JP2003296153A - Storage system and program therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the response to an access from a user to data in a storage system to enhance the power efficiency by performing a load distribution or integration according to load fluctuation. <P>SOLUTION: This storage system 100 is provided with a storage processing part 1 for managing the storage position of data, a temporary storage part 2 for temporarily storing data, and an accumulation part 3 for accumulating data, which are dispersively arranged through a network 4, wherein the data requested from a client 5 connected to the network 4 is read from the temporary storage part 2 or accumulation part 3 and transferred to the client 5. This system is further provided with a measuring means 11 for measuring the load of the storage processing part 1, and load distributing means 10, 12, 13, 14, 20 and 30 for distributing the load with another storage processing part 1 according to the measurement result of the measuring means 11. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a storage system for storing data on a network and a program therefor, and in particular, by distributing the load according to the load fluctuation of the storage that stores the data, The present invention relates to a storage system that improves response to access from a user and is highly power efficient, and a program therefor.

[0002]

2. Description of the Related Art IT (Information Technology) in recent years
The development of is exploding the digitization of various data. Along with this, the role of the storage system that processes and stores these data is becoming more and more important. In such a system, for example, NAS (Networ
k Attached Storage). This system is a large-capacity disk, a processing unit that manages document and image data as files, and communication that receives an access signal to a file from an external user via a network and sends the file to the external user. And a department.

However, as an operation mode of these storage systems, an IDC (Internet D) used by a company or an individual is not prepared by the company by itself.
ataCenter) and ASP (Application Service Provide)
r) is widespread. Since the IDC and ASP are used by a large number of users via the network, the load fluctuates greatly, and the response to the access from the user to the storage is deteriorated when the load is high.

In order to improve this response, load distribution is performed. Typical examples thereof include (i) a plurality of network file servers such as NAS installed, and (ii) a plurality of processing units in the storage system.

[0005]

(I) In the case where a plurality of network file servers are installed, the load may be concentrated on a specific file in a situation where load distribution is usually required. A file server is added. However, the heavily loaded file server has to copy the file having the concentrated load to the added file server, which is troublesome and the response to the user's access to the storage is delayed accordingly.

(Ii) In the case where a plurality of processing units are installed in the storage system, data always flows to the plurality of processing units via the communication unit, so that the communication unit is congested at the time of high load when access to the system is concentrated. However, there is still a problem that the response to the access from the user is delayed.

Therefore, the present invention solves the above problem and absorbs the load fluctuation by distributing the load within the system according to the above load fluctuation, and the data stored in the system from the user can be absorbed. It is an object of the present invention to provide a storage system that improves response to access and a program therefor.

Another object of the present invention is to improve the power efficiency of a storage system having a plurality of storages to save energy.

[0009]

A storage system according to the present invention that achieves the above object comprises a storage processing unit for managing the storage location of data and a storage unit for storing data, which are distributed over a network. In the storage system that retrieves the data requested by the client connected to the network from the storage unit and transfers the data to the client, the measuring unit that measures the load of the storage processing unit and the measurement result of the measuring unit And a load distribution means for distributing the load to and from another storage processing unit.

In the above storage system, the storage unit includes a temporary storage unit for temporarily storing data and a storage unit for storing data.

In the above storage system, when the load of the storage processing unit exceeds the upper limit value as a result of the measurement by the measuring means, the storage processing unit searches for another low load storage processing unit and is searched. The load is distributed to the storage processing unit.

In the above storage system, the storage processing unit searches for another low-load storage processing unit when the load of the storage processing unit falls below the lower limit value as a result of the measurement by the measuring means, and is searched. The load is integrated into the storage processing unit.

[0013]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

FIG. 1 is a schematic configuration diagram of a storage system of an embodiment according to the present invention. A storage system 100 shown in FIG. 1 as a whole includes a storage processing unit 1 that manages data storage positions, a data temporary storage unit 2 that temporarily stores data in a small-capacity but high-speed storage device such as a cache memory, and data such as a hard disk. It has a non-volatile, large-capacity but low-speed storage device for storing data.
A plurality of storage processing units 1, data temporary storage units 2 and data storage units 3 are provided (three are shown in FIG. 1, but different numbers may be provided), such as interconnect (INFINI band) and LAN. They are distributed and arranged via a high-speed network 4.

The storage system 100 comprises a measuring unit 11 for measuring the load of the storage processing unit 1 and a measuring unit 1.
Load balancing means 10, 12, 13, 14 for balancing the load with other storage processing units 1 according to the measurement result of 1.
20 and 30 are provided.

The storage processing unit 1 receives a request for transmission of data stored in the data temporary storage unit 2 or the data storage unit 3 from any one of the plurality of clients 5 connected to the network 4. A control is performed to search where the data is temporarily stored in the data temporary storage unit 2 or the data storage unit 3 that stores the data, and to transmit the searched data to the client 5.

The switch section 6 is installed in the network 4 and connects a transmission line of a command signal for accessing data from the client 5 to the storage processing section 1,
A switch for connecting a data transmission line from the data temporary storage unit 2 or the data storage unit 3 to the client 5. In the embodiment shown in FIG. 1, the switch 6 is provided inside the network 4, but the switch 6 may be provided outside the network 4.

The storage processing unit 1 has a virtual storage management unit 10, a processing load measuring unit 11, a processing division / integration determination unit 12, a processing node selection unit 13, a processing load notification unit 14, a control unit 15 and a communication unit 16. In addition, it has a virtual storage management table 20 and a processing load measurement table 30.

The virtual storage management unit 10 manages where the designated data address (virtual storage address) is in the real storage (the temporary data storage unit 2 or the data storage unit 3).

The processing load measuring unit 11 has a processing load for each virtual storage (for example, the number of accesses for each storage).
To measure.

The process division / integration determination unit 12 processes the virtual storage based on the value of the process load measurement table 30 according to whether the process load value exceeds the upper limit value, falls below the lower limit value, or is in between. Determine whether to split or integrate.

The processing node selection unit 13 inquires of the other storage processing units about the processing load and selects the storage processing unit to which the load is distributed. In the present invention, the storage processing unit with the lowest load is selected, and the load of the high-load storage processing unit is shared there.

The processing load notifying unit 14 notifies the state of the processing load in response to an inquiry from another storage processing unit.

The control unit 15 includes the virtual storage management unit 1
0, processing load measurement unit 11, processing division / integration determination unit 12,
The processing node selection unit 13 and the processing load notification unit 14 are controlled. In addition, the control unit 15 shifts the CPU of the storage processing unit 1 from the normal mode to the power saving mode when the storage processing unit 1 does not need to share the processing load. The CPU returns to the normal mode at an interrupt due to some processing request after entering a power saving mode, that is, a sleep state in which power consumption is lower than in the normal mode.

The communication unit 16 communicates with other storage processing units, data temporary storage units, data storage units and clients.

The virtual storage management table 20 is a table for storing the correspondence between virtual storage and real storage (see FIGS. 5, 6, and 8). This table has
For each virtual storage, virtual storage number, storage processing unit number, storage location type (data temporary storage unit such as cache memory or data storage unit such as disk), storage location number for identifying the installation location of the data storage unit, virtual Start LBA (Logical Block Address), virtual end LBA
Data of the actual start LBA and the actual end LBA are stored.

The processing load measurement table 30 is a table for storing the value of the processing load for each virtual storage measured by the processing load measuring unit 11, and is provided to determine whether to divide or integrate the virtual storage processing. 3 is a table that stores virtual storage numbers, processing load values, upper limit thresholds, and lower limit thresholds for each virtual storage (see FIG. 3). Next, the operation of load distribution by the storage processing unit 1 shown in FIG. 1 will be described below.

FIG. 2 is a flow chart showing the operation of the storage processing unit, and FIG. 3 is a diagram showing a concrete example of the processing load measurement table of the storage processing unit. In the following flow charts, the numbers following S are step numbers.

First, in step S10, the processing load value for each virtual storage is measured. As the measured value of processing load,
For example, it is obtained from the number of accesses processed by the virtual storage in a fixed time (for example, 1 minute), the temperature of the storage processing unit 1 (measured by a thermistor or the like) or the power consumption (load current is measured and calculated by supply voltage × load current).

In step S11, the data in the processing load measurement table (see FIG. 3) is updated with the processing load value measured in step S10.

In step S12, it is determined whether or not there is a virtual storage whose processing load value measured in step S10 exceeds the upper limit threshold value. If the determination result is YES, the process proceeds to step S13, which will be described later in step S13. The virtual storage division process is executed, and if the determination result is NO, the process proceeds to step S14.

In step S14, it is determined whether or not there is a virtual storage whose processing load value measured in step S10 is below the lower limit threshold value. If the determination result is YES, the process proceeds to step S15, which will be described later in step S15. The virtual storage integration process is executed, and if the determination result is NO, the process proceeds to step S16.

At step S16, after the rest for the designated time, the process returns to step S10. Thereafter, step S10
Repeat ~ 16.

Next, the virtual storage division processing / integration processing will be described. Before that, the virtual storage in load distribution by the storage system of the present invention will be defined. Each virtual storage (number 0, 1, 2,
, ...) is defined as the master storage processing unit 1 (for example, number 0) that is in charge of control, and the other units are designated as slave storage processing units 1 (numbers 1, 2, ...).

Next, the operation of the virtual storage division processing (step S13) shown in FIG. 2 will be described below.

FIG. 4 is a flow chart of the virtual storage dividing process in FIG. 2, and FIGS. 5 and 6 are diagrams showing first and second concrete examples in which the data in the virtual storage management table is changed by the dividing process. Yes, (A) is a diagram showing a table before division processing, and (B) is a diagram showing a table after division processing.

(I) When the load of the master storage processing unit exceeds the upper limit threshold First, in step S20, the processing node selecting unit 13 of the master storage processing unit 1 (No. 0) makes the slave storage processing unit 1 (No. 1). Inquire about the load. In response to this inquiry, the processing load notification unit 14 of the slave storage processing unit 1 notifies the processing load value stored in the processing load measurement table 30.

If there are other storage processing units, they are regarded as slave storage processing units and the same processing is performed.

In step S21, the master storage processing unit 1 (No. 0) selects the slave storage processing unit 1 (No. 1) having the lowest processing load value, based on the notified processing load value.

In step S22, the master storage processing unit 1 (No. 0) divides the virtual storage process (No. 1) currently in charge of the selected slave storage processing unit 1 (No. 1) into the virtual storage management table. 20 rewrites. This is shown in FIG. As can be seen from the virtual storage management table 20 before the division processing shown in FIG. 5A, the processing of the virtual storage (number 1) is currently being executed by the master storage processing unit 1 (number 0). Slave storage processing unit 1
It shall be divided into (number 1). The virtual storage management table 20 after this division processing is shown in FIG.
Shown in. In the example shown in FIG. 5B, the virtual storage (number 1) is divided at the center, but the division of the storage processing units may be different, for example, 1: 2.

In step S23, the master storage processing unit 1 (number 0) requests the slave storage processing unit 1 (number 1) selected in step S21 according to the virtual storage management table 20 after the division. The slave storage processing unit 1 (number 1) receives this and executes the requested processing.

(Ii) When the load of the slave storage processing unit exceeds the upper limit threshold First, before step S20, the slave storage processing unit 1 (No. 1) has the processing load stored in the virtual storage management table 20. Recognize that the value has exceeded the upper threshold. Then, the processing load notification unit 14 of the slave storage processing unit 1 (No. 1) notifies the master storage processing unit 1 (No. 0) that its own load exceeds the upper limit value.

Then, in step S20, the processing node selection unit 13 of the master storage processing unit 1 (No. 0) inquires the slave storage processing unit 1 (No. 2) about the load. The processing load notification unit 14 of the slave storage processing unit 1 (number 2) receives this inquiry and notifies the processing load value stored in the processing load measurement table 30.

If there are other storage processing units, they are regarded as slave storage processing units and the same processing is performed.

In step S21, the master storage processing unit 1 (No. 0) selects the slave storage processing unit 1 (No. 2) having the lowest processing load value, based on the notified processing load value.

In step S22, the master storage processing unit 1 (No. 0) sends the request to the selected slave storage processing unit 1 (No. 2) to the slave storage processing unit 1 (No. 2).
The virtual storage management table 20 is rewritten in order to divide the virtual storage process (number 1) which is in charge of (number 1). This is shown in FIG. As can be seen from the virtual storage management table 20 before the division processing shown in FIG. 6A, the processing of the virtual storage (number 1) is currently being executed by the slave storage processing unit 1 (number 1). It is assumed that the processing is divided into the slave storage processing unit 1 (number 2). The virtual storage management table 20 after this division processing is shown in FIG. In the example shown in FIG. 6B, the virtual storage (number 1) is divided into different allocations of the storage processing units, such as 2: 1.

In step S23, the master storage processing unit 1 (number 0) requests the slave storage processing unit 1 (number 2) selected in step S21 according to the virtual storage management table 20 after the division. The slave storage processing unit 1 (number 2) receives this and executes the requested processing.

The operation of the virtual storage integration process (step S15) shown in FIG. 2 will be described below.

(Iii) When the load of the slave storage processing unit falls below the lower limit threshold value FIG. 7 is a flowchart of the virtual storage integration processing in FIG. 2, and FIG. 8 changes the data in the virtual storage management table by the integration processing. FIG. 4A is a diagram showing a specific example of FIG.
(B) is a diagram showing a table after the integration processing.

First, before step S30, the slave storage processing unit 1 (number 1) recognizes that the processing load value stored in the virtual storage management table 20 has fallen below the lower limit threshold. Then, the processing load notification unit 14 of the slave storage processing unit 1 (No. 1) notifies the master storage processing unit 1 (No. 0) that its own load is below the lower limit value.

Then, in step S30, the processing node selection unit 13 of the master storage processing unit 1 (No. 0) inquires the slave storage processing unit 1 (No. 2) about the load. In response to this inquiry, the processing load notification unit 14 of the slave storage processing unit 1 (number 2) notifies the processing load value stored in the processing load measurement table 30.

In step S31, the master storage processing unit 1 (No. 0) selects the slave storage processing unit 1 (No. 2) having the lowest processing load value here based on the notified processing load value and its own processing load value. select.

In step S32, the master storage processing unit 1 (No. 0) rewrites the virtual storage management table 20 in order to integrate the virtual storage process (No. 1) into the selected slave storage processing unit 1 (No. 2). . The situation is shown in FIG. As can be seen from the virtual storage management table before the integrated processing shown in FIG. 8A, the processing of the virtual storage (number 1) is currently executed by the master and slave storage processing units 1 (numbers 0 and 1 and 2). However, of these, the slave storage processing unit 1
Since the processing load of (No. 1) is below the lower threshold,
This virtual storage process (number 1) is integrated into the slave storage processing unit 1 (number 2). The virtual storage management table after this integration processing is shown in FIG.

In step S33, the master storage processing unit 1 (number 0) follows the integrated virtual storage management table and the virtual storage (number 0) after integration into the slave storage processing unit 1 (number 2) selected in step S31. Request the processing of 1).

After integration of the processing loads, if the slave storage processing unit 1 (No. 1) is not executing other virtual storage processing, the control unit 15 of the slave storage processing unit 1 (No. 1) gives an instruction. By shifting the CPU of the slave storage processing unit 1 (number 1) to the power saving mode, it is possible to reduce the power consumption of the entire storage system having the plurality of storage processing units 1.

In the above-described embodiment, the master storage processing unit and the slave storage processing unit are described as being fixed for each virtual storage, but the present invention does not define a master / slave for each virtual storage, but rather a virtual storage It is also possible to adopt a configuration in which each storage processing unit that is processing determines its own processing load and requests another storage processing unit to divide / integrate the processing.

(Supplementary Note 1) A storage processing unit that manages the storage location of data and a storage unit that stores data are provided, and these are distributed over a network to store data requested by clients connected to the network. In a storage system that retrieves from the storage unit and transfers it to the client, a load that balances the load between a measuring unit that measures the load of the storage processing unit and another storage processing unit according to the measurement result of the measuring unit. A storage system comprising: a distribution unit.

(Supplementary Note 2) The storage system according to Supplementary Note 1, wherein the storage unit includes a temporary storage unit for temporarily storing data and a storage unit for storing data.

(Supplementary Note 3) When the load of the storage processing unit exceeds the upper limit value as a result of the measurement by the measuring means, the storage processing unit searches for another low-load storage processing unit and is searched. The storage system according to Appendix 1, which distributes the load to a storage processing unit.

(Supplementary Note 4) When the load of the storage processing unit falls below the lower limit value as a result of the measurement by the measuring means, the storage processing unit searches for another low-load storage processing unit and is searched. The storage system according to appendix 1 or 2, wherein the load is integrated in a storage processing unit.

(Supplementary Note 5) The measuring means measures the processing load as a frequency of access to the storage processing unit,
The storage system according to any one of appendices 1 to 4.

(Supplementary Note 6) The storage system according to any one of Supplementary Notes 1 to 4, wherein the measuring means measures the processing load by measuring the temperature of the storage processing unit.

(Supplementary Note 7) The storage system according to any one of Supplementary Notes 1 to 4, wherein the measuring unit measures the power supplied to the storage processing unit from the processing load.

(Supplementary Note 8) Among the storage processing units,
Any one of appendices 1 to 7, further comprising: a control unit that shifts a storage processing unit that has no load of a processing load due to load integration to a power saving state, thereby suppressing power consumption of the storage system and improving energy efficiency. Storage system described in one.

(Supplementary Note 9) A storage processing unit for managing the storage location of data, a temporary storage unit for temporarily storing data, and a storage unit for accumulating data are provided, and these are distributed over the network and connected to the network. In the storage control method of fetching the data requested by the client from the temporary storage unit or the storage unit and transferring the data to the client, the load of the storage processing unit is measured, and another storage process is performed according to the measurement result. A storage control method comprising: each step of distributing a load to and from a unit.

(Supplementary Note 10) A storage processing unit for managing the storage location of data, a temporary storage unit for temporarily storing data, and a storage unit for accumulating data are provided, and these are distributed over the network and connected to the network. In the storage system that retrieves the data requested by the client from the temporary storage unit or the storage unit and transfers the data to the client, the computer measures the load of the storage processing unit, and according to the measurement result,
A program for executing each step of load balancing with another storage processing unit.

[0067]

As described above, according to the present invention,
Since the load is distributed according to the load fluctuation of the storage that stores the data, it is possible to provide the storage system and the program for improving the response to the access from the user to the data stored in the storage system.

Further, according to the present invention, normal power is supplied only to the storage processing unit that is actually requested to share the load, and almost no power is supplied to the storage processing unit that is not required to share the load. Since it is good, the power consumption of the storage system is suppressed, a storage system with high energy efficiency can be provided, and it can contribute to energy saving.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a storage system of an embodiment according to the present invention.

FIG. 2 is a flowchart showing an operation of a storage processing unit.

FIG. 3 is a diagram showing a specific example of a processing load measurement table of a storage processing unit.

FIG. 4 is a flowchart of virtual storage division processing in FIG.

FIG. 5 is a diagram showing a first specific example in which data in a virtual storage management table is changed by division processing;
FIG. 9A is a diagram showing a table before division processing, and FIG.
FIG. 6 is a diagram showing a table after division processing.

FIG. 6 is a diagram showing a second specific example in which the data in the virtual storage management table is changed by the division processing;
FIG. 9A is a diagram showing a table before division processing, and FIG.
FIG. 6 is a diagram showing a table after division processing.

FIG. 7 is a flowchart of a virtual storage integration process in FIG.

FIG. 8 is a diagram showing a specific example in which data in a virtual storage management table is changed by an integration process, FIG. 8A is a diagram showing a table before the integration process, and FIG. 8B is a table after the integration process. FIG.

[Explanation of symbols]

1 ... Storage processing unit 2 ... Temporary data storage 3 ... Data storage unit 4 ... Network 5 ... Client 6 ... Switch section 10 ... Virtual storage management unit 11 ... Processing load measurement unit 12 ... Processing division / integration determination unit 13 ... Processing node selection unit 14 ... Processing load notification unit 15 ... Control unit 16 ... Communication unit 20 ... Virtual storage management table 30 ... Processing load measurement table 100 ... Storage system

Claims (5)

[Claims] 1. A storage processing unit for managing a storage location of data and a storage unit for storing data, which are distributed over a network and store the data requested by a client connected to the network. In a storage system that takes out from a storage unit and transfers it to the client, a load balancing unit that balances load between a storage unit that measures a load of the storage processing unit and another storage processing unit according to a measurement result of the measuring unit. A storage system comprising: 2. The storage system according to claim 1, wherein the storage unit includes a temporary storage unit that temporarily stores data and a storage unit that stores the data. 3. The storage processing unit searches for another low-load storage processing unit when the load of the storage processing unit exceeds an upper limit value as a result of the measurement by the measuring unit,
The storage system according to claim 1, wherein the load is distributed to the retrieved storage processing units. 4. The storage processing unit searches for another low-load storage processing unit when the load of the storage processing unit falls below a lower limit value as a result of the measurement by the measuring unit, and the searched storage processing is performed. The storage system according to claim 1, wherein the load is integrated into a unit. 5. A storage processing unit for managing a storage position of data, a temporary storage unit for temporarily storing data, and a storage unit for storing data, which are distributed over a network and connected to the network. In a storage system that retrieves data requested by a client from the temporary storage unit or the storage unit and transfers the data to the client, a computer measures the load of the storage processing unit, and another storage process is performed according to the measurement result. A program characterized by causing each step to be executed so that the load is distributed between the parts.