KR101375838B1 - Data storage system and data management method - Google Patents

Abstract

The present invention provides a memory device comprising: a first volatile memory in which data is stored; And a second volatile memory sequentially storing the updated contents of the data in an order of occurrence in an update contents storage space dynamically allocated according to the update frequency of the data, wherein the updated contents are stored in the first volatile memory. When the data is updated in the upper memory, a data storage system is provided, which is a difference between the data before updating and the updated data.

Description

TECHNICAL FIELD [0001] The present invention relates to a data storage system and a data management method,

The present invention relates to a data storage system, and more particularly, to a data storage system and a data management method using a flash memory and a phase change memory.

Flash memory-based storage devices are popular because they have faster read / write speeds than conventional disk storage devices. For example, in the table of FIG. 3, which compares the speed difference of various storage media, the disk takes 12.7 milliseconds to read and 13.7 milliseconds to write 2KB of data, while NAND flash memory. The access time for the same amount of data is only 75 microseconds for read and 250 microseconds for write, which is much faster.

Flash memory, however, has several drawbacks. For example, as shown in the table, unlike a hard disk, a flash memory must first delete the data before storing the data. The read / write operation of the flash memory is performed by page unit rather than byte unit, and data deletion operation should be performed by block unit. Since these operations deteriorate each time, excessive data updates reduce the lifespan of the flash memory.

Therefore, efforts have been made to improve the performance of flash memory by circumventing these shortcomings such as various methods of flash translation layer (FTL).

On the other hand, Phase Change Random Access Memory (PCRAM), a next-generation nonvolatile memory, enables random writes in bytes, and there is no restriction that data must be deleted first for data update. Shows read performance.

However, there are drawbacks to phase-conversion memory, so using flash memory and phase-conversion memory together to complement each other can create data storage systems and data management methods that provide better performance than single-use.

In this case, special processing of hot data with high update frequency may further improve the performance of the data storage device.

In this regard, Korean Patent Registration No. 10-2010-0037789 ("memory device and a management method of a memory device") discloses a configuration in which a data block for storing data and a log block for storing updated values of the data are disclosed.

In addition, Korean Patent No. 10-2010-0057346 ("memory device and a method of operating the memory device") includes a flash memory and NVRAM, NVRAM is a sequential write if the data size of the write command is larger than the threshold, random if the threshold is below The writing configuration is disclosed.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and an object thereof is to provide a data storage system and a data management method which provide improved performance through special processing of data with high complementary and update frequency of flash memory and phase conversion memory. will be.

A data storage system according to a first aspect of the present invention for achieving the above object includes a first volatile memory in which data is stored; And a second volatile memory sequentially storing the updated contents of the data in an order of occurrence in an update contents storage space dynamically allocated according to the update frequency of the data, wherein the updated contents are stored in the first volatile memory. When the data is updated in the upper memory, it is characterized in that the difference between the data before the update and the updated data.

A data management method using a data storage system including a first volatile memory and a second volatile memory according to a second aspect of the present invention for achieving the above object is (a) data stored in the first volatile memory Loading data to which the update contents stored in the update contents storage space of the second volatile memory are applied to the upper memory in the second memory; (b) storing updated contents of the data loaded and updated in the upper memory in an updated contents storage space of the second volatile memory; (c) determining whether the updated data is hot data which is frequently updated; (d) additionally allocating the update content storage space for data whose determination result of step (c) is hot data; And (e) performing an update merge operation to update the data stored in the first volatile memory by sequentially applying the update contents to the data stored in the first volatile memory in the order of occurrence, and performing the update merge operation. And emptying the update content storage space, wherein the update content is a difference between the data before updating and the updated data when the data stored in the first volatile memory is updated in the upper memory.

The present invention achieves the effect of improving performance by configuring the flash memory and the phase change memory to complement each other in the data storage system and the data management method.

Specifically, when updating data, instead of accessing the flash memory where the data is stored, the contents of the data are stored in the phase conversion memory and the flash memory data is reduced, thereby improving the random writing speed of the flash memory. Get

The high speed of writing small amount of data in the phase conversion memory can be utilized, and the data storage time is shorter than storing the update contents in the log area of the flash memory.

Since the number of flash memory data updates is reduced, the disadvantage of flash memory, which must be erased first for writing and a large amount of data must be written / deleted at a time, avoids the deterioration of the area storing frequently updated data. This can prevent the flash memory from shortening its lifespan.

In addition, since the log area is not fixedly allocated to each page in the phase change memory, the utilization of the phase change memory space is high, the load due to the mapping is low, and deterioration is increased in a specific area, thereby preventing the life of the phase change memory from being reduced. .

In addition, because it does not require complex operations, resource utilization is efficient, good performance, and easy to implement.

In addition, the update storage space for hot data, which is frequently updated, increases dynamically, further improving performance.

In addition, since the update storage space is dynamically allocated, deterioration of a specific area of the phase change memory can be further reduced.

1 shows a data storage system according to the invention.
2 shows an internal structure of a data storage system according to the present invention.
3 shows a comparison of the speed differences of various storage media.
4 illustrates a flow of a data management method according to the present invention.
Figure 5 shows the flow of the data loading step into the main memory of the data management method according to the present invention.
Figure 6 shows the flow of the hot block processing step of the data management method according to the present invention.
Figure 7 shows the flow of the update content merging step into the flash memory of the data management method according to the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, which will be readily apparent to those skilled in the art. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when a part is referred to as being "connected" to another part, it includes not only "directly connected" but also "electrically connected" with another part in between . Also, when an element is referred to as "comprising ", it means that it can include other elements as well, without departing from the other elements unless specifically stated otherwise.

1 shows a data storage system according to the invention.

The data storage system 10 according to the present invention includes a flash memory 100, which is a nonvolatile storage medium for storing data, and a phase change memory 200, which is a nonvolatile storage medium for storing update contents of data.

There is no limitation on the configuration of the data storage system 10 according to the present invention. For example, in one embodiment, the flash memory 100 may be configured in the form of a solid state disk (SSD), and the phase change memory 200 may be configured in the form included in the SSD. Alternatively, the flash memory 100 and the phase change memory 200 may be configured independently of each other. In addition, there is no limitation on the type, number and capacity of the flash memory 100 and the phase change memory 200. The data storage system 10 may also be used in connection with other auxiliary storage devices such as a hard disk.

Data stored in the data storage system 10 is loaded into an upper memory such as main memory (RAM) and used. In an exemplary embodiment, the main memory may include a dedicated buffer 20 for reading and using data stored in the data storage system 10, more specifically, a data area 300 in the buffer 20. In addition, the main memory buffer 20 may further include a log area 400 for storing updated contents when data loaded in the data area 300 is updated.

2 shows an internal structure of a data storage system according to the present invention.

Data is stored in the flash memory 100. As described above, the flash memory 100 reads and writes data in units of pages rather than bytes, and data erasing is performed by a block unit that is a set of pages.

As shown in the figure, if the i-th block of the flash memory 100 is Bi and the j-th page of each block is Pj, the j-th page of the i-th block may be represented as BiPj. Where BiPj is a logical address, not a physical address.

Since the frequency of updating the flash memory 100 data will be different for each block, a block having a high update frequency may be referred to as a hot block. For example, in the drawing, the first block B1 of the flash memory 100 is shown as a general block 102 and a second block B2 as a hot block 104.

The phase change memory 200 stores update contents of data stored in the flash memory 100. The update content is a difference between the data before updating and the updated data when the data is loaded into the upper memory (main memory) and updated, and may be referred to as difference data or log. That is, the update content may be defined as the difference between the pre-update version and the post-update version of the data.

The phase change memory 200 stores update contents for each block of the flash memory 100, and dynamically allocates more space to store update contents of the hot block 104. Detailed reasons for the dynamic allocation and the effects thereof will be described later, and an implementation method and a structure of the phase conversion memory 200 will be briefly described first.

The data storage system 10 according to the present invention uses the phase conversion memory 200 and the base area 202 and the spare area in order to obtain an effect of dynamically allocating update storage space while reducing the operation load caused by the dynamic allocation. Manage by dividing by (204). The phase change memory 200 fixedly allocates update storage space for each block of the flash memory 100 to the basic area 202, and additionally updates the space in the spare area 204 for the hot block 104. Allocate more storage space.

Since each block of the flash memory 100 and each update storage space of the phase-change memory 200 correspond one-to-one in the basic area 202, pages belonging to the i-th block Bi of the flash memory 100 may correspond. The update storage space allocated to the phase-conversion memory 200 with respect to may be represented as Li. On the other hand, each of the spaces R1, R2, ..., Rk, ... prepared in the spare area 204 is dynamically selected and allocated to the hot block 104. For example, when the second block B2 of the flash memory 100 is determined to be a hot block 104 in a state where there is no block previously determined as the hot block 104, the update storage space for the block is In addition to L2, R1 may be additionally allocated. The allocation time point is when L2 is full. If R1 is full, R2 may be additionally allocated.

In this case, the data storage system 10 according to the present invention manages an update content storage space additionally allocated using a pointer. According to an embodiment, the pointer may be configured to point to R1 at L2, R2 at R1, and vice versa. For convenience, in one embodiment, a space for storing such a pointer may be provided in the main memory buffer 20.

The reason why the data storage system 10 stores only the updated contents in the phase-conversion memory 200 without directly overwriting the updated data with the data stored in the flash memory 100 is as follows.

Data stored in the flash memory 100 is loaded into the main memory buffer 20 as described above. Since data is stored in the flash memory 100, the data unit loaded into the data area 300 of the main memory buffer 20 is a page. On the other hand, the unit in which the loaded data is updated is bytes, and typically only a fraction of the entire page is changed. For example, when applied to a database system, it is known that the data updated by a database operation is only an average of tens of bytes.

Considering that the page unit of a flash memory medium is usually larger than 2 KB and its capacity is becoming larger, the entire page is updated by accessing the flash memory 100 in order to apply the update content of only a few tens of bytes. Overwriting is very inefficient and has several disadvantages. As described above, the flash memory 100 may perform a write operation in units of blocks and then perform a write operation in units of pages, so that data may be updated so that changes of tens of bytes are applied, thereby further degrading performance. There is also a problem in that the memory life is reduced every time a degradation occurs.

The data storage system 10 according to the present invention effectively solves the above-mentioned problems by collecting the update contents in the phase conversion memory 200 without immediately updating the data stored in the flash memory 100 when the data is updated. Solve. The phase change memory 200 is a non-volatile storage medium capable of writing in bytes, showing a write speed close to DRAM for small amounts of data, and eliminating the need to perform a deletion before writing. Therefore, the phase change memory 200 is suitable for storing update contents. . Therefore, the data storage system 10 according to the present invention has better performance than the prior art in which a log area for each page is provided in the flash memory itself.

For example, if 2 KB of B1P1 data is loaded in the data area 300 of the main memory buffer 20, and 70 bytes of this page are changed by the database operation and 32 bytes are changed again, the phase change memory ( It is only necessary to sequentially record the update contents of 70 bytes and the update contents of 32 bytes in the L1 area of 200).

This keeps a record of the update contents without updating the data in the flash memory 100, so that the latest version of the data can be calculated when the data is needed. This is an operation performed when data is loaded into the data area 300 of the main memory buffer 20. For example, after reading the data from the flash memory 100, the updated contents are read from the L1 region of the phase conversion memory, 70 bytes of updated contents are applied, and 32 bytes of updated contents are applied again.

Whenever there is a read request, the cost of going through this latest version reconstruction process is low compared to the advantages such as a significant improvement in write speed as described above. This is because the phase-change memory reads the update content because the read speed of the small amount of data is also fast. Consideration of the trade-offs for faster write speeds and slower read speeds will be discussed later.

Accordingly, the data storage system 10 according to the present invention appends and stores the updated contents in the order of occurrence when the updated contents are stored in the phase conversion memory 200. This is a characteristic that has several advantages. Since there is no fixed size and position for each page, there is no computational load such as mapping. There is also no problem of deterioration in a particular area. Even if the end-of-life cell is a part of the entire memory, the entire memory cannot be used. Therefore, when using a storage medium that exhibits a deterioration phenomenon in which the life of the memory cell decreases every time the flash memory 100 or the phase change memory 200 is used. It is desirable to be careful not to deteriorate the specific area, but the data storage system 10 according to the invention has this advantage.

In addition, since the update contents are simply added to the end of the previously stored data and do not require any special operation, the update contents are collected in the log area 400 of the main memory buffer 20, and the phase conversion memory is stored at once. It becomes possible to write to 200. In such an embodiment, since I / O operations are further reduced, the performance may be further improved.

However, as described above, the data storage system 10 according to the present invention does not allocate the update contents storage space of the phase change memory 200 fixedly for each page, but allocates them separately for each block. The fixed allocation for each page is not preferable because of the above-mentioned reasons, but it is also undesirable to use the entire phase conversion memory 200 without any distinction. For example, if you simply store the update contents sequentially from the beginning of the medium without block division, you will suffer from the above-mentioned deterioration weighting problem.

More importantly, there arises a problem that the above-described read speed decreases. For example, if there are 2 updates for B1P1, 1 update for B1P2, 3 updates for B2P1, 1 update for B3P5, and 2 updates for B3P7, If all of these updates are stored sequentially in order of occurrence without block division, a total of 9 updates must be read to return the latest version of B1P1. This means that not only speed, but also the required temporary memory space can be increased. On the other hand, in case of storing by dividing by block, it is efficient because only 3 update contents need to be read.

That is, the main reason why the data storage system 10 according to the present invention fixedly allocates an area for storing update contents on a block basis is a trade-off between read and write. Accordingly, the allocation area threshold for each block may be set. Assuming that the page size of the flash memory 100 is 2KB, the number of pages per block is 32, and the read and write frequencies are the same, the read burden of the phase-conversion memory 200 = (differential data area size / 2/32) * Read speed, write burden = write speed. When calculating by applying the read speed and the write speed shown in FIG. 3, it is concluded that the difference data area size is approximately 2 KB or less.

Since the update content storage space is fixedly allocated for each block, an update content storage space in which the update content for the block can no longer be stored because of full data may occur. In this case, the data stored in the flash memory 100 of the block may be updated to the latest version in which the update contents are merged and then the corresponding update contents storage space of the phase change memory 200 may be emptied. That is, the update of the data stored in the flash memory 100 is made at this point.

However, in the case of the hot block 104, such update merging operation is postponed, and as described above, an additional update storage space is further allocated in the spare area 204.

The data storage system 10 according to the present invention can further improve performance in addition to the aforementioned effects by dynamically allocating additional update storage space dynamically for the hot block 104. This is because hot blocks 104, which frequently occur within a certain time window (e.g., the last hour), are often only a fraction of the total number of blocks. It is known that about 20% of the total updates occur for 5% of data. Thus, dynamically finding such hot blocks 104 and allocating additional update storage areas for them can further improve performance.

In addition, as described above, a phenomenon in which deterioration is increased in a specific area of the phase conversion memory 200 may be reduced. Since the hot block 104 is a block in which data is frequently updated, data must be frequently recorded in the update contents storage space of the corresponding phase conversion memory 200. Therefore, corresponding memory cells have a shorter lifespan than other memory cells. Therefore, in order to solve the lifetime problem of the storage medium as described above, it is desirable to prevent deterioration in the area concerned. In other words, by reducing the recording load on the update storage space through the additional update storage space allocation, the data storage system 10 according to the present invention allows the phase conversion memory 200 to be used longer.

At this time, the size of the additional update content storage space to be provided in the spare area 204 is set in consideration of the performance and the trade-off for the space. For example, in one embodiment, the size of the entire spare area 204 may be set within 10% of the total number of data blocks.

In order to additionally allocate the update content storage space to the hot block 104, a method of determining whether or not the hot block 104 is required is required. Since the hot block 104 is a block having a high update frequency, that is, a block in which updates are frequently performed within a specific time window, a time stamp in which the corresponding update contents are written to each update contents written in the phase-change memory 200 is performed. By additionally including the information, it may be determined whether the block is a hot block 104.

In this case, the determination of whether or not the hot block 104 may be performed only when the current update contents storage space in which the update contents are to be stored, that is, the update contents storage space most recently allocated by the corresponding block is full. Whether or not to allocate additional update storage space is only necessary when there is no more space available to store new update contents in the current update storage space. The most recently allocated update storage is full, which means that all update storage allocated to that block is full. In this case, the current update content storage space may be an update content storage space basically allocated in the basic area 202 or may be an update content storage space additionally allocated in the spare area 204.

For example, if the second block B2 of the flash memory 100 is the hot block 104, and R1 and R2 are further allocated in the spare area 204 in order, the current L1 and R1 are full and R2 is present. The update contents of the block B2 will be stored in the. At this point, if R2 is full, it is determined at this point whether the block B2 is still a hot block 104. If it is a hot block 104, an additional space (for example, R3) is further allocated, and at this time, In the non-hot block 104, the update data merging operation may be performed by updating the original data stored in the flash memory 100 to the latest version and then emptying the update storage spaces L1, R1, and R2. In this case, of course, R1 and R2 will be returned to the available space of the spare area 204.

In more detail, the determination whether the hot block 104 is performed is performed by comparing the time point information included in the oldest update content store in the update storage space R2 with the time point information of the current update to be stored, and being within a specific threshold. The block is determined as a hot block 104. At this time, the threshold value can be set by the user. For example, when the set threshold is 1 hour, when the oldest update storage in the update storage space R2 is stored 2 hours and 30 minutes ago, it is determined that it is not the hot block 104, and 30 minutes before In the case of the stored update contents, it is determined as the hot block 104.

If it is determined that the block B2 is the hot block 104, an additional space must be allocated as described above. The additional space of the spare area 204 is dynamically allocated unlike the base area 202, so any update is made. It is necessary to select whether to allocate the content storage space to the corresponding block B2. In particular, since the available space in the spare area 204 is limited, it is necessary to deal with it when there is no more available additional space that can be allocated.

If there is available space in the spare area 204, the method of selecting one is up to the implementor. For example, in one embodiment the first available space found in the spare area 204 can be selected (eg, R3 in the above example).

If there is no available space because all of the update storage space in the spare area 204 is in use, the block in which the update write has not occurred for the longest time among the data blocks allocated the additional update storage space in the spare area 204. Is selected as the victim to return to the basic block 102 rather than the hot block 104. That is, the victim is the least hot block of the existing hot blocks, that is, the block with the longest update. When the above-described update merging operation is performed on the victim, the victim may return the update storage space used in the spare area 204 to the available space. Accordingly, the update storage space in the spare area 204 used by the victim may be allocated to the current hot block 104.

Briefly summarizing the above operation, when the write request is caused by the data update in the main memory buffer 20, the data storage system 10 according to the present invention uses the original data stored in the flash memory 100 as updated data. Instead of overwriting, only update contents are stored in the current update contents storage space (which may be in the primary region 202 or in the additional region allocated in the spare region 204) allocated to the corresponding block of the phase-change memory 200. do.

At this time, if the update storage space is full, it is determined whether the block is a hot block (104). If it is not the hot block 104, the flash memory is the latest version that is reconstructed by sequentially merging all updates from all allocated update storage spaces (Li and one or more dynamically allocated storage spaces allocated dynamically) in order of occurrence. After updating the original data of (100), all allocated update storage spaces (Li and one or more dynamically allocated storage spaces allocated dynamically) are emptied, and if the additional space has been allocated, the space is reserved. Return the available space of.

In the case of the hot block 104, an additional space is allocated in the spare area 204. If there is no available space in the spare area 204, the hot block 104 is the least hot of other blocks using the spare area 204. The block, that is, the block whose update has been stored for the longest time, is allocated the space returned by the block selected as the victim to return to the basic block 102.

That is, update of original data through merging of updates is performed when it is necessary to free update storage space. Specifically, the hot data is not a hot block 104 or another hot block ( For the block that is selected as the victim for 104 and returns to basic block 102.

4, 5, 6, and 7 show the flow of the data management method according to the present invention.

4 shows the overall flow.

When the j-th page (BiPj) data of the i-th block Bi of the flash memory 100 to be loaded into the main memory buffer 20 is requested (S410), the contents of the request page (BiPj) stored in the flash memory 100 For the request page BiPj of the update contents stored in all update storage spaces (Li and one or more dynamically allocated storage spaces allocated dynamically) allocated to the corresponding block Bi of the conversion memory 200. The latest version of the data to which the update contents are sequentially applied is returned, and the data is loaded into the data area 300 of the main memory buffer 20 (S420).

When the page BiPj is updated in the main memory buffer 20 (S430), the update contents are recorded in the log area 400 of the main memory buffer 20 (S440), and the log area 400 is full or the checkpoint is updated. When it is necessary to write the update content data in the nonvolatile storage device for reasons such as an operation (S450), the page BiPj original data stored in the flash memory 100 is not updated, but the phase-conversion memory 200 is instead updated. The update content is stored in the current update area storage space (Li or most recently allocated update area storage space) of the device (S460). At this time, as described above, the data is stored in addition to the update contents data previously stored in the order of occurrence without overwriting.

If hot block processing is required (S470), an additional update content storage space is allocated to the hot block 104 (S480), and if update merge processing is required (S490), all update contents are stored in the original of the flash memory 100. Merge the data to free up the update storage area of the block.

As described above, hot block processing or update content merging processing is performed for blocks in which the current update content storage space is full when an update content write request is made. Compares the time point information included in the oldest update content stored in the current update content storage space with the time point information included in the write requested update content, and determines whether the corresponding block is hot. If the block processing operation is performed, and if not the hot block 104, the update content merging processing operation is performed, and the hot block 104 processing may include the update content merging processing operation for the victim block.

FIG. 5 shows the flow of the data loading step (420 in FIG. 4) into the main memory of the data management method according to the present invention.

The original data of the page requested to be read (BiPj) is read from the flash memory 100 (S510), and all update storage spaces Li and allotted for the block Bi belonging to the page in the phase-conversion memory 200 are included. Read all updates stored in the dynamically allocated one or more update storage space (S520), and sequentially update only the corresponding read request page (BiPj) to the original data read in the step (S510). If applied to (S530), since the page (BiPj) data is reconstructed to the latest version, it can be returned.

6 shows the flow of the hot block processing step (480 of FIG. 4) of the data management method according to the present invention.

Check whether there is available space in the spare area 204 (S610), if not, after merging the update contents as described above, the victim is selected to return to the basic block 102, and the additional space allocated to the victim is returned. An available space is provided in 204 (S620).

One of the available spaces in the spare area 204 is allocated to the hot block 104 as an additional update content storage space (S630).

7 shows the flow of the update content merging step (480 of FIG. 4) of the data management method according to the present invention.

First, an empty block is allocated to the flash memory 100 (S710).

After all of the update contents are read from all allocated update storage spaces (Li and one or more dynamically allocated update storage spaces) allocated for the merge required block Bi of the phase-change memory 200 (S720), Until all the pages of the required block Bi are updated (S750), the updated contents of the corresponding pages are sequentially applied to each page of the block (S730), and the page updated to the latest version is recorded in the allocated empty block. (S740).

The block mapping information is changed in the flash memory 100 and the previous block is returned as an empty block (S760).

In this case, since each page of the corresponding block Bi of the flash memory 100 has been updated and stored as the latest version, it is no longer necessary to maintain the existing update content, and thus the corresponding area allocated in the basic area 202 of the phase-change memory 200 is maintained. The default update storage space (Li) for the block is available again from the beginning, and if more than one additional update storage space has been allocated in the spare area 204, it is returned to the spare area 204 as available space. can do.

The foregoing description of the present invention is intended for illustration, and it will be understood by those skilled in the art that the present invention may be easily modified in other specific forms without changing the technical spirit or essential features of the present invention. will be. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive. For example, each component described as a single entity may be distributed and implemented, and components described as being distributed may also be implemented in a combined form.

The scope of the present invention is shown by the following claims rather than the above description, and all changes or modifications derived from the meaning and scope of the claims and their equivalents should be construed as being included in the scope of the present invention. do.

10: Memory system
100: flash memory
200: phase change memory

Claims (25)

In a data storage system,
A first nonvolatile memory in which data is stored; And
And a second nonvolatile memory configured to sequentially store update contents of the data in an order of occurrence in an update contents storage space dynamically allocated according to the update frequency of the data.
The update content is a difference between the data before updating and the updated data when the data stored in the first nonvolatile memory is updated in the upper memory,
The update storage space of the second nonvolatile memory is allocated for each block of the first nonvolatile memory. The method of claim 1,
And when the data is requested, return the latest version reconstructed by sequentially applying all updates to the requested data stored in the second nonvolatile memory to data stored in the first nonvolatile memory. The method of claim 1,
Sequentially applying the update contents to the data stored in the first nonvolatile memory in order of occurrence, and performing an update merge operation to update the data stored in the first nonvolatile memory, and performing the update through the update merge operation. Data storage system that frees up content storage. The method of claim 1,
Data stored in the first nonvolatile memory is accessed in units of pages, and the update contents are accessed in units of bytes. The method of claim 1,
Wherein said update content is variable in size and said storage location in said update storage space of said second non-volatile memory is not fixed. The method of claim 1,
And the update content includes information on when the update occurred. The method of claim 1,
The update content storage space may include a basic update content storage space allocated to the second nonvolatile memory for each block of the first nonvolatile memory; And
And additional update storage space that can be further dynamically allocated in the spare area of the second non-volatile memory. The method of claim 7, wherein
Comparing the time point of occurrence of the oldest update content in the update content storage space to which the update content is most recently allocated and the current time point, and the difference between the time points corresponding to the update content storage space within a preset threshold; 1 A data storage system that determines a nonvolatile memory block as a hot block. The method of claim 8,
And the additional update content storage space is allocated to a block that is a hot block as a result of the hot block determination. The method of claim 8,
The determination of whether or not the hot block is performed when an update content storage space for storing the update content is full. The method of claim 8,
And a data storage system for performing the merging of updated contents on a block other than a hot block as a result of determining whether the hot block is present. The method of claim 8,
If there is no available space that can be allocated as an additional update storage space in the spare area of the second nonvolatile memory, the update content is stored the longest time among the blocks that have been allocated additional update storage space in the spare area. And an available space in the spare area by performing the merging of updated contents on the blocks formed. The method of claim 7, wherein
And a pointer for accessing the additional update content storage space. The method of claim 1,
And data stored in the first nonvolatile memory is database data. The method of claim 1,
The first nonvolatile memory is a flash memory, and the second nonvolatile memory is a phase change random access memory (PCRAM). A data management method using a data storage system including a first nonvolatile memory and a second nonvolatile memory,
(a) loading data to which the update contents stored in the update contents storage space of the second nonvolatile memory into the upper memory to the data stored in the first nonvolatile memory;
(b) storing updated contents of the data loaded and updated in the upper memory in an updated contents storage space of the second nonvolatile memory;
(c) determining whether the updated data is hot data which is frequently updated;
(d) additionally allocating the update content storage space for data whose determination result of step (c) is hot data; And
(e) sequentially applying the update contents to the data stored in the first nonvolatile memory, in order of occurrence, to perform update merge operation to update the data stored in the first nonvolatile memory, and to perform the merge operation of the update contents. Including the step of emptying the update storage space;
The update content is a difference between the data before updating and the updated data when the data stored in the first nonvolatile memory is updated in the upper memory,
The update management space of the second nonvolatile memory is allocated for each block of the first nonvolatile memory. 17. The method of claim 16,
Data stored in the first nonvolatile memory is accessed in units of pages.
The data stored in the second nonvolatile memory is accessed in units of bytes, has a variable size, and has a fixed location in the second nonvolatile memory. 17. The method of claim 16,
And the update content includes information on when the update occurred. 17. The method of claim 16,
The step (c)
The data management method performed when all the update contents storage space allocated to the updated data are full. 17. The method of claim 16,
The step (e)
And the determination result of step (c) is performed on data other than hot data. 17. The method of claim 16,
The step (e)
The data management method is performed when there is no update content storage space that can be additionally allocated to data whose determination result in step (c) is hot data. 17. The method of claim 16,
The step (e)
Data management method performed on the oldest update of existing hot data. 17. The method of claim 16,
The step (c)
And determining that hot data is a difference between a time point of the oldest update content in the most recently allocated update content storage space for the updated data and a current time point or less. 17. The method of claim 16,
The data stored in the first nonvolatile memory is database data. 17. The method of claim 16,
And the first nonvolatile memory is a flash memory and the second nonvolatile memory is a phase change random access memory (PCRAM).

Images