KR100980667B1 - Storage device, method, and computer readable recording medium for improving random write performance in the SDD - Google Patents

Abstract

According to the random write request from the host system, the data is first recorded in the log format on the HDD, and the data is transferred to the original storage location SSD during idle time, so that the data read operation retains the advantages of the SSD. At the same time, a storage device, a method, and a computer-readable recording medium are provided to maintain the advantages of an HDD in a write operation. According to the present invention, a hard disk drive (HDD) for storing a log data unit converting the data into a log format for a random data write request from a host system; And a solid state disk (SSD) in which the data stored in the form of a log data unit is transferred to and stored in the HDD when an idle time occurs.

SSD, HDD, Random Write, Log Data Unit

Description

Storage device, method, and computer readable recording medium for improving random write performance in SSD, and the like, and a computer-readable recording medium TECHNICAL FIELD OF IMPROVING RANDOM WRITE PERFORMANCE OF SOLID STATE DISK}

The present invention relates to a storage device, a method, and a computer-readable recording medium for improving random write performance in an SSD. More particularly, the present invention relates to a HDD primarily in response to a random write request from a host system. Storage device and method for recording data in the log format and transferring the data to the SSD which is the original storage location during idle time, thereby maintaining the advantages of the SSD in the data read operation and maintaining the advantages of the HDD in the write operation. And a computer readable recording medium.

Solid state disk (SSD) is a semiconductor-based mass storage device that can input / output data to and from storage devices such as flash memory through an interface of a conventional block-based disk. In general, the performance of sequential read / write and random read operations in SSDs based on flash memory shows better performance than hard disk drives (HDDs). However, the random write operation shows a relatively low level of performance compared to the other operations.

Accordingly, there is a need for development of a technology capable of improving performance in random write operations on SSDs.

As one of the technologies developed in accordance with this effort, D. Narayanan et al. There is a technology introduced in "Write Off-Loading: Practical Power Management for Enterprise Storage", U SENIX FAST 2008. This technology uses a strategy of reducing or stopping the rotation of the disk to reduce energy consumption in enterprise systems with multiple disks, temporarily sending the write request to the disk that is spinning or stopped to another disk that is normally running. It is a technique for obtaining an energy saving effect when the disc is at rest, i.e. when its rotation is reduced or stopped, by recording and reading it later. However, this technique can only reduce the energy consumption of disk operation and can not improve the performance of random write operations to the SSD.

In addition, Korean Patent No. 0801015 discloses a "hybrid hard disk drive and a data storage method." This technology allows hybrid hard disk drives to increase the time that disks spin at low speeds by adding flash memory devices that are relatively large (for example, 64 Mbytes, 128 Mbytes, or 256 Mbytes) of nonvolatile memory devices to existing hard disk drives. In order to minimize power consumption and improve performance, the write pattern from the host is analyzed and data is stored in a storage location selected from a disk and a flash memory device. However, this technique also can reduce power consumption, but there is a problem in that the random write operation on the SSD cannot achieve the same performance as that of the sequential write method obtained from the HDD.

Meanwhile, Korean Unexamined Patent Publication No. 10-2006-0099429 discloses a "composite memory device, a data processing method and a program". This technique is a technique for efficiently separating and using a nonvolatile memory in order to compensate for the capacity limitation of a small capacity nonvolatile memory and the low reliability by using the mechanical structure of an HDD or other recording medium. Specifically, a technique in which two storage devices are used as equivalent recording mediums, a HDD is used as a main storage device in a normal state, and a nonvolatile memory is used only as a recording space for transient data or frequently accessed data. However, this technique is also not intended to improve random write performance to the SSD, and there is a problem that still cannot achieve the same level of performance as sequential write during random write operations.

The present invention is to solve the above-described problems of the prior art, and writes the data in a log format to the HDD primarily in response to a random write request from the host system, and then stores the data in the SSD as an original storage location at idle time. As the data is transferred, the objective of the present invention is to maintain the advantages of the SSD in the data read operation while maintaining the advantages of the HDD having the sequential write performance in the write operation, thereby improving the overall performance of the storage device.

According to an embodiment of the present invention for achieving the above object, a hard disk drive (HDD) for storing a log data unit converting the data into a log format for a random data write request from a host system; And a solid state disk (SSD) in which the data stored in the form of a log data unit is transferred to and stored in the HDD when an idle time occurs.

The log data unit may include version information of the data, logical sector number information indicating a location where the data is transferred to the SSD, and the logical sector number of the SSD. Data, and a commit indicating the end of the data.

The storage device may further include a disk interface connecting the HDD and the SSD to the host system.

The HDD preferably allocates a physical sector number to a log data unit for the corresponding data in order of occurrence of a random write request from the host system and sequentially stores the physical sector number.

The storage device may include logical sector number information indicating a location in an SSD in which the data is stored, physical sector number information of the log data unit, offset information indicating a starting point of a log data unit having the physical sector number, and the log data. It may further include a log search DS including log size information indicating the size of the unit.

The log search DS may be a hash table or a red-black tree.

The HDD may be a memory collection after the data is transferred to the SSD.

On the other hand, according to another embodiment of the present invention for achieving the above object, as a method for improving the random write performance for the data, (a) for the random write data request from the host system, the log for the data Storing the data unit on the HDD; And (b) transferring the data stored in the form of a log data unit in the HDD to an SSD when an idle time occurs.

The log data unit may include version information of the data, logical sector number information indicating a location where the data is transferred to the SSD, and the logical sector number of the SSD. Data, and a commit indicating the end of the data.

Data stored by transferring to the SSD may be the most recent version of data.

Step (a) may include converting the data into the log data unit; Assigning a physical sector number to the log data unit; Storing the log data unit at a location of the HDD corresponding to the physical sector number; And adding an entry including the physical sector number information and metadata to a log search DS.

The metadata may include logical sector number information indicating a position in the SDD where the data is stored, offset information indicating a starting point of the log data unit, and log size information indicating a size of the log data unit.

The method may further include (c) performing memory collection on the HDD to merge space created in the HDD by the data migration.

In addition, the present invention provides another storage device, method for improving the random write performance of an SSD, or a computer readable recording medium for recording a computer program for executing the above methods.

According to the present invention, in response to a random write request from a host system, the corresponding data is first recorded in a log format on the HDD, and then the data is transferred to the SSD, which is a storage location of the original storage time, during idle time. At the same time, the advantages of the HDD having sequential write performance are maintained in the write operation, thereby improving the overall performance of the storage device.

DETAILED DESCRIPTION The following detailed description of the invention refers to the accompanying drawings that show, by way of illustration, specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that the various embodiments of the present invention are different but need not be mutually exclusive. For example, certain features, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the invention in connection with an embodiment. It is also to be understood that the position or arrangement of the individual components within each disclosed embodiment may be varied without departing from the spirit and scope of the invention. The following detailed description, therefore, is not to be taken in a limiting sense, and the scope of the present invention, if properly described, is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. Like reference numerals in the drawings refer to the same or similar functions throughout the several aspects.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention.

Configuration of the entire system

1 is a view showing the overall configuration of a storage system included in a host system according to an embodiment of the present invention.

As shown in FIG. 1, the storage system of the present invention includes a hard disk drive (HDD) (hereinafter, referred to as 'HDD') 110, one or more solid state disks (SDDs). And a disk interface 130 connecting the HDD 110 and the SDD 120 to the host system 100.

The HDD 110 according to an embodiment of the present invention is a space in which data and metadata to be stored in the storage device from the host system 100 are stored in a log format. All created data requested by the host system 100 are preferentially recorded in the HDD 110 in a log manner, and when the idle time occurs, the data is transferred to the SSD 120 and recorded. Accordingly, the SSD 120 functions as a primary storage device, and the HDD 110 functions as an auxiliary storage device that stores only data recorded in a log format. Meanwhile, although only one HDD 110 is illustrated in FIG. 1, the HDD 110 may be provided more than one according to the needs of those skilled in the art.

Next, the SSD 120 is a data storage device that stores data using a semiconductor such as a solid state memory, and is a storage device that can input and output data at high speed without a search time by random access. As described above, in the storage system of the present invention, all data requesting a recording from the host system 100 are first recorded in a log form on the HDD 110, and then the SSD 120 is stored at a later idle time. The data will be moved.

The disk interface 130 connects the HDD 110 and the SSD 120 with the host system 100. That is, the data recording request from the host system 100 is transmitted to the HDD 110 so that the corresponding data can be recorded on the HDD 110 in a log format, and the data recorded in the log format on the HDD 110 is later recorded. Read it and transfer it to SSD 120 to record. The disk interface 130 may use an interface method used in a typical computer such as Serial ATA (SATA) or Parallel ATA (PATA).

HDD Data units written to

Hereinafter, a log data unit stored in the HDD 110 in response to a data recording request from the initial host computer will be described.

2 is a diagram illustrating a configuration of a log data unit recorded in the HDD 110 according to an embodiment of the present invention.

As shown in FIG. 2, the log data unit 200 recorded on the HDD 110 may include version information 210, logical sector number information 220, actual data 230, and the like. And a commit 240.

The version information 210 indicates data recorded on the HDD 110, that is, version information of the actual data 230. When data is recorded, the same data may be stored several times and changed. In this case, version information 210 of each data is required to determine what the changed data is.

The logical sector number information 220 is information about a number indicating a position of data to be recorded when the actual data 230 is later transferred to the SSD 120 and recorded. The logical sector number information 220 is later converted into a physical address by a flash translation layer (FTL) in the SSD 120 and used. When the data recorded in the log format on the HDD 110 by the logical sector number information 220 is transferred to the SSD 120 serving as a main storage device, the data may be recorded and stored in the correct physical location.

The actual data 230 is data that is to be stored in the log format on the HDD 110 and then actually recorded and stored in the location of the SSD 120 indicated by the logical sector number information 220 later.

The commit 240 notifies the end of the data recording of the corresponding version and at the same time performs a function of notifying that the recorded log data unit 200 is completely written.

HDD In which log data units are written to

Hereinafter, a method and a process of recording a log data unit on the HDD 110 in response to a data write request from the host system 100 will be described.

First, the order in which log data units are recorded in the HDD 110 will be described with reference to FIG. 3. As shown in FIG. 3, the log data units 310 recorded in the HDD 110 may have a recording request. In order, they are inserted into a request queue (RQ) of the HDD 110. At this time, each log data unit 310 is assigned a physical sector number (Physical Sector Number) that contains information about the position actually recorded on the HDD (110), and is sequentially recorded on the HDD (110) .

Hereinafter, referring to FIG. 4, a process of writing a log data unit to the HDD 110 in response to a data write request from the host system 100 will be described.

First, when the host system 100 receives the corresponding data together with the data write request from the SSD 120 (S410), the data is converted into a log format and a physical sector number is allocated (S420). The result of log format conversion for arbitrary data is the same as that of the log data unit 200 as described above.

After converting the data into a log format and allocating a physical sector number, which is information about a position on the HDD 110, in which the resulting log data unit is actually recorded, the data is transferred to an I / O scheduler (S430). ).

Thereafter, for access to the data recorded in the HDD 110 in log format, the log search is performed on an entry in which the physical sector number, which is information about the position of the log data unit, and the meta information related to the data are recorded. Add to DS (Data Set) (S440). The log search DS contains information on whether data corresponding to a logical sector number exists in the HDD 110 in the form of a log or whether the data is transferred to the original location of the SSD 120. 5 is a data structure containing location information for easily finding the last version of data recorded at 110, and the data structure of an entry stored in the log search DS is shown in FIG.

Referring to FIG. 5, the data structure of entry 500 includes logical sector number information 510, physical sector number information 520, offset information 530, and log size information 540.

The logical sector number information 510 contains logical position information of a data unit in the HDD 110. The logical sector number information 510 is the same type of information as the logical sector number information 220 of the log data unit 200. On the other hand, the logical sector number information 510 in the entry functions as a key for distinguishing each entry, that is, as identification information.

The physical sector number information 520 contains location information in the HDD 110 where the log data unit is actually located. Through this, the log data unit recorded on the HDD 110 can be easily accessed.

The offset information 530 contains information regarding from which position the log data unit existing at the position corresponding to the physical sector number information 520 in the HDD 110 starts.

The log size information 540 contains information on the total size of the log data unit actually recorded on the HDD 110. Since the log data unit includes not only the actual data but also metadata associated with the corresponding data, the size of the actual data indicated by the logical sector number information 510 and the size of the log data unit for reading the data are inevitably different. Accordingly, the physical sector number information 520 corresponding to one logical sector number information 510 may be one or more consecutive numbers. Therefore, log size information 540 containing information on the size occupied by the corresponding log data unit is required, so that even if there is one or more physical sector number information 520 corresponding to the logical sector number information 510. The log data unit can then be identified.

Reading process of log data unit

Hereinafter, a process of reading a log data unit will be described with reference to FIG. 6.

First, all the log data units recorded on the HDD 110 are scanned when the host system 100 boots up, and then a log search DS is generated and possessed. As described above, the log search DS is a data structure having physical sector number information or the like which is information on the location where data exists in the HDD 110 in the form of a log data unit.

Subsequently, when a read request is received from the host system 100 and a logical sector number of the data to be read is obtained (S610), the latest version of data corresponding to the obtained logical sector number is obtained from the log search DS. Check whether the position (S620).

It is determined whether the corresponding data exists in the SSD 120 (S630). If it is determined that the data exists in the HDD 110 in the form of a log data unit, it corresponds to the physical sector number information included in the log search DS. Read the data from the position (S640). On the other hand, if it is determined that the data exists in the SSD 120, the data corresponding to the logical sector number is read from the SSD 120 (S650).

On the other hand, the log search DS that contains information for access to the log data unit may be configured as a data structure such as a hash table or a red-black tree in order to enable a quick search. . However, the present invention is not limited thereto, and other efficient data structures may be used as the log search DS.

HDD from SSD Data to the server

Hereinafter, referring to FIG. 7, a process in which data recorded in the log data unit format on the HDD 110 is transferred to the SSD 120 will be described.

First, when a command to transfer the data recorded in the log data unit format to the SSD 120 to the HDD 110 is received (S710), the respective versions are compared for each log data unit managed by the log search DS, and the final version is compared. After reading only the data (S720), it writes to the original position of the SSD 120 (S730). At this time, logical sector number information may be referred to.

Thereafter, the log data unit of the previous version recorded on the HDD 110 is deleted, and information on the data recorded on the SSD 120 is also deleted from the log search DS (S740).

Next, after determining whether there are any more log data units to deliver (S750), if remaining, repeat steps S720 to S740, and if there are no more log data units to deliver to the SSD 120, the HDD 110 In step S760, memory collection is performed on the space generated by the log data unit transfer, and merge is performed on the empty space. The empty spaces generated by data transfer by the memory collection are merged to secure a continuous space, thereby securing the space efficiency of the HDD 110.

In the present invention, for a random write request from the host system 100, the corresponding data is converted into a log format and sequentially written to the HDD 110 first, and then returned to the original position of the SSD 120 at idle time later. By allowing them to be moved, the performance of random writes can be improved.

In addition, the read operation reads the data recorded in the SSD 120, so that the advantages of the SSD 120 can be maintained, and at the same time, the write operation is primarily performed on the HDD 110. Can improve the performance of the entire storage device.

On the other hand, by writing data in log form primarily to the HDD 110 for random write requests, it is possible to reduce the decrease in recording speed due to log writing, merging, and erasing operations in the SSD 120. have.

Although the present invention has been described by specific embodiments such as specific components and the like, but the embodiments and the drawings are provided to assist in a more general understanding of the present invention, the present invention is not limited to the above embodiments. For those skilled in the art, various modifications and variations can be made from these descriptions.

Therefore, the spirit of the present invention should not be construed as being limited to the above-described embodiments, and all of the equivalents or equivalents of the claims, as well as the following claims, I will say.

1 is a view showing the overall configuration of a storage system included in a host system according to an embodiment of the present invention.

2 is a diagram illustrating a configuration of a log data unit recorded in an HDD according to an embodiment of the present invention.

3 is a view for explaining the order in which log data units are recorded in the HDD according to an embodiment of the present invention.

4 is a flowchart illustrating a process of writing a log data unit to an HDD in response to a data write request from a host system according to an embodiment of the present invention.

5 is a diagram illustrating a data structure of entries stored in a log search DS according to an embodiment of the present invention.

6 is a flowchart illustrating a process of reading a log data unit according to an embodiment of the present invention.

FIG. 7 is a flowchart illustrating a process of transferring data recorded in a log data unit format to an SSD according to an embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

100: host system

110: HDD

120: SSD

130: disk interface

Claims (14)

A hard disk drive (HDD) for storing a log data unit converting the data into a log format for a random write request from a host system; And A solid state disk (SSD) in which the data stored in the form of log data units are transferred to and stored in the HDD when an idle time occurs; And the HDD sequentially allocates physical sector numbers to log data units for corresponding data in the order in which random write requests from the host system occur. The method of claim 1, The log data unit, Version information of the data, logical sector number information indicating a location to be stored when the data is transferred to the SSD, the data to be actually stored in the logical sector number of the SSD, and the data A storage device that includes a commit informing the end. The method of claim 1, And a disk interface connecting the HDD and the SSD to the host system. delete The method of claim 1, Logical sector number information indicating a position in the SSD where the data is stored, physical sector number information of the log data unit, offset information indicating a starting point of a log data unit having the physical sector number, and a size of the log data unit And a log search DS containing log size information. The method of claim 5, The log search DS is a hash table or a red-black tree. The method of claim 1, The HDD is a memory collection (Garbage Collection) after the data is transferred to the SSD. As a method for improving random write performance on data, (a) for a random write request from the host system, storing a log data unit for the data on the HDD; And (b) transferring the data stored in the form of a log data unit in the HDD to an SSD when an idle time occurs; And storing the HDD sequentially by assigning physical sector numbers to log data units of corresponding data in the order in which random write requests from the host system occur. The method of claim 8, The log data unit, Version information of the data, logical sector number information indicating a location to be stored when the data is transferred to the SSD, the data to be actually stored in the logical sector number of the SSD, and the data How to include a commit to the end. 10. The method of claim 9, And the data stored in the SSD is the most recent version of the data. The method of claim 8, In step (a), Converting the data into the log data unit; Assigning a physical sector number to the log data unit; Storing the log data unit at a location of the HDD corresponding to the physical sector number; And Adding an entry containing the physical sector number information and metadata to a log search DS. The method of claim 11, The metadata includes logical sector number information indicating a location in the SDD where the data is stored, offset information indicating the starting point of the log data unit, and log size information indicating the size of the log data unit. The method of claim 8, (c) performing memory collection on the HDD to merge space created on the HDD by the data migration. A computer-readable recording medium for recording a computer program for executing the method according to any one of claims 8 to 13.

Images