriately changed along with the rearrangement, the partition is deleted from the rearrangement candidate list storage unit 3, and the process is terminated.

  Note that the file rearrangement process for reducing the load on the high busy rate sections by the processes A and B described above is not limited to the process described above.

(Process C)
After executing the processes A and B, the process C is executed. In this process C, a partition with a low busy rate (the lowest partition or a partition from the lowest order to a predetermined order) is selected from the rearrangement candidate list storage unit 3, and a file group on that partition is selected as another file group. Estimate and calculate an evaluation value that represents the effect of moving to a working disk. When the calculated busy rate or the like falls within a predetermined range, the file is rearranged and the section is deleted from the rearrangement candidate list storage unit 3. At this time, the relocation of the file in the selected partition is executed by moving to a partition with a busy rate higher than the busy rate of the selected partition, especially after the file is moved. It is preferable that the disk device has a high possibility of being able to stop the disk device in which the partition is formed.

  In the above-described storage system, as shown in FIG. 1, the case where the function for performing each process is provided in one computer has been described, but each function may be provided in a plurality of computers. . For example, the file management unit 1 may exist in an independent server computer and may be configured as a distributed file system that serves as a metadata server that hides the physical location of a file from a client.

  Further, in the above-described storage system, a configuration in which each disk device is an independent file system and the file management unit 1 performs overall configuration is also conceivable. For example, when an image search system is considered, the file management unit 1 can be considered as a search processing unit, and disk arrangement of image data optimal for an access pattern can be performed.

[Operation]
Next, the operation of the above-described storage system will be described. First, the operation executed at regular time intervals (for example, every 10 minutes) in the performance monitoring unit 2 will be described with reference to the flowchart of FIG.

  First, the busy rate of each disk device and the read / write I / O frequency distribution of each partition are acquired and recorded (step S1). Subsequently, when the busy rate of the disk device is not “0” but is determined to be an active disk (Yes in step S2), the busy rate of each partition formed in the disk device is determined as the busy rate of the disk. Calculation is performed by proportional distribution in the I / O frequency distribution of each section (step S3). If the calculated partition busy rate is not “0”, it is input to the rearrangement candidate list storage unit 3 (step S4). At this time, in the rearrangement candidate list storage unit 3, as described above, the lists are sorted in order from the highest to the lowest. The above processing is executed at regular time intervals.

  Next, operations that are executed periodically (for example, once a day) by the file relocation control unit 4 will be described with reference to the flowcharts shown in FIGS. To do.

  First, the process A described above is executed. The operation at this time will be described with reference to FIGS. First, a partition with a high busy rate is selected from the rearrangement candidate list storage unit 3 (step S11). At this time, for example, it is assumed that the section # 11 in the disk device indicated by reference numeral 6 in FIG. Then, the effect when the load in this partition is distributed to other currently operating disk devices (partitions) is predicted and calculated based on the busy rate (step S12).

  Here, as shown by the arrow in FIG. 7A, the file in the selected partition # 11 is moved to the partition # 12 in the other operating disk device 7 to distribute the load. Think. The effect when such load distribution processing is performed is calculated as an evaluation value based on evaluation information such as an evaluation function stored in advance. Then, it is determined whether or not the evaluation value satisfies a preset criterion that can be determined that the effect is improved by the rearrangement (step S13). If the criterion is satisfied (Yes in step S13), the predicted load distribution process is actually executed (step S14). Then, as shown in FIG. 7B, one file stored in the partition # 11 is moved to and stored in the partition # 12 in the other disk device 7 in operation.

  Thereafter, the information in the file management unit 1 is appropriately changed as the files are rearranged. Further, the section is deleted from the high busy rate list in the rearrangement candidate list storage unit 3 (step S15), and the process ends.

  Here, when the effect of the dispersion process predicted in step S12 described above does not satisfy the standard (No in step S13), the process B is executed. The operation at this time will be described with reference to FIGS.

  First, a partition with a high busy rate is selected from the rearrangement candidate list storage unit 3 (step S21). At this time, for example, it is assumed that the section # 11 in the disk device indicated by reference numeral 6 in FIG. 9A is selected. Then, unlike the process A described above, the load in this partition is calculated by predicting the effect when the load is distributed, including the disk devices that are currently inactive (step S22).

  Here, as shown by the arrow in FIG. 9A, the file in the selected partition # 11 is moved to the partition # 14 in another non-operating disk device 9 to distribute the load. think of. The effect when such load distribution processing is performed is calculated as an evaluation value based on evaluation information such as an evaluation function stored in advance. Then, it is determined whether or not the evaluation value satisfies a preset criterion that can be determined that the effect is improved by the rearrangement (step S23). If the criterion is satisfied (Yes in step S23), the predicted load distribution process is actually executed (step S24). Then, as shown in FIG. 9B, one file stored in the partition # 11 is moved to and stored in the partition # 14 in the other non-operating disk device 9, and The non-operating disk device 9 is in an operating state.

  Thereafter, the information in the file management unit 1 is appropriately changed as the files are rearranged. Further, the section is deleted from the high busy rate list in the rearrangement candidate list storage unit 3 (step S25), and the process is terminated.

  As described above, in the storage system according to the present embodiment, data is rearranged by distributing files stored in high-load partitions or by copying and storing them in other disk devices. Yes. As a result, the load on the disk device in which the corresponding section is formed is reduced. As a result, power is efficiently used in the entire storage system, and power saving can be achieved. In particular, when distributing the load by increasing the number of non-working disks, the performance and power saving can be balanced by performing performance prediction using an evaluation function or the like before the actual distribution processing.

  Subsequently, after the process A or the process B is executed, the process C is executed. The operation at this time will be described with reference to FIGS.

  First, a partition with a low busy rate is selected from the rearrangement candidate list storage unit 3 (step S31). At this time, for example, it is assumed that the section # 13 in the disk device indicated by reference numeral 8 in FIG. Then, the effect of moving the file arranged in this partition to another operating disk device (partition) is predicted and calculated (step S32).

  Here, consider a case where the file in the selected partition # 13 is moved to the partition # 12 in the other operating disk device 7 as indicated by the arrow in FIG. An effect obtained when such load distribution processing is performed is calculated as an evaluation value based on evaluation information such as an evaluation function stored in advance, and this evaluation value is set in advance so that it can be determined that the effect is improved by rearrangement It is determined whether or not the set criteria are satisfied (step S33). If the criterion is satisfied (Yes in step S33), the predicted file movement is actually executed (step S34). Then, as shown in FIG. 11B, the file stored in the partition # 13 is moved to and stored in the partition # 12 in the other disk device 7 in operation.

  Thereafter, the information in the file management unit 1 is appropriately changed as the files are rearranged. Further, the section is deleted from the low busy rate list in the rearrangement candidate list storage unit 3 (step S35), and the process is terminated.

  After that, as described above, since the file in the partition # 13 that originally had a low load is moved to another partition, the subsequent load on the partition # 13 is further reduced. Therefore, when the busy rate of the disk device is later measured by the performance monitoring unit 2 and the busy rate of the disk device 8 in which the partition # 13 is located is “0”, the disk device 8 is not used. It can be determined that the disk is a working disk, and thereafter the power supply is stopped.

  As described above, according to the present invention, it is possible to efficiently increase the number of non-operating disks by moving the files in a partition with a low load to another disk device. Therefore, it is possible to stop the power supply to the disk device that has become inactive, and to achieve power saving of the entire storage system.

Here, as described above, the partition busy rate is calculated, and the processing amount for managing the sorted list is estimated.
Processing per unit time: When 1000 1TB disk units are divided into 1000 areas, all the disks are active disks, and the busy rate is aggregated once every 10 minutes, the number of disk busy rate acquisition times: 1K / 10 minutes = 1.67 pieces / sec. Size of each area: 1GB
Number of areas: 1M busy rate calculation: 1M / 10 minutes = 1.67K pieces / second Of the above, if 10% entered the list, area insertion and re-sorting to the list was 167 times / second is there. And if the length of the list is short, it will not be a heavy load on modern CPUs. If the number of active disks becomes 1/10, the number of managed partitions will also become 1/10, and management processing will be greatly reduced. Thus, even if 1000 disks are connected, the rearrangement process can be executed with a small amount of calculation and statistical information.

<Embodiment 2>
A first embodiment of the present invention will be described with reference to FIG. FIG. 12 is a functional block diagram showing the configuration of the storage system. In this embodiment, an outline of a storage system will be described.

As shown in FIG. 1, the storage system 100 in this embodiment is
For each of the disk devices 111 and 112, the operation status per unit time for the disk device is measured, the operating disk device is determined as an operating disk, the non-operating disk device is determined as a non-operating disk, A performance monitoring unit 101 that measures the load in the disk device per unit time and stores the measurement result;
A data rearrangement unit that stores data stored in the disk device determined to be an active disk based on a measurement result by the performance monitoring unit in another disk device having a higher load than the load in the disk device. 102,
Is provided.

And in the above storage system,
The performance monitoring unit measures a load in each section in which the storage area in the disk device is divided into a plurality of parts, stores the measurement result,
The data rearrangement unit, in the partition formed in the disk device determined to be an active disk based on the measurement result by the performance monitoring unit, in the partition having a lower load than the other partition. Store the stored data in the other disk device determined to be an active disk.
The structure is taken.

In the above storage system,
The data rearrangement unit, based on the measurement result by the performance monitoring unit, data stored in the sections from the order of low load to the preset order, the other disks determined as active disks Store in the device,
The structure is taken.

  According to the above invention, first, in the storage system, the operating status per unit time is measured for each disk device. Then, the disk device operating per unit time is determined as an active disk, and the disk device not operating per unit time is determined as a non-operating disk. As a result, for example, by stopping the disk device determined to be a non-working disk, it is possible to save power in the entire storage system.

  Further, in the storage system, as described above, the operation status for each disk device is measured, and the load on the disk device, for example, the load for each partition formed in the disk device is measured and stored. Then, the data stored in the partition having a lower load than the other partitions among the partitions formed in the disk device that is the working disk is stored in the other disk device that is the working disk. Note that the above-described data movement may be executed in the order of sections with a low load.

  As a result, the disk device that has moved the data to another disk device that is in operation has a low load from the beginning, so that there is a high possibility that the access will be reduced and become a non-working disk. If the disk becomes a non-working disk, the disk device is stopped as described above. Thereby, further power saving of the entire storage system can be achieved.

In the above storage system,
The data rearrangement unit, in the partition formed in the disk device determined to be an active disk based on the measurement result by the performance monitoring unit, in the partition having a higher load than the other partition. To relocate the data stored in each of the above partitions so as to reduce the load,
The structure is taken.

In another aspect of the storage system,
The data rearrangement unit is a part or all of the data stored in the partition having a higher load than the other partition among the partitions formed in the disk device determined to be an active disk. Store it in the other disk devices above, and rearrange the data.
The structure is taken.

In the above storage system,
The data rearrangement unit rearranges the data including the disk device determined to be a non-working disk.
The structure is taken.

In the above storage system,
The data rearrangement unit rearranges the data stored in each of the partitions so as to reduce the load on the partition from the highest load to a preset order based on the measurement result by the performance monitoring unit. I do,
The structure is taken.

In the above storage system,
The performance monitoring unit measures the accumulated time of turnaround time, which is the time required from the start of input per unit time in each section to the completion of output for the input, as the load in each section.
The structure is taken.

  According to the above invention, in the storage system, data is rearranged by distributing or replicating data stored in a high-load partition and storing it in another disk device. As a result, the load on the disk device in which the corresponding partition is formed is reduced, so that power is efficiently used in the entire storage system, and power saving can be achieved.

The storage system described above can be realized by incorporating a program into a computer. Specifically, a program according to another embodiment of the present invention is stored in a computer.
For each disk device, measure the operating status per unit time for the disk device, determine that the disk device that was operating is an active disk, determine that the disk device that is not operating is a non-operating disk, and A performance monitoring unit for measuring the load on the disk device and storing the measurement result;
A data rearrangement unit that stores data stored in the disk device determined to be an active disk based on a measurement result by the performance monitoring unit in another disk device having a higher load than the load in the disk device. When,
It is a program for realizing.

And in the above program,
The performance monitoring unit measures a load in each section in which the storage area in the disk device is divided into a plurality of parts, stores the measurement result,
The data rearrangement unit, in the partition formed in the disk device determined to be an active disk based on the measurement result by the performance monitoring unit, in the partition having a lower load than the other partition. Store the stored data in the other disk device determined to be an active disk.
The structure is taken.

In the above program,
The data rearrangement unit, in the partition formed in the disk device determined to be an active disk based on the measurement result by the performance monitoring unit, in the partition having a higher load than the other partition. To relocate the data stored in each of the above partitions so as to reduce the load,
The structure is taken.
A data storage method according to another embodiment of the present invention, which is executed when the above-described storage system operates,
For each disk device, measure the operating status per unit time for the disk device, determine that the disk device that was operating is an active disk, determine that the disk device that is not operating is a non-operating disk, and Measure the load on the above disk device, store the measurement results, perform performance monitoring,
Based on the measurement result, data stored in the disk device determined to be an active disk is stored in another disk device having a higher load than the load in the disk device, and data is rearranged. .

A data storage method according to another embodiment of the present invention, which is executed when the above-described storage system operates,
During the performance monitoring, measure the load in each partition into which the storage area in the disk device is divided into a plurality, and store the measurement results;
At the time of rearrangement of the data, among the partitions formed in the disk device determined to be an active disk based on the measurement result by the performance monitoring, the load is stored in the partition having a lower load than the other partitions. Store the stored data in the other disk device determined to be an active disk,
The structure is taken.

In another aspect of the data storage method,
At the time of the data rearrangement, among the partitions formed in the disk device determined to be an active disk based on the measurement result of the performance monitoring, the load in the partition is higher than that of the other partitions To relocate the data stored in each of the above partitions so as to reduce
The structure is taken.

  Even the invention of the program or the data storage method having the above-described configuration can achieve the above-described object of the present invention because it has the same operation as the above storage system.

  The present invention can be applied to a storage system including a plurality of disk devices such as a disk array device, a NAS device, an aggregate NAS device, and a grid storage, and has industrial applicability.

DESCRIPTION OF SYMBOLS 1 File management part 2 Performance monitoring part 3 Reallocation candidate list memory | storage part 4 File reallocation control part 5 File group selection control part 6, 7, n Disk apparatus 10,100 Storage system 101 Performance monitoring part 102 Data reallocation control part 111 112 disk unit

Claims (7)

The operation status per unit time for each disk device is measured for each disk device, the operating disk device is determined as an operating disk, the non-operating disk device is determined as a non-operating disk, and the disk device the load per unit time in each compartment storage area of the inner is divided into a plurality measures, and performance monitoring unit for storing the measurement result,
Based on the measurement result by the performance monitoring unit , among the partitions formed in the disk device determined to be an active disk , the load is reduced so as to reduce the load in the partition that is higher than the other partitions. A part or all of the data stored in the partition that is higher than the other partition is stored in the other disk device, the data is rearranged, and then the measurement result by the performance monitoring unit Based on the above, among the partitions formed in the disk device determined to be an active disk, data stored in the partition having a lower load compared to the other partitions is determined to be an active disk. A data relocation unit for storing in the disk device ,
Storage system with The storage system according to claim 1 ,
The data rearrangement unit, based on the measurement result of the performance monitoring unit, data stored in the partition from the order of low load to the preset order, the other disks determined as active disks Store in the device,
Storage system. The storage system according to claim 1 or 2 ,
The data rearrangement unit rearranges the data including the disk device determined to be a non-working disk.
Storage system. The storage system according to any one of claims 1 to 3 ,
The data rearrangement unit rearranges the data stored in each partition so as to reduce the load in the partition from the order of higher load to the preset order based on the measurement result by the performance monitoring unit. I do,
Storage system. The storage system according to any one of claims 1 to 4 , wherein
The performance monitoring unit measures, as the load in each partition, the accumulated time of turnaround time, which is the time required from the start of input per unit time in each partition to the completion of output for the input,
Storage system. On the computer,
The operation status per unit time for each disk device is measured for each disk device, the operating disk device is determined as an operating disk, the non-operating disk device is determined as a non-operating disk, and the disk device the load per unit time in each compartment storage area of the inner is divided into a plurality measures, and performance monitoring unit for storing the measurement result,
Based on the measurement result by the performance monitoring unit , among the partitions formed in the disk device determined to be an active disk , the load is reduced so as to reduce the load in the partition that is higher than the other partitions. A part or all of the data stored in the partition that is higher than the other partition is stored in the other disk device, the data is rearranged, and then the measurement result by the performance monitoring unit Based on the above, among the partitions formed in the disk device determined to be an active disk, data stored in the partition having a lower load compared to the other partitions is determined to be an active disk. A data relocation unit for storing in the disk device ,
A program to realize The operation status per unit time for each disk device is measured for each disk device, the operating disk device is determined as an operating disk, the non-operating disk device is determined as a non-operating disk, and the disk device storage area of the inner is to measure the load per unit time in each compartment is divided into a plurality, and stores the measurement result, performs performance monitoring,
Based on the measurement result, among the partitions formed in the disk device determined to be an active disk , the load is other than the other so that the load in the partition is high compared to the other partitions. A part or all of the data stored in the partition higher than the partition is stored in the other disk device, and then formed on the disk device determined to be an active disk based on the measurement result. Among the partitions, data stored in the partition having a lower load than the other partitions is stored in another disk device determined to be an active disk, and data is rearranged.
Data storage method.

Images