JP5924482B2 - Storage device - Google Patents

Description

  Some embodiments according to the present invention relate to a storage apparatus.

  In recent years, a storage device that stores large-capacity data including a disk drive such as a plurality of HDDs (Hard Disk Drives) has become widespread (see, for example, Patent Document 1). Patent Document 1 discloses an apparatus for storing a data block obtained by dividing data requested for writing into blocks of a predetermined size in a buffer area in the disk device and transferring the data stored in the buffer area to the storage area. It is disclosed.

JP 07-210330 A

  Here, many storage apparatuses include a cache memory that temporarily stores data to be written, in order to improve data read performance and write performance. However, since the cache memory has a limited storage capacity, when it receives a write request for large-capacity data such as multimedia data (for example, moving image data), the cache capacity is compressed.

  Some aspects of the present invention have been made in view of the above-described problems, and an object of the present invention is to provide a storage apparatus capable of performing a suitable writing process depending on the situation.

  The storage device according to the present invention includes a primary storage device that temporarily stores data written from a host device, a first storage device that has a larger storage capacity than the primary storage device, and a storage device that is larger than the first storage device. Depending on the entropy of the data stored in the primary storage device and the second storage device having a large storage capacity and slow reading speed, the data is stored in either the first storage device or the second storage device. And a control unit for recording.

  In the present invention, “part”, “means”, and “apparatus” do not simply mean physical means, but the functions of the “part”, “means”, and “apparatus” are realized by software. This includes cases where In addition, even if the functions of one “unit”, “means”, and “device” are realized by two or more physical means or devices, two or more “parts”, “means”, and “devices” The function may be realized by one physical means or apparatus.

  According to the present invention, it is possible to provide a storage apparatus capable of performing a suitable writing process depending on the situation.

1 is a functional block diagram showing a schematic configuration of a recording system in which a disk device according to an embodiment of the present invention is used. 3 is a flowchart showing a flow of processing of the disk device shown in FIG. 1. FIG. 2 is a diagram for explaining a specific example of processing of the disk device shown in FIG. 1.

  Embodiments of the present invention will be described below. In the following description and the description of the drawings to be referred to, the same or similar components are denoted by the same or similar reference numerals.

(Outline of 1 embodiment)
Generally, a storage device is equipped with a plurality of disk drives, for example, and is used for storing various data. In recent years, disk drives have become larger in capacity and lower in price due to an increase in recording density, and in recent years, those of several hundred gigabytes to several terabytes per disk drive have become mainstream. In addition, since the disk drive itself has been reduced in size, the storage device is reduced in space and capacity. Accordingly, large-size data such as multimedia data including moving image data can be handled on the computer.

  The multimedia data such as moving image data, audio data, and still image data has been increased in capacity, and the compression technology has been improved. For example, in video compression, MPEG-4 or H.264 is used. Conversion technology (encoding technology) that boasts a high compression rate while maintaining fine image quality, such as H.264, is becoming widespread. For this reason, when the multimedia data is stored in the storage device, it is often stored in the storage after being subjected to some compression conversion processing on the host device side, rather than being stored uncompressed.

On the other hand, in order to improve data read performance (read performance) and write performance (write performance), a storage device is often provided with a cache memory for temporarily storing target data for a write request. In many cases, a DRAM (Dynamic Random Access Memory) capable of high-speed access is used as the cache memory. Here, when the storage capacity of the cache memory is increased, the read / write performance can be improved. However, since the DRAM is expensive, the cost of the cache memory increases. Therefore, it is conceivable to use a memory such as a high-speed but small-capacity DRAM as a primary cache, and use a memory or a flash SSD (Solid State Drive) that is slower than a DRAM but has a large capacity and a low price as a secondary cache. As described above, when the secondary cache is used, as a data storage method, for example, when the data on the primary cache overflows and a cache write to be written to the secondary cache becomes necessary, all the overflowed data is deleted. It is possible to store in the secondary cache.
However, such mounting causes the following problems.

  The first problem is that when a large-volume data write request such as several tens of megabytes to several gigabytes is received, such as multimedia data, if all the data is stored in the secondary cache, the secondary cache There is a possibility of squeezing the capacity. This problem becomes particularly noticeable when the storage device is used as an archive of multimedia data, for example. In this regard, for example, in NAS (Network Attached Storage), a technique for performing secondary cache placement control by referring to file characteristic information has been put into practical use. In the case of SAN storage, a write request is a file Since it is performed not in units but in block data units smaller than that, it is difficult to control placement in the secondary cache based on information such as a file header.

  The second problem is that when a flash SSD is used for a secondary cache, the flash SSD has a write life limit, so that data cannot be frequently written as a cache.

  Therefore, in the disk device 100 that is the storage device according to the present embodiment, when the block data is ejected from the primary cache memory used as the primary cache in order to efficiently perform cache control of the mounted secondary cache. Then, entropy is calculated for the block data. As a result, block data whose entropy exceeds a certain threshold is directly written to the disk drive, and other block data is written to the secondary cache memory. In general, compressed multimedia data has a high entropy, and this implementation limits the writing of multimedia data to the secondary cache, thus reducing the capacity of the secondary cache memory. In addition, when the flash SSD is used for the secondary cache memory, the life of the flash SSD is not shortened. That is, the cache control of the secondary cache memory mounted on the storage device can be suitably performed.

(2 Functional configuration of the disk device 100)
Hereinafter, a schematic configuration of the disk device 100 as a storage device according to the embodiment will be described with reference to FIG. FIG. 1 is a functional block diagram showing a configuration of a recording system in which the disk device 100 is used.

  In this recording system, the disk device 100 and the host device 200 (host devices 200A to 200M are collectively referred to as “host device 200”. Hereinafter, the controller 101, the host interface 103, the primary cache unit 105, and the secondary cache unit 107 are called. The same applies to the disk interface 109, the disk drive 150, and the like.), And the disk device 100 and the host device 200 are connected via a SAN (Storage Area Network) 300. As an example of SAN, FC (Fibre Channel) -SAN, iSCSI-SAN, etc. can be considered, for example.

  The host device 200 is a device that uses the disk device 100, and is, for example, a file server or a database server. In this embodiment, a case is considered in which a plurality of host devices 200A to 200M can access the disk device 100. However, the present invention is not limited to this, and for example, a single device may be used.

  In response to a write request (write request) or read request (read request) output from the host device 200, the disk device 100 performs a data write process and a data read process. The disk device 100 includes a controller 101 including a host interface 103, a primary cache unit 105, a secondary cache unit 107, and a disk interface 109, and one or more disk drives 150. The disk interface 109 and the disk drive 150 are connected by SAS (Serial Attached SCSI), SATA (Serial ATA), or the like. The controller 101 in the example of the disk device 100 in FIG. 1 is configured in a cluster configuration with two controllers 101A and 101B so that the operation can be continued even if a failure occurs on one side.

  The host interface 103 is an interface for receiving read (read) and write (write) requests from the host device 200 using the disk device 100. As a specific example of the host interface 103, for example, an FC or iSCSI host bus adapter (HBA) can be considered.

  The primary cache unit 105 includes a primary cache control unit 111 and a primary cache storage unit 115. The primary cache storage unit 115 temporarily stores data according to a write request from the host device 200, or temporarily stores data read according to a read request. The primary cache storage unit 115 is a memory that can be accessed at high speed on both the read and write sides, such as a DRAM.

  The primary cache control unit 111 is a module that controls writing and reading of block data in the primary cache storage unit 115 to and from the host device 200, the secondary cache unit 107, and the disk drive 150. Here, the primary cache control unit 111 includes an entropy calculation unit 113. The entropy calculation unit 113 can calculate the entropy of the block data stored in the primary cache storage unit 115, and the primary cache control unit 111 responds to the calculation result by the primary cache control unit 111. It is determined whether the data write destination is the secondary cache storage unit 119 or the disk drive 150.

Here, entropy is a measure that generally represents disorder, ambiguity, and uncertainty, but in the field of information science, it can be described as a measure that indicates the degree of redundancy of information. Data with higher redundancy has lower entropy, and data with lower redundancy, that is, data that is messy and difficult to compress, has higher entropy. As described above, multimedia data such as moving image data is generally written on the disk device 100 after being compressed on the host device 200 side, but a codec used for the recent compression processing (encoding) of multimedia data. Since the compression performance (including the encoder) is very high, even if another compression algorithm is further applied to the compressed data, the compression performance is limited. That is, multimedia data after compression (after encoding) generally has a low compressibility rate (compression rate that can be further compressed) and a high entropy. Therefore, the level of entropy can be approximated to the level of compressibility, and the level of entropy can be used to determine whether or not it is multimedia data (after compression). It can be regarded as a scale. Hereinafter, description will be made assuming that the entropy calculation unit 113 calculates the compressibility rate of the target block data as a scale indicating entropy.
The compressibility ratio can be obtained by, for example, 100 − {(size of block data after compression) ÷ (size of block data before compression) × 100}.

  The secondary cache unit 107 includes a secondary cache control unit 117 and a secondary cache storage unit 119, and is disposed between the primary cache unit 105 and the disk drive 150. Therefore, by accumulating frequently used data on the secondary cache unit 107, access to the disk drive 150 that is slower than the secondary cache unit 107 is reduced, and overall write / read performance is improved. It becomes possible. The secondary cache storage unit 119 can be a DRAM as well as the primary cache storage unit 115, and can also be a flash SSD. Here, the secondary cache storage unit 119 has a larger capacity than the primary cache storage unit 115.

  The secondary cache control unit 117 is a module that controls writing and reading of block data on the secondary cache storage unit 119 to and from the primary cache unit 105 and the disk drive 150.

  The disk interface 109 is an interface for performing writing and reading processing from the controller 101 to the disk drive 150. The disk drive 150 is a storage device such as an HDD (Hard Disk Drive). In the example of the disk device 100 of FIG. 1, a plurality of disk drives 150A to 150N are prepared as the disk drive 150, and these constitute a disk array such as a RAID (Redundant Arrays of Inexpensive Disks), for example. Also good. Further, the number of disk drives 150 may be one. The disk drive 150 has a lower access speed than the secondary cache storage unit 119, but has a larger capacity.

(3 Process flow)
The processing flow of the disk device 100 will be described below with reference to FIG. FIG. 2 is a flowchart showing a processing flow of the disk device 100. The processing of FIG. 2 is performed by the host device 200 on the primary cache storage unit 115 in a state where all the cache pages on the primary cache storage unit 115 of the primary cache unit 105 are used (when there is no empty cache page). This is a flow of processing when there is an access request for block data that does not exist in.

  Each processing step to be described later can be executed in any order or in parallel as long as there is no contradiction in processing contents, and other steps can be added between the processing steps. good. Further, a step described as a single step for convenience can be executed by being divided into a plurality of steps, and a step described as being divided into a plurality of steps for convenience can be executed as one step.

  First, the primary cache control unit 111 of the primary cache unit 105 selects one block data stored in any cache page on the primary cache storage unit 115 (S201), and from the primary cache storage unit 115 The process of creating an empty cache page in the primary cache storage unit 115 is started. Here, the cache page storing the block data to be ejected is accessed in the farthest past based on the access order of the cache pages recorded in the management information list managed in the primary cache control unit 111, for example. The page number of the cache page storing the block data can be obtained and used as the page number of the target cache page. That is, the primary cache unit 105 selects a cache page to be discarded (ejected) based on an LRU (Least Recently Used) method. However, other cache page selection methods may be used. For example, an LFU (Least Frequently Used) that is a target for discarding (pushing out) a cache page storing block data with the lowest frequency of reference. A method may be used.

  After selecting block data to be purged (discarded), the primary cache control unit 111 uses the entropy calculator 113 to calculate the entropy (compressible rate) of the block data to be discarded (S203). Then, the primary cache control unit 111 determines whether or not the entropy is sufficiently high (whether or not the compressibility rate is equal to or less than a threshold value) (S205).

  Here, as described above, the entropy can be measured based on the compressibility rate. Since the compressed (encoded) multimedia data has a low compressibility rate, it has a high entropy. On the other hand, non-compressed data such as text data has a high compressibility rate, and is considered to have low entropy. That is, in S203 and S205, the primary cache control unit 111 obtains the compressibility rate of the target block data (S203), and whether the entropy is high depending on whether the compressibility rate is below the threshold (S205). Yes) It can be determined whether it is low (No in S205).

  If the entropy is high (Yes in S205), the block data is considered to be a part of the multimedia data and is likely to have a large capacity. Therefore, the primary cache unit 105 outputs the write request to the disk drive 150 using the disk interface 109, transmits the target block data (S207), and waits until a write completion response is received from the disk drive 150 (S209). ).

  On the other hand, when the entropy of the target block data is low (No in S205), the block data is considered to be data other than multimedia data. Therefore, the primary cache unit 105 transmits a write request to the secondary cache control unit 117 of the secondary cache unit 107 and transmits block data (S211), and then receives a write completion response from the secondary cache unit 107. (S213).

(4 Specific examples of processing)
Hereinafter, a specific example of the processing of the disk device 100 will be described with reference to FIG. In the example of FIG. 3, the processing in the controller 101A is illustrated. However, since the processing is the same in the controller 101B, each configuration of the controller 101A, the primary cache unit 105A, and the like will be described below. It will be described as “controller 101” and “primary cache unit 105”.

  Here, it is assumed that there are 4 kilobytes of block data A to C in the primary cache storage unit 115 and all the cache pages on the primary cache storage unit 115 are used. At this time, when a write request is transmitted from the host device 200, the primary cache control unit 111 of the disk device 100 selects block data to be ejected ((1) in FIG. 3, corresponding to S201 in FIG. 2). Here, it is assumed that block data C is selected as a target, and this data is a part of text data (uncompressed data).

  Next, the entropy calculation unit 113 calculates entropy for the selected block data C ((2) in FIG. 3, corresponding to S203 in FIG. 2). Here, for example, an LZSS compression algorithm is used for entropy calculation, and the threshold of the compressibility ratio is 3%. Here, the reason why the threshold of the compressibility ratio is set to 3% is that, if header information of about several hundred bytes is included in the block data, even the multimedia data can be compressed. Because it is normal.

  In the case of text data, the compressibility ratio is often high, and it is assumed here that the block data C can be compressed by 30%. In this case, since the compressibility ratio is 30% and greatly exceeds the threshold value 3%, the primary cache control unit 111A determines that the entropy is low (No in S205 in FIG. 2), and the secondary cache unit 107 A block data write request to the secondary cache storage unit 119 is transmitted to the secondary cache control unit 117 ((3) in FIG. 3, corresponding to S211 in FIG. 2).

  When the writing of the block data in the secondary cache unit 107 is completed, the secondary cache control unit 117 transmits a write completion to the primary cache control unit 111 ((4) in FIG. 3). When the primary cache control unit 111 receives the write completion (corresponding to Yes in S213 in FIG. 2), the process ends.

(5 Effects of this embodiment)
As described above, according to the present embodiment, it is possible to avoid recording multimedia data such as moving image data and audio data, which often have a huge size, in the secondary cache unit 107. As a result, the used capacity of the secondary cache storage unit 119 is not suddenly compressed, and the number of data stores / restores in the secondary cache unit 107 can be reduced.

  Here, in general, the disk access 150 such as HDD is slower in data access than the secondary cache unit 107 such as flash SSD, but multimedia data such as moving image data and audio data is sequentially accessed. It is normal. Here, when performing a sequential read, there is no significant difference in performance between the secondary cache unit 107 such as a flash SSD and the disk drive 150 such as an HDD. Therefore, even if the primary cache control unit 111 directly arranges the multimedia data (without placing it in the secondary cache unit 107) in the disk drive 150, no significant performance degradation occurs. Further, if multimedia data once arranged in the disk drive 150 is frequently read, the multimedia data may be rearranged in the secondary cache unit 107.

  Further, when a flash SSD is used for the secondary cache unit 107, the number of cache writes to the secondary cache unit 107 can be suppressed, so that the life of the flash SSD can be extended.

(6 Other embodiments)
In the above embodiment, the compressibility ratio is used for entropy calculation by the entropy calculation unit 113. Accordingly, it is conceivable to further improve the utilization efficiency of the disk drive 150 using the information. Hereinafter, the method will be described.

  When entropy calculation is performed by the entropy calculation unit 113 (S203 in FIG. 2), when the compressibility rate using a predetermined compression algorithm is high (entropy is low), when writing block data to the secondary cache unit 107 In addition, a flag indicating that the compressibility rate of the block data of the page number of the cache page of the written data is high is added to the management information list managed by the secondary cache control unit 117.

  When the block data on the secondary cache unit 107 is to be purged and written to the disk drive 150, a flag indicating that the block data has a high compressibility rate is displayed. If it is given, the data is written to the disk drive 150 after the target block data is compressed. At this time, a flag indicating compression is added to the management table of the disk drive 150.

When reading the data on the disk drive 150 to the primary cache unit 105 or the secondary cache unit 107, the primary cache control unit 111 or the secondary cache control unit 117 refers to the management table of the disk drive 150 and compresses it. Is attached to the primary cache storage unit 115 or the secondary cache storage unit 119 after decompressing the data.
By mounting in this way, the storage capacity of the data written to the disk drive 150 is reduced, and the utilization efficiency can be increased.

(7 Appendix)
Note that the configurations of the above-described embodiments may be combined or some of the components may be replaced. The configuration of the present invention is not limited to the above-described embodiment, and various modifications may be made without departing from the scope of the present invention.
A part or all of each of the above-described embodiments can be described as in the following supplementary notes, but is not limited thereto.

(Appendix 1)
A primary storage device that temporarily stores data written from the host device, a first storage device having a storage capacity larger than that of the primary storage device, a storage capacity larger than that of the first storage device, and a reading speed A slow second storage device, and a control unit that records the data in either the first storage device or the second storage device in accordance with the entropy of the data stored in the primary storage device A storage device.

(Appendix 2)
The control unit records the data in the first storage device when the entropy of data stored in the primary storage device is lower than a threshold value, and stores the data in the second storage device when higher than the threshold value. The storage device according to appendix 1, which is recorded.

(Appendix 3)
The storage device according to appendix 1 or appendix 2, wherein the control unit calculates the entropy as a data compressibility rate.

(Appendix 4)
The control unit according to any one of appendix 1 to appendix 3, wherein when the data stored in the first storage device is recorded in the second storage device, the data is compressed and recorded. Storage device.

(Appendix 5)
The storage device according to appendix 4, wherein the control unit assigns a flag corresponding to a compressibility ratio when compressing the data stored in the primary storage device and recording the compressed data in the first storage device.

  DESCRIPTION OF SYMBOLS 100 ... Disk apparatus, 101 ... Controller, 103 ... Host interface, 105 ... Primary cache part, 107 ... Secondary cache part, 109 ... Disk interface, 111 ... Primary cache Control unit 113 ... Entropy calculation unit 115 ... Primary cache storage unit 117 ... Secondary cache control unit 119 ... Secondary cache storage unit 150 ... Disk drive 200 ...・ Host device

Claims (3)

A primary storage device for temporarily storing data written from the host device;
A first storage device having a larger storage capacity than the primary storage device;
A second storage device having a storage capacity larger than that of the first storage device and having a slower reading speed;
A control unit that controls whether to record the data in the first storage device or the second storage device according to the compressibility ratio of the data stored in the primary storage device ;
The control unit records the data in the first storage device when the compressibility ratio of the data stored in the primary storage device is lower than a threshold value, and records the data in the second storage when higher than the threshold value. Record to storage device,
Storage device. The controller compresses and records the data stored in the second storage device when recording the data stored in the first storage device.
Claim 1 Symbol placement of the storage device. Wherein, when compressing the data stored in the primary storage device for recording in the first storage device, imparts a flag corresponding to the compressible rate,
The storage apparatus according to claim 2 .