JP2015200930A - Storage device, data reading method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a storage device capable of suppressing reduction in throughput even when a plurality of clients simultaneously access identical data.SOLUTION: A storage device includes: a plurality of storage units 2 each of which stores identical data; a management unit 4 that selects any one of the plurality of storage units 2 on the basis of first load information indicating the load state of each of the storage units; and reading units 3 that read data from the selected storage units 2.

Description

  The present invention relates to a storage device, a data reading method, and a program.

  In recent years, opportunities to handle time-series data such as data generated by various sensors, video data of surveillance cameras, video data of television broadcasting, audio data of radio broadcasting, and the like are increasing. By analyzing such time-series data in real time, it is possible to know the current event in real time, or to predict an event that will occur in the future based on past data.

  For example, in order to analyze time-series data in real time or to reproduce video data or audio data without delay, it is necessary to read out stored data with a constant throughput. In order to read the stored data with a constant throughput, it is important to distribute the processing load in order to prevent the processing load of the data reading process from being concentrated on a specific resource.

  Patent Document 1 describes a distributed data management system that moves data between storages in order to prevent the processing load from concentrating on a specific storage that stores a plurality of frequently accessed data, as a technique related to load distribution. Has been. In this distributed data management system, the timing for moving data is determined based on the load status of the source server.

  Patent Document 2 describes a load distribution system that distributes and executes a job to a plurality of computers as a technique related to load distribution. In this load balancing system, data necessary for job execution is stored in a high-speed storage that can read data at high speed or a large-capacity storage that includes a large-capacity disk drive. When the data necessary for executing the designated job is stored in the large-capacity storage, the data is moved from the large-capacity storage to the high-speed storage, and the data is read from the high-speed storage.

JP 2013-222221 A JP 2008-015888 A

  However, in the distributed data management system described in Patent Document 1 and the load distribution system described in Patent Document 2, when a plurality of clients access the same data at the same time, the concentration of processing load on the same storage is suppressed. I couldn't. For example, when analyzing time-series data in real time and obtaining multiple analysis results, multiple analysis processes are performed on the same data at the same time, so multiple clients may access the same data at the same time. is there. For this reason, when reading the stored data, the throughput may decrease.

  An object of the present invention is to provide a storage device, a data reading method, and a program capable of suppressing a decrease in throughput even when a plurality of clients simultaneously access the same data.

The storage device according to the present invention comprises:
A plurality of storage units each storing the same data;
A management unit that selects one of the plurality of storage units based on first load information indicating a load status of each storage unit;
And a reading unit that reads the data from the selected storage unit.

The data reading method according to the present invention includes:
Based on the first load information indicating the load status of each of a plurality of storage units each storing the same data, select one of the plurality of storage units,
The data is read from the selected storage unit.

The program according to the present invention is
Computer
A plurality of storage units each storing the same data;
A management unit that selects one of the plurality of storage units based on first load information indicating a load status of each storage unit;
A program for causing a storage device to include a reading unit that reads the data from the selected storage unit.

  According to the present invention, it is possible to suppress a decrease in throughput even when a plurality of clients simultaneously access the same data.

It is a block diagram which shows the structure of the real-time data recording device 1 concerning the 1st Embodiment of this invention. It is a figure for demonstrating the 1st load information of each storage 2, and the selection method of the storage 2. FIG. It is a figure for demonstrating the 1st load information of each storage 2 at the time of adjusting an upper limit, and the selection method of the storage 2. FIG. 4 is a flowchart for explaining data recording processing of the real-time data recording apparatus 1. 5 is a flowchart for explaining a load status confirmation process of the real-time data recording apparatus 1; 6 is a sub-flowchart for explaining details of step S201 in FIG. 6 is a sub-flowchart for explaining details of step S204 in FIG. 5. 5 is a flowchart for explaining a read request response process of the real-time data recording apparatus 1. It is a block diagram which shows the structure of the memory | storage device 100 which concerns on the 2nd Embodiment of this invention. 4 is a flowchart for explaining an operation example of the storage device 100.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In addition, in this specification and drawing, the description which overlaps may be abbreviate | omitted by attaching | subjecting the same code | symbol about the component which has the same function.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of a real-time data recording apparatus 1 according to the first embodiment of the present invention. The real-time data recording device 1 has a plurality of storages 2, a plurality of read servers 3, a management server 4, and an internal network 5. In FIG. 1, the real-time data recording apparatus 1 includes a plurality of read servers 3, but may include a single read server 3.

  The storage 2 is a device for storing data, and includes, for example, a non-volatile memory such as a flash memory, a magnetic recording medium such as an HDD (Hard Disk Drive), and a CPU (Central Processing Unit) that is a control device. The storage 2 is connected to the read server 3 and the management server 4 via the internal network 5. Further, the storage 2 outputs stored data to the read server 3 in accordance with an instruction from the read server 3.

  The read server 3 is a device that reads data from the storage 2. The read server 3 includes a CPU that is a control device. The read server 3 is connected to the management server 4 and the storage 2 via the internal network 5, and is connected to the client 72 via the network 62. When the read server 3 reads the data according to the instruction of the management server 4, the read server 3 transmits the data to the client 72 that has requested the data read.

  The storage 2 and the read server 3 are monitored and controlled according to SNMP (Simple Network Management Protocol), and the CPU usage rate, the number of accesses, and the like are recorded and held in an MIB (Management Information Base).

  The management server 4 is a management unit that manages the entire real-time data recording apparatus 1. The management server 4 is connected to the real-time data input device 71 via the network 61, is connected to the storage 2 and the read server 3 via the internal network 5, and is connected to the client 72 via the network 62.

  The management server 4 includes a microprocessor and a memory (not shown), and controls the operation of the real-time data recording apparatus 1 by reading and executing a program stored in the memory or the like. Specifically, when the real-time data input device 71 acquires real-time data and inputs it to the management server 4, the management server 4 accepts the input real-time data and multicasts the real-time data to the plurality of storages 2. As a result, real-time data is recorded in the plurality of storages 2. As a result, each of the plurality of storages 2 stores the same data.

  In addition, when the management server 4 receives a read request for requesting reading of data stored in the storage 2 from the client 72, the management server 4 performs a data read process corresponding to the read request based on the load status of the real-time data recording apparatus 1. Determine whether to do it. Further, when performing the data reading process, the management server 4 distributes the loads of the plurality of storages 2 and the plurality of reading servers 3 based on the load status of the real-time data recording device 1. The information indicating the load status of the real-time data recording device 1 is, for example, first load information indicating the load status of the storage 2 and second load information indicating the load status of the read server 3. Further, the first load information may include a plurality of types of information each indicating the load status of each resource in the storage 2. The first load information includes, for example, a usage rate of a CPU (Central Processing Unit) that is a control device included in the storage 2 and a storage access number that is the number of accesses to each storage 2. Similarly, for the read server 3, the second load information includes, for example, a usage rate of a CPU that is a control device included in the read server 3 and a read server access number that is the number of accesses to each read server 3.

  The management server 4 uses the first load information and the second load information as information indicating the load status of the real-time data recording apparatus 1 to determine whether or not to perform the data reading process. For example, when the total number of storage accesses to each storage 2 is less than or equal to the upper limit value and the total number of read server accesses to each read server 3 is less than or equal to the upper limit value, the management server 4 performs data read processing. At this time, the management server 4 acquires the upper limit value of the number of storage accesses for each storage 2 and sets the sum as the upper limit value for the total number of storage accesses. The same applies to the read server 3, and the management server 4 acquires the upper limit value of the read server access number for each read server 3, and sets the sum as the upper limit value for the total read server access number.

  Further, when performing the data read process, the management server 4 selects the storage 2 used for the data read process based on the first load information, thereby distributing the load among the plurality of storages 2. Further, the management server 4 selects the read server 3 used for the data read process based on the second load information, thereby distributing the load among the plurality of read servers 3. For example, the management server 4 can select the storage 2 based on the number of storage accesses. Specifically, the management server 4 calculates the difference between the storage access number of each storage 2 and the upper limit value set for each storage 2, and based on the calculation result, Select. More specifically, the management server 4 selects the storage 2 with the largest calculation result.

  FIG. 2 is a diagram for explaining the first load information of each storage 2 and the selection method of the storage 2. In the example of FIG. 2, the real-time data recording apparatus 1 has three storages 2 of storages A to C. It is assumed that the upper limit values provided for each of the storages A to C are all 5. Here, it is assumed that the number of storage accesses is 2, 3, and 4 in order for each of the storages A to C. In this case, for each of the storages A to C, the difference between the upper limit value and the number of storage accesses is 3, 2, and 1 in order. Therefore, in this case, the management server 4 selects the storage A having the largest difference.

  However, in the storage 2, the processing load of the storage 2 may increase due to factors other than data access processing for writing and reading data. For example, when a failure occurs in the storage A in FIG. 2, a rebuild process may occur as a background process. In this case, the difference between the number of storage accesses and the upper limit value is not necessarily a value reflecting the processing load of the storage A, and this difference is a large value even though the processing load is increased. Therefore, the management server 4 adjusts the upper limit value of the number of storage accesses based on the CPU usage rate of each storage 2.

  When the CPU usage rate of each storage 2 is equal to or greater than a predetermined threshold, the management server 4 limits the upper limit value of the storage access number to the difference between the upper limit value and the storage access number, and the CPU usage of the storage 2 Reflect the load situation based on the rate. In addition, the management server 4 counts the number of times the CPU usage rate of each storage 2 is continuously greater than or equal to a predetermined threshold value. It may be determined that the load is high, and the upper limit value of the number of storage accesses may be limited. “Count” in FIG. 2 indicates the number of times the CPU usage rate has continuously exceeded the threshold value. By using the number of times the CPU usage rate is continuously equal to or higher than the predetermined threshold, the upper limit value is not limited when the CPU usage rate is instantaneously increased and the CPU usage rate is immediately decreased. Therefore, it is possible to prevent an access error from occurring in response to a read request from the client 72 even though the storage 2 is not actually in a high load state.

  FIG. 3 is a diagram for explaining the first load information of each storage 2 and the selection method of the storage 2 when the upper limit value is adjusted. In the adjustment of the upper limit value, in this example, the threshold value for the CPU usage rate is set to 70%, and the upper limit value is limited when the CPU usage rate becomes three times or more and 70% or more continuously. FIG. 2 shows that the CPU usage rate of the storage A is “90%” and the count is “2”. In this case, since the CPU usage rate of the storage A is 70% or more, the management server 4 increments the count to “3”. Since the count is 3 or more, the management server 4 limits the upper limit value of the number of storage accesses for this storage A. Here, the limited upper limit value is “2”. Similarly, the management server 4 determines whether the CPU usage rate is 70% or more for the storages B and C. Since the CPU usage rate of the storage C is 70% or more, the management server 4 increments the count to “1”. Since the count has not yet reached 3 or more, the management server 4 does not limit the upper limit value for the storage C.

  When the upper limit value is limited, as shown in FIG. 3, the difference between the upper limit value and the number of storage accesses for storage A is “0”. Thereby, among the storages A to C, the storage 2 having the largest difference between the upper limit value and the number of storage accesses becomes the storage B.

  Further, when the management server 4 selects the read server 3 using the second load information, it is the same as when the storage 2 is selected. In the above description, the selection of the read server 3 is described by replacing the storage 2 with the read server, the first load information with the second load information, and the storage access number with the read server access number. When the management server 4 selects the storage 2 and the read server 3 that perform data read processing, the management server 4 notifies the selected read server 3 of the selected storage 2 and causes the selected storage 2 to read data. Thus, only the selected read server 3 among the plurality of read servers 3 executes data reading.

  Next, an operation example when the real-time data recording apparatus 1 performs various processes will be described with reference to FIGS. The processing performed by the real-time data recording device 1 mainly includes data recording processing for recording real-time data, load status confirmation processing for confirming the load status of the real-time data recording device 1, and reading in response to a read request from the client 72. Request response processing. The read request response process includes a data read process.

  FIG. 4 is a flowchart for explaining an operation example when the real-time data recording apparatus 1 performs a data recording process. The data recording process is started when the real-time data input device 71 inputs real-time data to the real-time data recording device 1.

  When the real time data input device 71 inputs real time data to the real time data recording device 1, the management server 4 of the real time data recording device 1 accepts the input real time data (step S100).

  The management server 4 converts the data format of the received real-time data into a file (step S101).

  Subsequently, the management server 4 multicasts and records the filed real-time data in all the storages 2 (step S102). As a result, the same data is stored in all the storages 2.

  FIG. 5 is a flowchart for explaining an operation example when the real-time data recording apparatus 1 performs the load status confirmation process. The load status confirmation process is periodically executed by the management server 4 of the real-time data recording apparatus 1.

  The management server 4 selects one storage 2 from the storages 2 for which the storage load confirmation process has not been completed (step S200). Then, the management server 4 performs a storage load confirmation process for confirming the load status for the selected storage 2 (step S201). The detailed contents of the storage load confirmation process will be described later with reference to FIG. The management server 4 determines whether there is a storage 2 that has not been subjected to the storage load confirmation process (step S202).

  When there is a storage 2 that has not been subjected to the storage load confirmation process, the management server 4 performs the processes from step S200 to step S202 again. Thereby, the processing of step S200 to step S202 is repeated until the storage load confirmation processing is performed for all the storages 2.

  When there is no storage 2 for which the storage load confirmation process has not been completed, the management server 4 subsequently selects the read server 3 for which the server load confirmation process has not been completed (step S203). Then, the management server 4 performs server load confirmation processing for the selected read server 3 (step S204). The detailed contents of the server load confirmation process will be described later with reference to FIG. The management server 4 determines whether or not there is a read server 3 that has not finished the server load confirmation process (step S205).

  When there is a read server 3 that has not finished the server load confirmation process, the management server 4 performs the processes of steps S203 to S205 again. Thereby, the process of step S203-step S205 is repeated until the server load confirmation process is performed about all the read servers 3. FIG. As a result, the load status is periodically confirmed for all storages 2 and all read servers 3.

  FIG. 6 is a sub-flowchart of step S201 in FIG. 5, and is a diagram for explaining an operation example when the management server 4 performs a storage load confirmation process. In the storage load confirmation processing, for each storage 2, whether the load status is confirmed, first load information is generated and recorded in the management table, and the upper limit value of the number of storage accesses included in the first load information is limited. And a process of determining whether or not and recording the current upper limit value in the management table.

  The management server 4 acquires the current CPU usage rate, the current number of storage accesses, and the upper limit value of the number of storage accesses for the selected storage 2 (step S300). The management server 4 can obtain the current CPU usage rate and the current storage access number from the MIB of the storage 2, for example. Further, the management server 4 can acquire the upper limit value of the number of accesses of the selected storage 2 from the management table.

  The management server 4 determines whether or not the acquired CPU usage rate is greater than or equal to a predetermined threshold (step S301). When the CPU usage rate is equal to or greater than the threshold, the management server 4 determines whether or not the number of times the CPU usage rate is continuously equal to or greater than the threshold is equal to or greater than the specified number (step S302). When the number of times that the threshold is continuously exceeded is equal to or more than the specified number, the management server 4 limits the upper limit value of the number of storage accesses (step S303).

  When the number of times the CPU usage rate has continuously exceeded the threshold is less than the specified number, the management server 4 counts up the number of times the CPU usage rate has continuously exceeded the threshold (step S304). If the CPU usage rate is less than the threshold, the management server 4 releases the restriction on the upper limit value of the number of storage accesses (step S305). When the restriction on the upper limit value of the storage access number is released, the management server 4 initializes the number of times that the threshold value is continuously equal to or greater than the threshold value to “0” (step S306).

  When the process of step S303, the process of step S304, or the process of step S306 ends, the management server 4 rewrites the management table and updates the current CPU usage rate and the number of storage accesses. If the upper limit value of the storage access number has changed, the management server 4 also rewrites and updates the management table for the upper limit value of the storage access number (step S307).

  FIG. 7 is a sub-flowchart of step S204 in FIG. 5 and is a diagram for explaining an operation example when the management server 4 performs server load confirmation processing. In the server load confirmation process, for each read server 3, the load status is confirmed, second load information is generated and recorded in the management table, and the upper limit value of the number of read server accesses included in the second load information is limited. And a process of determining whether to perform the process and recording the current upper limit value in the management table.

  The management server 4 obtains the current CPU usage rate, the current number of read server accesses, and the upper limit value of the number of read server accesses for the selected read server 3 (step S400). The management server 4 can obtain, for example, the current CPU usage rate and the current number of read server accesses from the MIB of the read server 3. Further, the management server 4 can acquire the upper limit value of the number of accesses of the selected read server 3 from the management table.

  The management server 4 determines whether or not the acquired CPU usage rate is greater than or equal to a predetermined threshold (step S401). When the CPU usage rate is equal to or higher than the threshold value, the management server 4 determines whether or not the number of times the CPU usage rate is continuously equal to or higher than the threshold value is equal to or higher than the specified number (step S402). When the number of times that the threshold value is continuously exceeded is equal to or more than the specified number, the management server 4 limits the upper limit value of the number of read server accesses (step S403).

  When the number of times the CPU usage rate has continuously exceeded the threshold is less than the specified number, the management server 4 counts up the number of times the CPU usage rate has continuously exceeded the threshold (step S404). If the CPU usage rate is less than the threshold value, the management server 4 releases the restriction on the upper limit value of the read server access number (step S405). When the restriction on the upper limit value of the number of read server accesses is canceled, the management server 4 initializes the number of times that the threshold value is continuously exceeded or more to “0” (step S406).

  When the process of step S403, the process of step S404, or the process of step S406 ends, the management server 4 rewrites the management table and updates the current CPU usage rate and the number of read server accesses. Further, when the upper limit value of the read server access number has changed, the management server 4 rewrites and updates the management table for the upper limit value of the read server access number (step S407).

  FIG. 8 illustrates an operation example when the management server 4 performs a read request response process when the client 72 inputs a read request for requesting reading of data stored in the real-time data recording device 1 to the management server 4. It is a flowchart for doing.

  The management server 4 receives the read request input by the client 72 (step S500). The management server 4 determines whether the total number of storage accesses is equal to or less than the upper limit. For example, the management server 4 acquires the storage access number and the upper limit value of each storage 2 from the management table, and determines whether or not the total storage access number is equal to or less than the total upper limit value (step S501). .

  When the total number of storage accesses is equal to or less than the upper limit, the management server 4 selects one of the plurality of storages 2 as the storage 2 used for the data read process based on the first load information. For example, as described with reference to FIG. 2, the management server 4 selects the storage 2 having the largest difference between the upper limit value of each storage 2 and the number of storage accesses (step S502).

  The management server 4 determines whether or not the total number of read server accesses is equal to or less than the upper limit. For example, the management server 4 acquires the number of read server accesses and the upper limit value of each read server 3 from the management table, and determines whether or not the sum of the read server access numbers is equal to or less than the sum of the upper limit values ( Step S503).

  When the total number of read server accesses is equal to or less than the upper limit, the management server 4 selects one of the plurality of read servers 3 as the read server 3 used for the data read process based on the second load information. The management server 4 also selects, for example, the read server 3 that has the largest difference between the upper limit value of each read server 3 and the number of read server accesses (step S504).

  Then, the management server 4 notifies the selected read server 3 of the selected storage 2 to read data from the selected storage 2 (step S505).

  If the total number of read server accesses is less than the upper limit, the management server 4 cancels the selection of the storage 2 selected in step S502 (step S506). Then, the management server 4 returns an error response to the client 72. Also, in the case where the total number of storage accesses is less than the upper limit in step S501, the management server 4 returns an error response to the client 72 (step S507).

(Modification)
While the present invention has been described with reference to the embodiments, the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  For example, in the above embodiment, the real-time data recording apparatus 1 includes the plurality of read servers 3, but the present invention is not limited to such an example. For example, the real-time data recording apparatus 1 may have one read server 3. The storage 2 and the read server 3 can take various configurations according to the processing capability of each resource.

  In the above embodiment, the read request response process in FIG. 8 is executed in response to a read request from the client 72, but the present invention is not limited to this example. For example, the management server 4 can periodically select the storage 2 based on the first load information even when the reading process has already been started. For example, the management server 4 selects one of the plurality of storages 2, and if the first load information of the selected storage 2 satisfies a predetermined condition, the management server 4 selects another storage 2 to store data. The storage 2 used for reading is switched. Thereby, even after the reading process has already started, if the load on the storage 2 used for the reading process increases, the storage 2 used for the reading process can be switched. In this case, when the read request is received, any of the plurality of storages 2 may be selected in a predetermined order, or any of the plurality of storages 2 may be selected based on the first load information. Good. In addition, the management server 4 repeatedly acquires the first load information, counts the number of times the first load information continuously satisfies the predetermined condition, and if the number exceeds the specified number, You may choose. As a result, it is possible to suppress frequent switching of the storage 2 used for reading data.

  In the above embodiment, the configuration and operation example of the real-time data recording apparatus 1 have been mainly described. However, a computer program for realizing each function of the real-time data recording apparatus 1 can be created. In addition, a computer-readable recording medium storing such a computer program can be provided. Further, the computer program may be distributed via a network by, for example, wired or wireless communication without using a recording medium.

  As described above, according to the first embodiment of the present invention, the real-time data recording device 1 that is an example of the storage device includes the storage 2 that is the storage unit, the management server 4 that is the management unit, and the reading unit. And a read server 3. The storage 2 stores the same data, and the management server 4 selects one of the plurality of storages 2 based on the first load information indicating the load status of each storage 2. The read server 3 reads data from the selected storage 2. Thereby, since the same data is stored in the plurality of storages 2, the data stored in the real-time data recording apparatus 1 can be taken out from any storage 2 without moving the data. Since data can be extracted from the storage 2 selected based on the load status of each storage 2, it is possible to suppress the concentration of access to the same storage 2 and to extract data from the storage 2 with a low load. it can. Therefore, even when a plurality of clients access the same data at the same time, it is possible to suppress a decrease in throughput.

  Further, according to the modification of the above embodiment, when the management server 4 selects any one of the plurality of storages 2 and the first load information of the selected storage 2 satisfies a predetermined condition, By selecting the storage 2, the storage 2 used for reading data is switched. Thereby, even after the reading process is started, the storage 2 used for the reading process can be switched according to the load status of the storage 2. Therefore, it is possible to more reliably suppress a decrease in throughput.

  In addition, according to the above embodiment, the management server 4 repeatedly acquires the first load information, counts the number of times the first load information continuously satisfies the predetermined condition, and the number of times exceeds the specified number. If this is the case, another storage 2 is selected. As a result, when the period during which the first load information satisfies the predetermined condition is short, the storage 2 cannot be switched. Therefore, it is possible to prevent the reliability of the storage 2 from being frequently switched and lowering the reliability. Become.

  Further, according to the embodiment, the management server 4 selects one of the plurality of storages 2 based on the first load information when receiving a read request for requesting data read. As a result, the storage 2 used for the reading process is selected according to the load status of each storage 2 at the start of reading, so that it is possible to more reliably suppress a decrease in throughput.

  Further, according to the above embodiment, each of the plurality of storages 2 has a CPU that is a control device, and the management server 4 has at least the usage rate of each CPU and the number of accesses to each storage 2 as the first load information. Either one is used. This makes it possible to select the storage 2 with a low load based on the CPU usage rate or the number of accesses. Therefore, it is possible to more reliably suppress a decrease in throughput.

  Further, according to the embodiment, the management server 4 uses the usage rate and the number of accesses as the first load information, and sets the number of accesses to the storage 2 and the storage 2 for each storage 2. The difference between the calculated upper limit value is calculated, one of the plurality of storages 2 is selected based on the calculation result, and the upper limit value is adjusted for each storage 2 based on the CPU usage rate. Thereby, the load status of the storage 2 can be grasped more accurately based on the usage rate and the number of accesses. In addition, the processing load increases due to internal processing that does not involve external access, and even if the CPU usage rate is increasing although the number of accesses does not increase, it becomes a value according to the increase in processing load. The storage 2 can be selected based on the difference. Therefore, it is possible to more reliably suppress a decrease in throughput.

  Further, according to the above embodiment, the management server 4 repeatedly acquires the CPU usage rate, counts the number of times the CPU usage rate continuously exceeds a predetermined threshold, and when the number of times exceeds a specified value, Lower the upper limit. Thereby, when the state where the processing load is determined to be high based on the CPU usage rate continues for a predetermined time or more, the number of accesses is limited, and when the processing load increases instantaneously, the number of accesses is not limited. Therefore, it is possible to prevent the number of accesses from being unnecessarily limited, and it is possible to more reliably suppress a decrease in throughput.

  In addition, according to the above embodiment, the real-time data recording apparatus 1 includes the plurality of reading servers 3, and the management server 4 uses the plurality of reading servers based on the second load information indicating the load status of each reading server 3. 3 is selected. Then, the selected read server 3 reads data from the selected storage 2. As a result, the processing load is also distributed to the read server 3, and it is possible to more reliably suppress a decrease in throughput.

(Second Embodiment)
FIG. 9 is a block diagram showing a configuration of the storage device 100 according to the second embodiment of the present invention.

  The storage device 100 includes a plurality of storage units 120, a management unit 130, and a reading unit 140. Each storage unit 120 stores the same data. The management unit 130 selects one of the plurality of storage units 120 based on the first load information indicating the load status of each storage unit 120. The reading unit 140 reads data from the selected storage unit 120.

  FIG. 10 is a flowchart for explaining the operation of the storage device 100. The management unit 130 of the storage device 100 selects one of the plurality of storage units 120 based on the first load information (Step S600). The reading unit 140 reads data from the selected storage unit 120 (step S601).

  As described above, according to the second embodiment of the present invention, the same data is stored in the plurality of storage units 120. The data can be extracted from any storage unit 120 without moving the data. In addition, since data can be extracted from the storage unit 120 selected based on the load status of each storage unit 120, the concentration of access to the same storage unit 120 is suppressed, and data from the storage unit 120 with low load is reduced. Can be taken out. Therefore, even when a plurality of clients access the same data at the same time, it is possible to suppress a decrease in throughput.

DESCRIPTION OF SYMBOLS 1 Real time data recording device 2 Storage 3 Reading server 4 Management server 71 Real time data input device 72 Client

Claims (10)

A plurality of storage units each storing the same data;
A management unit that selects one of the plurality of storage units based on first load information indicating a load status of each storage unit;
And a reading unit that reads the data from the selected storage unit.   After selecting one of the plurality of storage units, the management unit selects another storage unit when the first load information of the selected storage unit satisfies a predetermined condition. The storage device according to claim 1, wherein the storage unit used for reading the data is switched.   The management unit repeatedly acquires the first load information, counts the number of times the first load information continuously satisfies the predetermined condition, and if the number exceeds the specified number, the other storage The storage device according to claim 2, wherein the storage unit is selected.   4. The control unit according to claim 1, wherein the management unit selects any of the plurality of storage units based on the first load information when receiving a read request for requesting reading of the data. 5. Storage device. Each of the plurality of storage units has a control device,
5. The storage device according to claim 1, wherein the management unit uses at least one of a usage rate of each control device and the number of accesses to each storage unit as the first load information. 6.   The management unit uses the usage rate and the number of accesses as the first load information, and for each storage unit, the number of accesses of the storage unit and an upper limit value set for the storage unit The memory according to claim 5, wherein a difference is calculated, one of the plurality of storage units is selected based on the calculation result, and the upper limit value is adjusted for each of the storage units based on the usage rate. apparatus.   The management unit repeatedly acquires the usage rate, counts the number of times the usage rate has continuously exceeded a predetermined threshold, and lowers the upper limit value when the number of times exceeds a specified value. The storage device described in 1. A plurality of the reading unit,
The management unit selects one of the plurality of reading units based on second load information indicating a load status of each reading unit,
The storage device according to claim 1, wherein the selected reading unit reads the data from the selected storage unit. Based on the first load information indicating the load status of each of a plurality of storage units each storing the same data, select one of the plurality of storage units,
A data reading method for reading the data from the selected storage unit. Computer
A plurality of storage units each storing the same data;
A management unit that selects one of the plurality of storage units based on first load information indicating a load status of each storage unit;
A program for functioning as a storage device comprising: a reading unit that reads out the data from the selected storage unit.

Images