KR101353968B1 - Data process method for replacement and garbage collection data in non-volatile memory cache having ring structure - Google Patents

Abstract

A data processing method is provided. According to this data processing method, a replacement operation is performed when a read access request for new data corresponding to K blocks (K is a natural number of 1 or more) is requested while there is only one free block in the read cache area of the nonvolatile memory cache. It is possible to provide a data processing method for performing garbage collection operations when all pages of the write cache area of the nonvolatile memory cache are used, so that the storage device can have lower power consumption and higher processing speed of data. do.

Description

Data process method for replacement and garbage collection data in non-volatile memory cache having ring structure}

The present invention relates to a data processing method, and more particularly, to a data processing method for replacing and garbage collection of data written to an annular nonvolatile memory cache.

Recently, as interest in green IT has increased due to environmental problems, various studies on efficient energy management of high power consumption storage systems have been conducted. With the development of the Internet, network services such as web servers and e-commerce have been greatly activated, and high-performance large-capacity disks are used to store large amounts of data, and data backup is also replacing disks. Disks can be classified into an active mode for performing I / O operations, an idle mode for waiting for an IO request, and a standby mode for stopping disk rotation when there is no IO request for a certain period of time. . The power consumed in idle mode saves only 20% of the power consumed in active mode. As a result, if the disk does not enter standby mode, the disk consumes a large amount of power even without performing IO operation.

As a method for significantly reducing the power consumption of a disk, a method of replacing a flash memory with a disk has been used. Flash memory has lower latency and power consumption than disks. However, flash memory is much more expensive than disk, making it unsuitable for storage systems running on large disks. Instead, flash memory can be used as a cache to save data and improve performance by putting frequently reused data in the cache to keep the disk on standby. This may have lower latency than using an extended DRAM cache, but it is a good alternative to disk cache because it is cheaper and has better random access read performance than DRAM.

As such, the use of flash memory caches to improve disk performance and energy savings has been studied earlier. However, this study proposes only basic system structure and algorithm and does not consider data allocation and wear-leveling according to flash memory characteristics.

As described above, in using the flash memory as a cache, a search for a method for providing a flash memory management method in consideration of the characteristics of the flash memory is required.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to correspond to K blocks (K is a natural number of 1 or more) in a state in which one empty block of a read cache area of a nonvolatile memory cache remains. The present invention provides a data processing method of performing a replacement operation when read access to new data is requested, and garbage collection when all pages of the write cache area of the nonvolatile memory cache are used.

According to an embodiment of the present invention, a data processing method of a storage device including a nonvolatile memory cache having an annular structure includes a state in which one empty block of a read cache area of the nonvolatile memory cache remains. If read access is requested for new data corresponding to K blocks (K is a natural number of 1 or more), selecting K blocks among the blocks of the read cache area and performing a replacement operation of recording the new data ; When all pages of the write cache area of the nonvolatile memory cache are used, garbage collection, which writes all valid data written to the write cache area to the disk array, and moves the predetermined number of blocks after deleting all the write cache areas. It includes; performing a task.

In the performing of the replacing operation, when read access is requested for new data corresponding to K blocks in a state in which one empty block remains in the read cache area, K replacement blocks in the read cache area are requested. Selecting K; Deleting the K replacement blocks to replace the empty blocks; Recording the new data into one empty block and the selected K empty blocks; Displaying a value indicating no access to a spare area of blocks in which the new data is recorded; And associating a read allocation pointer with one remaining empty block among the selected K blocks.

In the selecting step, the replacement target block may be selected using a read replacement pointer and a bit value of a spare area of the block.

In the selecting step, the replacement target block may be selected using the access list.

The garbage collection operation may include writing all valid pages in invalid blocks among blocks of the write cache area in a block of the first reserved area when all pages of the write cache area are used; Deleting all invalid blocks in the write cache area; Copying data of a valid block in the write cache area and data written in a block of a first reserved area in the nonvolatile memory cache to a disk array of the storage device; Deleting a valid block in the write cache area and at least one block in which data of the first reserved area is recorded; And moving the write cache area by the number of at least one block in which data of the first reserved area is recorded.

The invalid block may indicate a block in which the spare area includes an invalid page, and the valid page may indicate a page in which the data is valid in the spare area of the page.

According to various embodiments of the present disclosure, when a read access to new data corresponding to K blocks (K is a natural number of 1 or more) is requested, when there is only one free block in the read cache area of the nonvolatile memory cache, it is replaced. It is possible to provide a data processing method for performing tasks and performing garbage collection operations when all pages of the write cache area of the nonvolatile memory cache are used, so that the storage device has lower power consumption and higher processing speed of data. It becomes possible.

1 is a diagram illustrating a structure of a storage device using a nonvolatile memory cache according to an embodiment of the present invention;
2 is a diagram illustrating a structure of a nonvolatile memory cache 200 having an annular structure, according to an embodiment of the present invention;
3A is a view showing the structure of one block according to an embodiment of the present invention;
3B illustrates a block including a plurality of pages according to an embodiment of the present invention;
4A is a diagram illustrating a state of a nonvolatile memory cache having an annular structure before data is processed according to an embodiment of the present invention;
4B is a diagram illustrating a case where data processing for reading and writing is performed according to an embodiment of the present invention;
4C is a diagram illustrating a case in which data of a second block is out-place updated according to an embodiment of the present invention;
FIG. 5A illustrates a case in which access to new data corresponding to three blocks occurs in a state in which one empty block of a read cache area remains, according to an embodiment of the present invention; FIG.
5B is a diagram illustrating a state in which three blocks are selected as empty blocks, according to an embodiment of the present invention;
FIG. 6A illustrates a case in which access to new data corresponding to three blocks occurs in a state in which one empty block of a read cache area remains, according to an embodiment of the present invention; FIG.
6B illustrates a state in which three blocks are selected as empty blocks using an access list according to an embodiment of the present invention;
7A is a diagram illustrating a case where all write cache regions are used, according to an embodiment of the present invention;
FIG. 7B is a diagram illustrating a state in which a write cache region is moved clockwise by one block in a garbage collection process according to an embodiment of the present invention.

Hereinafter, the present invention will be described in detail with reference to the drawings.

1 is a diagram illustrating a structure of a storage device 100 using a nonvolatile memory cache according to an embodiment of the present invention. As shown in FIG. 1, the storage device 100 includes a storage controller 110 and a disk array 120.

As shown in FIG. 1, the storage controller 110 includes a CPU 112, a memory 114, and a nonvolatile memory cache 116. The CPU 112 controls the overall operation of the storage device 100. The memory 114 is a space for temporarily storing data processed by the CPU and includes a RAM and a cache memory.

The nonvolatile memory cache 116 corresponds to a large capacity cache using nonvolatile memory such as an SSD. The nonvolatile memory used in the nonvolatile memory cache 116 may be an SSD using a flash memory. In addition, a solid state drive (SSD) using various types of nonvolatile memory may be used. For example, the nonvolatile memory cache 116 may be applied to any nonvolatile memory cache such as SSD and flash memory.

The disk array 120 includes a plurality of disks to implement a large storage space.

When a data access request is made, the CPU 112 first searches whether data exists in the memory 114. The CPU 112 accesses the data if there is data in the memory 114. On the other hand, if there is no data in the memory 114, the CPU 112 searches whether the data exists in the nonvolatile memory cache 116. If there is data in the nonvolatile memory cache 116, the CPU 112 accesses the data. On the other hand, if there is no data in the nonvolatile memory cache 116, the CPU 112 accesses the data in the disk array 120.

As such, the CPU 1120 attempts to access data in the order of the memory 114-> nonvolatile memory cache 116-> disk array 120. Accordingly, the storage device 100 may improve disk IO performance by using the nonvolatile memory cache 116 and may reduce power consumption of the disk array 120.

The storage device 100 of FIG. 1 may be applied to any electronic device including the nonvolatile memory cache 116. For example, the storage device 100 may be a computer device including a nonvolatile memory cache 116, a disk array device, a server device, a notebook computer, a mobile phone, or the like.

Hereinafter, the structure and operation of the nonvolatile memory cache 116 will be described in detail. The structure and operation of the nonvolatile memory cache 116 described below may be performed by the CPU 112 in accordance with the algorithm described below.

2 is a diagram illustrating a structure of a nonvolatile memory cache 200 having an annular structure according to an embodiment of the present invention. Here, the nonvolatile memory cache 200 corresponds to the nonvolatile memory cache 160 of FIG. 1, and may be an SSD cache, a flash memory cache, or the like.

As shown in FIG. 2, the nonvolatile memory cache 200 has a block structure in a storage area. In practice, the nonvolatile memory cache 200 may have an annular structure by connecting the last block of data with the first block.

As shown in FIG. 2, the nonvolatile memory cache 200 includes a write cache area 210, a read cache area 220, a first reserved area 230, and a second reserved area 240.

The write cache area 210 is an area used to perform a write access operation, and the read cache area 220 is an area used to perform a read access operation. As described above, since the read area and the write area are divided in the nonvolatile memory cache 200, the miss rate of the cache can be reduced.

In addition, as shown in FIG. 2, the read cache area 220 may confirm that more storage areas are allocated than the write cache area 210. Since the read cache area 220 has a high hit rate, is not sensitive to out place updates, and has more read accesses than write accesses, the read cache area 220 may be allocated a large number of read cache areas to maximize IO efficiency. .

Specifically, the read cache area 220, the write cache area 210, the first reserved area 230, and the second reserved area 240 each represent 70%, 10%, 10%, and 10% of the total cache space. Can occupy.

In addition, the first reserved area 230 and the second reserved area 240 are reserved areas for out-place update and garbage collection.

Out-place update corresponds to a method of updating data due to the electrical characteristics of flash memory, which is a nonvolatile memory. The electrical characteristics of flash memory make it impossible to overwrite data on a written page. Therefore, when modifying data in flash memory, new data is written by allocating an empty area, and old data pages are marked as invalid in the spare area. This method is called an out-place update. At this time, the areas for recording new data for out-place updating are the first reserved area 230 and the second reserved area 240.

In addition, garbage collection is required when most pages in the write cache area 210 are written out and the empty area becomes insufficient. Garbage collection is to delete the block including the invalidated page spread in the block in the write cache area 210 to secure an empty block area. The first reserved area 230 and the second reserved area 240 are used for such garbage collection. To this end, the first reserved area 230 and the second reserved area 240 are each allocated with the same size as the write cache area 210 or within 1 to 2 times.

As shown in FIG. 2, it can be seen that the nonvolatile memory cache 200 having an annular structure is divided in block units. Hereinafter, a block unit structure will be described in detail with reference to FIGS. 3A and 3B.

3A is a diagram illustrating the structure of one block 300 according to an embodiment of the present invention. As shown in FIG. 3A, one block 300 includes a data area 310 and a spare area. The data area 310 is an area where substantial data is recorded in units of pages. In the spare area 320, metadata associated with the corresponding block unit and the page unit within the block are recorded. In detail, the spare area 320 may record whether data is recorded in the data area 310, whether access has been made, and whether data recorded in the data area 310 is valid. For example, the spare area 320 of one block of the read cache area 220 indicates an empty block when there is no value, and when the value is "0", data is recorded in the data area 310, but access is not possible. When the value is " 1 ", the data is recorded in the data area 310 and there is access.

Meanwhile, one block 300 includes a plurality of pages. 3B illustrates a block 300 including a plurality of pages according to an embodiment of the present invention. As shown in FIG. 3B, one block 300 may include a plurality of pages. One page includes a page data area 311 and a page spare area 321. The page data area 311 actually becomes a unit area for reading and writing data. In the page spare area 321, information indicating the state of the corresponding page is recorded.

For example, the page spare area 321 of one page of the write cache area 210 indicates a blank page when there is no value, and when the value is "I", data is recorded in the page data area 311 but invalidated. If the value is " V ", the data is recorded in the page data area 311 and is valid data (Valid).

In addition, the spare area 320 of one block of the write cache area 210 indicates an empty block in which all pages are empty when there is no value, and when the value is "I", one of a plurality of pages included in the block. Even if the page is invalidated, the value "V" indicates that the plurality of pages included in the block are all valid.

Hereinafter, a data processing method for reading and writing data to and from the nonvolatile memory cache 200 having an annular structure will be described. The annular nonvolatile memory cache 200 accesses the write cache area 210 in units of pages and accesses the read cache area 220 in units of blocks.

First, a method of reading data from the nonvolatile memory cache 200 having an annular structure will be described.

When a read access is requested, the storage device 100 searches in block units whether the first data, which is data corresponding to the requested read access, exists in the read cache area 220 of the nonvolatile memory cache 200. .

When the first data exists in the read cache area 220, the storage device 100 reads the data from the first block that is a block corresponding to the first data. The storage device 100 indicates that access has been made to the spare area of the first block. For example, the storage device 100 changes the bit indicating whether to access the spare area of the first block from '0' to '1'. Thereafter, the storage device 100 associates a read allocation pointer R_alloc_ptr with an empty block located after the first block. Here, the read allocation pointer is a pointer used to find an empty block in the read cache area 220.

In contrast, when the first data does not exist in the read cache area 220, the storage device 100 reads the first data from the disk array 120. The storage device 100 writes the first data to a second block that is an empty block to which the read allocation pointer R_alloc_ptr corresponds. In addition, the storage device 100 indicates that data was recorded in the spare area of the second block but no access was made. For example, the storage device 100 displays a bit indicating whether access to the spare area of the first block is '0'. Thereafter, the storage device 100 associates a read allocation pointer R_alloc_ptr with an empty block located after the second block.

Through this process, the storage device 100 reads data from the nonvolatile memory cache 200 having an annular structure.

Secondly, a method of writing data in the nonvolatile memory cache 200 having an annular structure will be described.

When a write access is requested, the storage device 100 searches whether the second data, which is data corresponding to the requested write access, exists in the write cache area 210 of the nonvolatile memory cache 200 in units of pages. .

When the second data exists in the write cache area 210, the storage device 100 indicates that data written in the spare area of the first page, which is a page corresponding to the second data, is invalidated. For example, the storage device 100 changes a bit indicating the state of the spare area of the first page from 'V' to 'I' (from Valid to Invalid). In addition, the storage device 100 indicates that there is a page invalidated in the spare area of the third block, which is a block including the first page. For example, the storage device 100 sets a bit indicating the state of the spare area of the third block from 'V' (or '0') to 'I' (or '1') (Valid to Invalid). Will change. Thereafter, the storage device 100 writes the second data in the first reserved area 230. Then, the storage device 100 associates a write allocation pointer W_alloc_ptr with a blank page located after the first page. Here, the write allocation pointer is a pointer used to find a blank page in the write cache area 210. This process corresponds to a method of updating changed data, which is called an out place update.

On the other hand, when the second data does not exist in the write cache area 210, the storage device 100 writes the second data to the second page, which is a blank page to which the write allocation pointer W_alloc_ptr corresponds. In addition, the storage device 100 indicates that data recorded in the spare area of the second page is valid. For example, the storage device 100 displays a bit indicating the state of the spare area of the second page as 'V' (or '0') (as Valid). If there is a blank page in the block including the second page, the storage device 100 moves the write allocation pointer W_alloc_ptr to the next blank page. On the other hand, if there is no blank page in the block including the second page, the storage device 100 moves the write allocation pointer W_alloc_ptr to the next block.

Through this process, the storage device 100 writes or updates data (out-place update) to the nonvolatile memory cache 200 having an annular structure.

Hereinafter, a data processing method for reading and writing data to and from the nonvolatile memory cache 200 having an annular structure will be described with reference to FIGS. 4A to 4C.

4A is a diagram illustrating a state of a nonvolatile memory cache 200 having an annular structure before data is processed according to an embodiment of the present invention.

As shown in FIG. 4A, the read cache area 220 includes an empty block and a block in which data is written, and blocks in which data is written are '0' indicating that there is no access to data in the spare area. You'll see that '1' is displayed, indicating that there was access to the data. In addition, it can be seen that the read allocation pointer R_alloc_ptr corresponds to the first block 410 which is an empty block.

In addition, as shown in FIG. 4A, the write cache area 210 includes an empty block and a block in which data is recorded. When data is recorded in at least one of the included pages and all of the recorded pages are valid pages, 'V' indicating that the block is valid for the data is displayed in the spare area. In addition, it can be seen that the write allocation pointer W_alloc_ptr corresponds to the second block 420 which is an empty block.

4B is a diagram illustrating a case where data processing for reading and writing is performed according to an embodiment of the present invention. In FIG. 4B, 1 indicates that a read access is performed on data recorded in the third block 430 of the read cache area 220, and 2 indicates a first block of data not recorded in the read cache area 220. 410 denotes to perform a read access, and (3) denotes writing data to the second block 420 to perform a write access.

As can be seen by comparing FIG. 4A with FIG. 4B, the third block 430 may check that the value of the spare area is changed from '0' to '1' while performing the process of ①.

In addition, the first block 410 may check that data is recorded in the data area and '0' is displayed in the spare area while performing step ②. In this process, it may be confirmed that the read allocation pointer R_alloc_ptr has been changed from a first block 410 to a fourth block 440 which is the next empty block.

In addition, the second block 420 may check that data is recorded in the data area and 'V' is displayed in the spare area while performing step (3). In addition, since the second block 420 has written all the pages in the process of ③, it is confirmed that the write allocation pointer W_alloc_ptr has been changed from the second block 420 to the fifth block 450 which is the next empty block. Can be.

Meanwhile, FIG. 4C is a diagram illustrating a case where data of the second block 420 is out-place updated according to an embodiment of the present invention.

In FIG. 4C,? Indicates that out-place update is performed on data recorded in the second block 420 of the write cache area 210. As shown in FIG. 4C, when data of a page corresponding to some or all of the data recorded in the second block 420 is updated, the updated data is the sixth block 460 of the first reserved area 230. ) Is recorded in units of updated pages. In addition, the updated page of the second block 420 displays 'I' to indicate that data is invalid in the spare area of the corresponding page, and the data is invalid in the spare area of the second block 420. 'I' will be displayed to indicate that there is an in-page.

Through this process, the storage device 100 performs read and write access to data.

Hereinafter, a data processing method for replacing and garbage collection of data recorded in the nonvolatile memory cache 200 having an annular structure will be described. Here, the replacement operation is performed when the free cache is insufficient in the read cache area 220, and garbage collection is performed when the empty cache is insufficient in the write cache area 210.

First, a data processing method for replacing data written in the nonvolatile memory cache 200 having an annular structure will be described.

The storage device 100 includes at least one block (hereinafter, referred to as K: K is a natural number of 1 or more) in a state in which one empty block remains in the read cache area 220 of the annular nonvolatile memory cache 200. When read access is requested for new data corresponding to the data, K blocks of the blocks of the read cache area 220 are selected and a replacement operation of writing new data is performed.

In detail, when the read access request for new data corresponding to the K blocks is requested with one empty block remaining in the read cache area 220, the storage device 100 may store K in the read cache area 220. Select K replacement blocks. There are two methods for selecting a block to be replaced, which will be described below.

In a first method, the storage device 100 may apply a method of selecting a replacement block using a read replacement pointer R_rep_ptr and a bit value of a spare area of a block. In detail, the storage device 100 checks the value of the spare area of the block while sequentially moving the read replacement pointer R_rep_ptr in a specific direction from one remaining empty block. If the read replacement pointer is a value indicating that the spare area of the corresponding block has been accessed (for example, '1'), the storage device 100 has no access to the spare area of the corresponding block. Change to a value that represents (for example, '0'). On the other hand, if the read replacement pointer is a value (eg, '0') indicating that the spare area of the corresponding block does not have access, the storage device 100 deletes the block and changes it to an empty block. The storage device 100 repeats this process until K empty blocks are secured.

As a second method, the storage device 100 may apply a method of selecting a replacement target block using an access list. Here, the access list is in accordance with the Least Recently Used (LRU) algorithm and represents an approximate frequency of access of the block. Herein, items corresponding to all blocks of the read cache area 220 are sequentially arranged in the access list. In addition, the access list is referred to as a head (Head) and the tail is called a tail (Tail), the closer to the head indicates a block that has recently accessed history, the closer to the tail indicates a block that is the last access point is older.

In detail, the storage device 100 sequentially checks the spare area value of the block in the head direction from the block corresponding to the tail of the access list. If the spare area of the corresponding block has a value indicating that there is an access (for example, '1'), the storage device 100 determines a value indicating that there is no access to the spare area of the corresponding block ( For example, change to '0'. On the other hand, if the read replacement pointer is a value (eg, '0') indicating that the spare area of the corresponding block does not have access, the storage device 100 deletes the block and changes it to an empty block. The storage device 100 repeats this process until K empty blocks are secured.

As such, the storage device 100 may select a replacement target block in two ways.

Thereafter, the storage device 100 records data corresponding to K blocks in one existing empty block and newly secured empty blocks, and there is no access to the spare area of the block in which the data is recorded. It will display a value (for example, '0'). The storage device 100 associates the read allocation pointer R_alloc_ptr with the remaining one free block.

At this time, when the replacement target block is selected by the second method, the storage device 100 moves items corresponding to the blocks in which data is recorded in the access list to the head.

Through this process, the storage device 100 performs a data processing method of replacing data written in the nonvolatile memory cache 200 having an annular structure.

5A to 6B, a data processing method for replacing data recorded in the nonvolatile memory cache 200 having the annular structure described above will be described by way of example.

5A and 5B illustrate a process of selecting a block to be replaced by the first method. FIG. 5A illustrates a case in which access to new data corresponding to three blocks occurs in a state in which one empty block of a read cache area remains, according to an embodiment of the present invention. 5B is a diagram illustrating a state in which three blocks are selected as empty blocks according to an embodiment of the present invention.

As illustrated in FIG. 5A, when one empty block 510 is left in the read cache area, when a read access of ① (access to new data corresponding to three blocks) occurs, the storage device 100 is generated. ) First checks the value of the spare area of the block while sequentially moving the read replacement pointer (R_rep_ptr) clockwise. If the value of the spare area of the block is '1', the storage device 100 changes the value to '0', and if the value is '0', the storage device 100 deletes the block to make the block empty.

Thus, as shown in FIG. 5B, the storage device 100 makes the second block 520, the third block 530, and the fifth block 550 an empty block, and the fourth block 540 It can be seen that the value of the spare area is changed from '1' to '0'.

Thereafter, the storage device 100 reads data to be accessed from the disk array 120 and then accesses the first block 510, the second block 520, and the third block 530. The data will be recorded. In addition, the storage device 100 may correspond to the fifth block 550, which is a free block, with the read allocation pointer R_alloc_ptr and the read replacement pointer R_rep_ptr.

6A and 6B illustrate a process of selecting a replacement target block by the second method. FIG. 6A illustrates a case in which access to new data corresponding to three blocks occurs in a state in which one empty block of a read cache area remains, according to an embodiment of the present invention. 6B is a diagram illustrating a state in which three blocks are selected as empty blocks using an access list according to an embodiment of the present invention.

As illustrated in FIG. 6A, when one empty block 610 is left in the read cache area, a read access of 1 (access to new data corresponding to three blocks) occurs, the storage device 100. ) Sequentially checks the value of the spare area of the block while sequentially moving from the block of the item close to the tail of the access list 600 to the head direction. If the value of the spare area of the block is '1', the storage device 100 changes the value to '0', and if the value is '0', the storage device 100 deletes the block to make the block empty.

Thus, as shown in FIG. 6B, the storage device 100 makes the second block 520, the third block 530, and the fifth block 550 an empty block, and the fourth block 540 It can be seen that the value of the spare area is changed from '1' to '0'.

Thereafter, the storage device 100 reads data to be accessed from the disk array 120 and then accesses the first block 510, the second block 520, and the third block 530. The data will be recorded. The storage device 100 moves items corresponding to the first block 510, the second block 520, and the third block 530 to the head in the access list 600. In addition, the storage device 100 maps the read allocation pointer R_alloc_ptr and the read replacement pointer R_rep_ptr to the fifth block 550 which is an empty block.

So far, the data processing method of replacing data recorded in the nonvolatile memory cache 200 having an annular structure has been described with reference to FIGS. 5A to 6B.

Secondly, a data processing method for garbage collection of data recorded in the annular nonvolatile memory cache 200 will be described.

When all the pages of the write cache area 210 of the annular nonvolatile memory cache 200 are used, the storage device 100 stores all valid data written in the write cache area 210 to the disk array 120. After deleting all of the write and write cache areas 210, a garbage collection operation is performed to move a predetermined number of blocks.

First, when all pages of the write cache area 210 are used, the storage device 100 stores all valid pages in the invalid blocks among the blocks of the write cache area 210 in the block of the first reserved area 230. Record it. Here, an invalid block indicates a block having an indication (for example, "I") indicating that a spare area contains a page where data is invalid, and a valid page indicates that data is valid in a spare area of a page. Represents a displayed page (e.g., "V").

In detail, the storage device 100 places the write replacement pointer W_rep_ptr in the first invalid block in the write cache area 210. The storage device 100 searches for valid pages in the invalid block. In addition, the storage device 100 records the valid page in the invalid state in the first reserved area 230. Thereafter, the storage device 100 places the write replacement pointer W_rep_ptr in the next invalid block, and then repeats the above-described process, and repeats the above process until the last invalid block of the write cache area. Accordingly, when all pages of the write cache area 210 are used, the storage device 100 blocks all valid pages in the invalid blocks among the blocks of the write cache area 210 of the first reserved area 230. Will be written in.

Then, the storage device 100 deletes all invalid blocks in the write cache area 210. The storage device 100 copies both the data of the effective block in the write cache area 210 and the data recorded in the block of the first reserved area 230 to the disk array 120. The storage device 100 also deletes at least one block in which the valid block in the write cache area 210 and the data in the first reserved area 230 are recorded.

Thereafter, the storage device 100 moves the write cache area 210 by at least one block in which data of the first reserved area 230 is recorded. Specifically, it is assumed that at least one block in which data in the first reserved area 230 is recorded is L (L is a natural number of 1 or more). The storage device 100 copies the blocks of the L read cache areas 230 adjacent to the first reserved area 230 to the second reserved area 240, and copies the blocks of the copied L read cache areas 230. Delete it to make an empty block. In this case, blocks of the L read cache areas 230 adjacent to the first reserved area 230 are sequentially copied from blocks adjacent to the read cache area 230 among the second reserved areas 240. Next, the storage device 100 moves the write cache area 210 by L blocks toward the first reserved area 230. Through this process, the storage device 100 moves the write cache area 210 by at least one block in which data of the first reserved area 230 is recorded.

Through this process, when all pages of the write cache area 210 of the nonvolatile memory cache 200 of the annular structure are used, all of the valid data recorded in the write cache area 210 are used. After the write to the disk array 120 and the write cache area 210 are all deleted, the garbage collection operation may be performed by moving the predetermined number of blocks.

Among the nonvolatile memories, the flash memory is limited to about 10 to 1 million block erases, so the erase operation may be concentrated on a specific block and may be unusable due to wear-out. However, the write cache area 210 By periodically moving), the likelihood of wear-out can be minimized.

Hereinafter, with reference to FIGS. 7A and 7B, a data processing method for garbage collection of data recorded in the nonvolatile memory cache 200 having the annular structure described above will be described as an example.

FIG. 7A is a diagram illustrating a case where all write cache regions are used, according to an embodiment of the present invention. FIG. 7B is a diagram illustrating a state in which a write cache region is moved clockwise by one block in a garbage collection process according to an embodiment of the present invention.

As shown in FIG. 7A, when data is written to the first block 710 so that the write cache area is completely used, the storage device 100 starts garbage collection. In this case, as shown in FIG. 7A, the second block 720 of the first reserved area 230 is in a state in which data is written to some pages through out-place updating.

In this state, as shown in FIG. 7B, the storage device 100 writes valid pages included in all invalid blocks 760, 770, and 780 in the write cache area 210 to blocks in the first reserved area 230. Delete all invalid blocks 760, 770, 780. In the case of FIG. 7B, all valid pages included in all invalid blocks are recorded in the second block 720. Thereafter, the storage device 100 copies the data recorded in the first block 710 and the second block 720 to the disk array 120, and then the first block 710 and the second block 720 are stored. Will be deleted.

Next, the storage device 100 moves the write cache area 210 by one block clockwise to incorporate the second block 720 into the write cache area 210. To this end, after the storage device 100 moves the third block 730 to the second reserved area 240 and the fourth block 740, the storage device 100 incorporates the third block 730 into the first reserved area 230. Let's do it. The storage device 100 incorporates the fourth block 740 into the read cache area 220. In addition, the storage device 100 incorporates the sixth block 760 into the second reserved area 240 so that the second reserved area 240 is between the fifth block 750 and the sixth block 760. Move one block.

Through this process, the storage device 100 finishes garbage collection, and the write cache area 210 moves by one block clockwise.

 So far, the data processing method for garbage collection of the data recorded in the nonvolatile memory cache 200 having the annular structure described above has been described with reference to FIGS. 7A and 7B.

Meanwhile, the technical idea of the present invention may be applied to a computer-readable recording medium including a nonvolatile memory cache having an annular structure and a computer program implementing the above-described data processing method. In addition, the technical idea according to various embodiments of the present invention may be embodied in computer-readable code form recorded on a computer-readable recording medium. The computer-readable recording medium is any data storage device that can be read by a computer and can store data. For example, the computer-readable recording medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical disk, a hard disk drive, or the like. In addition, the computer readable code or program stored in the computer readable recording medium may be transmitted through a network connected between the computers.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention.

100: storage device 110: storage controller
112: CPU 114: memory
116, 200: nonvolatile memory cache 210: write cache area
220: read cache area 230: first reserved area
240: second reserved area

Claims (6)

A data processing method of a storage device including a nonvolatile memory cache having an annular structure,
If read access is requested for new data corresponding to K blocks (K is a natural number equal to or greater than 1) when only one empty block of the read cache area of the nonvolatile memory cache is left, K of the blocks of the read cache area is requested. Selecting two blocks to perform a replacement operation of recording the new data;
If all pages of the write cache area of the nonvolatile memory cache are used, garbage collection operation is performed to write valid data written to the write cache area to the disk array, and move the predetermined number of blocks after deleting the write cache area. Performing;
The garbage collection work step,
When all pages of the write cache area are used, writing all valid pages in invalid blocks among blocks of the write cache area into a block of a first reserved area;
Deleting all invalid blocks in the write cache area;
Copying data of a valid block in the write cache area and data written in a block of a first reserved area in the nonvolatile memory cache to a disk array of the storage device;
Deleting a valid block in the write cache area and at least one block in which data of the first reserved area is recorded; And
And moving the write cache area by the number of at least one block in which data of the first reserved area is recorded.
The method of claim 1,
Performing the replacement operation,
Selecting K replacement blocks in the read cache area when a read access request for new data corresponding to K blocks is requested when only one empty block remains in the read cache area;
Deleting the K replacement blocks to replace the empty blocks;
Recording the new data into one empty block and the selected K empty blocks;
Displaying a value indicating no access to a spare area of blocks in which the new data is recorded; And
And associating a read allocation pointer with one remaining free block among the selected K blocks.
3. The method of claim 2,
Wherein the selecting step comprises:
And selecting the replacement block using a read replacement pointer and a bit value of a spare area of the block.
3. The method of claim 2,
Wherein the selecting step comprises:
A data processing method comprising selecting a block to be replaced using an access list.
delete The method of claim 1,
The invalid block represents a block in which the spare area has an indication indicating that a page whose data is invalid is included,
And the valid page indicates a page on which a valid data is displayed in a spare area of the page.

Images