KR101554550B1 - Memory management apparatus and control method thereof - Google Patents

Abstract

The present invention discloses a technology capable of not only improving a whole performance of a storage system, but also minimizing IO overhead caused by data scrubbing by performing a garbage collection which is performed in a flash-based device like a SSD necessarily and the data scrubbing which is performed to improve data reliability based on a storage system. A memory control device includes: a data reading unit for reading a particular data segment recorded in a memory device when a garbage collection performing event occurs; a defect state determining unit for determining a data defect state related to the particular data segment while reading the particular data segment; and a memory control unit for recording after restoring particular data once the particular data, determined as a data defect among effective data by reflecting a result of the data defect state determination, is verified when performing the garbage collection, which records valid data with respect to the particular data segment in another position in the memory device.

Description

[0001] MEMORY MANAGEMENT APPARATUS AND CONTROL METHOD THEREOF [0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a storage system configured with a plurality of flash-based SSDs, and more particularly, to a storage system configured by a garbage collection which is essentially performed in a flash-based device such as an SSD and data scrubbing The present invention relates to a technique capable of minimizing IO overhead due to data scrubbing as well as enhancing overall performance of a storage system by performing a storage system based operation.

Recently, SSD (Solid State Drive) with improved performance compared to HDD (Hard Disk Drive) has been widely used, and a storage system adopting a plurality of flash-based SSDs has been configured to provide data service through a network .

For example, a storage system may serve as a SAN (Storage Area Network) / NAS (Network Attached Storage) to provide a data service by simply supporting a block-level storage service, It can also provide data services directly (eg DB, Web service) by operating an application server.

In such a storage system, it is very important to ensure the reliability of recorded (stored) data.

Accordingly, a data scrubbing technique is employed as a technique employed for ensuring data reliability in a storage system.

However, since the conventional data scrubbing method periodically performs data scrubbing on all the data stored in the storage medium (disk, flash memory) in the storage system, the performance of the storage system deteriorates due to the occurrence of the IO overhead There is a problem.

On the other hand, in the case of a flash-based device such as an SSD, a flash memory (page) which is wasted by arranging invalid data recorded in the flash memory due to the characteristics of the flash memory is secured as a usable flash memory area (Garbage Collection) function.

However, in order to perform this garbage collection function, an operation (IO) of reading and writing data in a flash-based device is indispensable. However, the internal IO of such a device causes a problem of deterioration in performance from the outside of the device.

Therefore, in the present invention, data scrubbing performed for enhancing garbage collection and data reliability, which are essentially performed in a flash-based device such as an SSD, is performed based on a storage system, thereby improving overall performance of the storage system, We propose a method to minimize IO overhead.

SUMMARY OF THE INVENTION The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide data scrubbing, which is performed in order to improve garbage collection and data reliability, Thereby improving overall performance of the storage system and minimizing IO overhead due to data scrubbing.

According to a first aspect of the present invention, there is provided a memory control apparatus for reading a specific data segment recorded in the memory device when a garbage collection performing event for a memory device occurs A data lead portion; A defect determination unit for determining whether a data defect is present in the specific data segment during the reading of the specific data segment; And a memory controller for performing garbage collection on the specific data segment by reflecting a result of determining whether the data is defective.

Advantageously, the memory device comprises a plurality of flash-based memory chips, and the memory chips may be a solid state drive (SSD).

Preferably, the specific data segment may be composed of two or more pieces of data distributedly recorded in a plurality of memory chips in the memory device, and a hash value calculated in the distributed recording for each of the two or more pieces of data.

Preferably, the defect determination unit may calculate a hash value for each of the two or more data being sequentially read by the data lead unit, and for each of the two or more data, the calculated hash value is stored in the hash value And if it is different, it can be judged as a data defect.

Preferably, the specific data segment includes two or more data distributedly recorded in a plurality of memory chips in the memory device, and the memory control unit selects valid data from the two or more data, If the specific data determined as a data defect among the valid data is confirmed by reflecting the determined result, a new data segment including valid data may be restored and recorded in the memory device after recovering the specific data.

Preferably, the memory device stores the same original or duplicate data segment as the specific data segment, and the memory control unit stores the specific data as corresponding data corresponding to the specific data of the original or duplicated data segment Alternatively, the specific data can be recovered.

Advantageously, the new data segment can be configured such that the valid data is written to a memory chip in which a plurality of memory chips in the memory device had been written before performing the garbage collection.

Preferably, the defect determination unit may determine whether a data segment in which the garbage collection is not performed in the memory device is read from the memory device to determine whether or not a data defect exists, so that a predetermined time elapses.

According to a second aspect of the present invention, there is provided a method for operating a memory controller, the method comprising: when a garbage collection execution event for a memory device occurs, data for reading a specific data segment recorded in the memory device Lead stage; Determining whether a data defect is present in the specific data segment during the reading of the specific data segment; And a garbage collection control step of performing garbage collection on the specific data segment by reflecting a result of determining whether the data is defective.

Preferably, the specific data segment includes two or more pieces of data distributedly recorded in a plurality of memory chips in the memory device, and the garbage collection control step may include selecting valid data from the two or more pieces of data, And if the specific data determined as a data defect among the valid data is identified, the new data segment including the valid data may be restored to the memory device after restoring the specific data.

Preferably, the memory device records the same original or duplicate data segment as the specific data segment, and the garbage collection control step stores the specific data in correspondence with the specific data of the original or duplicated data segment Data can be replaced with data to recover the specific data.

According to the memory control device and the operation method of the memory control device of the present invention, data scrubbing performed for enhancing garbage collection and data reliability, which are essentially performed in a flash-based device such as an SSD, Not only improves the overall performance of the system but also minimizes the IO overhead due to data scrubbing.

1 is a block diagram illustrating a configuration of a storage system including a memory controller according to a preferred embodiment of the present invention.
2 is a block diagram illustrating a memory control apparatus according to an exemplary embodiment of the present invention.
3 is a flowchart illustrating an operation method of a memory control apparatus according to a preferred embodiment of the present invention.
FIGS. 4 and 5 illustrate a process of performing garbage collection and data scrubbing on a specific data segment according to a preferred embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

First, a configuration of a storage system including a memory controller according to a preferred embodiment of the present invention will be described with reference to FIG.

1, the storage system includes a memory device 10 as a storage space, a file system or application (hereinafter referred to as a file system 200) for attempting memory operation to the memory device 10, And a memory control device (100) for controlling the memory device (10) between the file system (200) and the memory device (10).

At this time, the memory device 10 includes a plurality of flash-based memory chips, wherein the memory chip is preferably a flash-based device such as a solid state drive (SSD).

That is, the memory device 10 is composed of a plurality of flash-based memory chips, that is, a plurality of flash-based SSDs, as shown in Fig.

The memory device 10 may employ a redundant algorithm (for example, a Redundant Array of Independent Disks (RAID) algorithm) for grouping memory chips, that is, SSDs, into a predetermined number of memory groups.

Hereinafter, it will be described that the memory device 10 employs a RAID algorithm.

The memory control device 100 may be a control module or control software that controls the memory device 10 between the file system 200 and the memory device 10. [

Currently, a memory array system having a plurality of arrays, that is, a storage system, is constructed by using a storage system employing a plurality of flash-based SSDs as described above as an array, and an external client computer Not shown) to provide data services.

In such a storage system, it is very important to ensure the reliability of recorded (stored) data.

Accordingly, a data scrubbing technique is employed as a technique employed for ensuring data reliability in a storage system.

The data scrubbing technique calculates and stores the data hash value for the data when the data is stored in the memory, and then reads the data stored in the memory periodically to calculate the hash value, (Judging) whether data is defective or not.

However, since the conventional data scrubbing method periodically performs data scrubbing for all the data stored in the memory in the storage system, the IO overhead is caused thereby, and the memory performance is deteriorated.

On the other hand, in the case of a flash-based device such as an SSD, a flash memory (page) which is wasted by arranging invalid data recorded in the flash memory due to the characteristics of the flash memory is secured as a usable flash memory area (Garbage Collection) function.

However, in order to perform this garbage collection function, an operation (IO) of reading and writing data in a flash-based device is indispensable. However, such an internal IO of the device causes a problem of deteriorating the performance from the outside of the device.

Therefore, in the present invention, data scrubbing performed for enhancing garbage collection and data reliability, which are essentially performed in a flash-based device such as an SSD, is performed based on a storage system, thereby improving overall performance of the storage system, We propose a method to minimize IO overhead, and specifically, we propose a memory control device that realizes this.

Hereinafter, a memory control apparatus according to an embodiment of the present invention will be described in detail with reference to FIG. For convenience of explanation, the description will be made using the reference numeral of the memory control device 100 corresponding to FIG. 1 described above.

The memory control apparatus 100 according to the present invention includes a data read unit 110 for reading a specific data segment recorded in the memory device 10 when a garbage collection execution event for the memory device 10 occurs, A defect judgment unit 120 for judging whether or not a data defect exists in the specific data segment while reading the specific data segment; and a defect judgment unit 120 for judging whether or not the specific data segment is defective, And a memory control unit 130 for performing a collection.

At this time, the memory device 10 includes a plurality of flash-based memory chips, wherein the memory chip is preferably a flash-based device such as a solid state drive (SSD).

That is, the memory device 10 is composed of a plurality of flash-based memory chips, that is, a plurality of flash-based SSDs, as shown in Fig. Hereinafter, for convenience of explanation, a memory chip and an SSD can be used in combination.

The memory device 10 may employ a redundant algorithm (e.g., RAID (Redundant Array of Independent Disks) algorithm) in which memory chips, i.e., SSDs, are grouped into a certain number . Hereinafter, it will be described that the memory device 10 employs a RAID algorithm.

The memory control device 100 according to the present invention may be in the form of a control module or control software for controlling a memory device 10 composed of a plurality of flash based SSDs, (E.g., a data segment buffer, a GC Gathering Queue, etc.) necessary to control the application.

The data read unit 110 reads a specific data segment recorded in the memory device 10 when a garbage collection execution event for the memory device 10 occurs.

The garbage collection may be performed whenever there is insufficient memory available in the memory device 10, and the above-described garbage collection execution event may mean that the memory device 10 is judged to be deficient in the available memory area.

That is, for example, the memory control device 100 monitors whether or not data exceeding a predetermined threshold level (hereinafter referred to as a GC threshold value) is recorded in the memory device 10, and when data exceeding the GC threshold value is recorded, It can be determined that a memory area usable in the memory device 10 is insufficient and that a garbage collection execution event has occurred.

In this case, the memory control device 100 executes the garbage collection function for the memory device 10.

As described above, the memory device 10 employs a plurality of memory chips or SSDs including a plurality of flash memory (pages).

Among the memory operations for flash memory, the most basic read / write operations are performed in page units (eg, 4 KB, 8 KB, 16 KB), and write operations are performed on pages in which data is not necessarily written, ie, erased pages. On the other hand, erasing during the memory operation for the flash memory is performed in units of blocks in which a plurality of pages are gathered. As described above, since the write operation unit and the erase unit are different, the garbage collection is indispensable in the memory device 10 configured by the flash-based device such as the SSD.

When the garbage collection execution event occurs, the memory control device 100 selects data (or data segment) to be rearranged in the memory device 10 and performs garbage collection on the selected data (or data segment) . Here, the selected data (or data segment) is referred to as victim data (or a victim data segment).

At this time, the victim data (or the victim data segment) to be rearranged is selected in accordance with the adopted garbage collection algorithm as data (or data segment) having a large amount of invalid data, and the victim data Data segment) is the same as that of the conventional method, so a detailed description will be omitted.

As a result, when the garbage collection execution event for the memory device 10 occurs, the data lead unit 110 stores the specific data segment recorded in the memory device 10, that is, the victim data segment selected by the garbage collection as described above Lead.

This is an essential operation for reading the data stored in the memory device 10, that is, the victim data segment, in order to perform the garbage collection function.

Hereinafter, for convenience of description, a specific data segment will be referred to as a victim data segment.

The defect judgment unit 120 judges whether or not there is a data defect with respect to the victim data segment while reading the specific data segment, that is, the victim data segment.

That is, the defect determination unit 120 intervenes in the operation of reading the victim data segment from the memory device 10 in accordance with the execution of the garbage collection as described above, and reads the victim data segment in the victim data segment And data scrubbing is performed to determine whether data is defective.

Hereinafter, a process of determining whether data is defective in the victim data segment during the process of reading the victim data segment as garbage collection is performed will be described in more detail.

The data segment is composed of a plurality of memory chips in the memory device 10, that is, two or more pieces of data distributedly recorded in a plurality of SSDs, and a hash value calculated in the distributed recording for each of the two or more pieces of data.

Therefore, the specific data segment, i.e., the victim data segment, will also be composed of two or more data distributedly recorded in a plurality of SSDs in the memory device 10 and a hash value calculated in the distributed recording for each of the two or more data.

The process of configuring such a data segment will be described with reference to FIG.

For example, the memory control apparatus 100 sequentially writes data requested to be written to a data segment buffer (not shown) every time a write request for the memory device 10 is received, The hash value is calculated for each data recorded in the data segment buffer (not shown).

Thus, the memory control device 100 can constitute a data segment including two or more pieces of data sequentially recorded in a data segment buffer (not shown) and a hash value calculated for each piece of data. At this time, the hash value is included in the last part of the data segment.

Thereafter, the memory control device 100 distributes the constituent data segments to a plurality of memory chips.

For example, as shown in FIG. 4, four memory chips are operated as a group according to a RAID algorithm, and hash values H_1, H_2, H_3, and H_4), the memory control device 100 distributes the data segment A thus configured to a plurality of memory chips, for example, four memory chips # 0, # 1, # 2, and # 3 Can be recorded. In this case, H_A in FIG. 4 denotes a hash value including the hash values H_1, H_2, H_3, and H_4 of the data 1, 2, 3, and 4, respectively.

Then, in the process of writing the data segment A to the memory device 10, for example, the memory chips # 0, # 1, # 2, and # 3, the memory control device 100 reads the data 1, 2, 3 or 4 is stored in a mapping table (not shown) so that the data can be found when a read request for data 1 or 2 or 3 or 4 is received.

4, when the data segment A is recorded in the memory device 10, for example, the memory chips # 0, # 1, # 2, and # 3, the memory control device 100 stores the data segment A It is possible to further record the duplicated data segment A 'in the memory device 10.

At this time, the memory control device 100 may write the data segment A 'to the memory chips # 0, # 1, # 2, and # 3 on which the original data segment A is written as shown in FIG. 4, It is also possible to record the data segment A 'in the memory chips # 4, # 5, # 6, and # 7 different from the segment A.

Hereinafter, as an example for convenience of description, an original data segment A and a data segment A 'are recorded on the same memory chips # 0, # 1, # 2 and # 3 as shown in FIG.

The data block A selected in the memory device 10 as a result of the execution of the garbage collection function as described above, that is, the victim data segment, is the data segment A of FIG. 4.

In this case, when the garbage collection execution event for the memory device 10 is generated, the data lead unit 110 described above determines the victim data segment of the memory device 10 selected by the garbage collection, that is, the data segment A leads.

For example, the data lead unit 110 sequentially reads the data 1, 2, 3, and 4 and the hash value H_A from the beginning to the end of the data segment A, can do.

In addition, the defectiveness judgment unit 120 judges whether or not there is a data defect in the data segment A while reading the victim data segment, that is, the data segment A.

That is, the defect determination unit 120 intervenes in the operation of reading the data segment A from the memory device 10 in accordance with the execution of the garbage collection as described above, and reads the data segment A in the data segment A And data scrubbing is performed to determine whether data is defective.

More specifically, the defectiveness determining unit 120 calculates a hash value for each of two or more data being sequentially read by the data lead unit 110, i.e., data 1, 2, 3, and 4, respectively.

That is, the defect determination unit 120 calculates the hash values (hash) of the data 1, 2, 3, and 4 in the process of sequentially reading the data 1, 2, 3, and 4 in the data segment A .

The defect determination unit 120 determines whether or not the hash value hash calculated according to two or more pieces of data 1, 2, 3, and 4 is different from the hash values H_1, H_2, H_3, and H_4 in the data segment A It is judged that there is a data defect.

For example, if the hash value (hash) calculated for the data 1 is equal to the hash value (H_1) of the data 1 in the data segment A and the hash value hash is the same as the hash value H_3 of the data 3 in the data segment A but the hash value hash calculated for the data 2 is different from the hash value H_2 of the data 2 in the data segment A, (Hash) calculated for the data segment A is different from the hash value (H_4) of the data 4 in the data segment A, it can be judged as a data defect for the data 2 and 4.

As described above, in the present invention, at the time of reading data (data segment) from the memory device 10 by garbage collection, which is essentially performed in the memory device 10 configured with a flash-based device such as an SSD, Data segment), data garbage collection is performed to judge whether a data defect is incidental, so that garbage collection and data scrubbing can be performed concurrently on the basis of a storage system rather than individually.

Thus, in the present invention, the IO overhead caused by data scrubbing can be effectively minimized.

If it is determined that the data 2 and 4 are defective as described above, the defectiveness determining unit 120 determines whether or not the data 2 and 4 are defective or the hash values H_2 and H_4 of the data 2 and 4 in the data segment A ) Is defective.

For example, the defect determination unit 120 identifies the original or duplicated data segment of the data segment A. In this case, since the data segment A is the original, the replicated data segment A 'can be confirmed.

Accordingly, the defect judgment unit 120 judges whether or not the calculated hash value (hash) differs from the hash value (H_2, H_4) in the data segment A ' (H_2, H_4) of the data 2 and 4 in the data segment A, which is not a data defect, in the same case.

The memory controller 130 performs garbage collection on a specific data segment, i.e., a victim data segment, by reflecting a result of determining whether the data is defective or not by the defectiveness determining unit 120. [

Referring to FIG. 4, the memory control unit 130 performs garbage collection on a data segment A as a specific data segment, that is, a victim data segment, by reflecting the result of determining whether a data defect exists.

More specifically, the memory control unit 130 selects two or more data of the data segment A, that is, data effective among the data 1, 2, 3, and 4.

For example, the memory control unit 130 can select data 1 and 2 as valid data and data 3 and 4 as invalid data.

At this time, selection of valid data among data 1, 2, 3, and 4 is selected according to the adopted garbage collection algorithm, and selection of such valid data is the same as that of the conventional method, so a detailed description will be omitted.

When the specific data determined as a data defect among the valid data is identified by reflecting the result of determining whether the data is defective, the memory controller 130 restores the specific data, and then generates a new data segment including valid data And writes them to the memory device 10. [

That is, the memory controller 130 reflects the result of determining whether the data is defective by the defectiveness determining unit 120, and determines whether there is data that is determined to be a data defect among the valid data (e.g., data 1 and 2).

If the defect determination unit 120 determines that the data 1, 2, 3, and 4 of data 2 and 4 are data defects as described above, the memory control unit 130 determines that the data is defective Specific data, i.e., data 2, can be identified.

In this case, after recovering the specific data determined as a data defect, that is, data 2, the memory control unit 130 constructs a new data segment including valid data, i.e., data 1 and recovered data 2, Can be recorded.

At this time, the memory control unit 130 can recover the data 2 using the original or duplicated data segment of the data segment A.

That is, when the specific data, i.e., data 2, determined as a data defect among the valid data 1 and 2, i.e., data 2, is confirmed, the memory control unit 130 checks the original or duplicate data segment of the data segment A. In this case, since the data segment A is the original, the replicated data segment A 'can be confirmed.

On the other hand, when the parity information capable of recovering the data 2 is stored in the memory device 10 in correspondence with the specific data determined as a data defect, that is, the data 2, the memory control unit 130 uses the parity information Data 2 can also be recovered. In this case, the recovered data 2 will also be referred to as data 2 'for convenience of explanation.

Then, the memory control unit 130 constitutes a new data segment including valid data, that is, data 1 and recovered data 2 (= data 2 ').

Here, the new data segment is configured so that the above-mentioned valid data 1, 2 (= 2 ') are recorded in the same memory chip recorded before performing the garbage collection among a plurality of memory chips in the memory device 10 desirable.

This is because, when valid data is recorded in the same memory chip previously recorded, it is possible to cope with the IO request that occurs later in parallel, thereby improving the memory performance.

To this end, the memory controller 100 of the present invention holds a GC gathering queue.

More specifically, the memory control device 100 can hold GC gathering queues # 0 to # 3 as shown in FIG. This corresponds to four SSDs, that is, four memory chips # 0 to # 3, in which the above-described data segment is recorded.

5, for the valid data, that is, the data 1 and the recovered data 2 (= data 2 '), the memory control unit 130 sets the data 1 to the memory chip For example, in the GC Gathering Queue # 0 corresponding to the memory chip # 0, and the recovered data 2 (= data 2 ') is stored in the GC Gathering Queue # 1 corresponding to the memory chip .

Accordingly, the memory control device 100 fetches the data 1 from the GC gathering queue # 0 and writes it to the memory location corresponding to the memory chip # 0 in the data segment buffer (not shown) 2 'in the memory segment corresponding to the memory chip # 1 in the data segment buffer (not shown), and writes the remaining data in the data segment buffer (not shown) 7) will be recorded.

Accordingly, when the data segment buffer (not shown) becomes full, the memory control device 100 calculates a hash value for each data recorded in the data segment buffer (not shown) as described above, A new data segment B including two or more pieces of data sequentially recorded on the data 1, 2 ', 5, 6, and 7 and a hash value H_B calculated for each piece of data may be configured.

Thereafter, the memory control device 100 distributes the data segment B over the memory chips # 0, # 1, # 2, and # 3 as shown in FIG.

At this time, the data 1 will be recorded on the same memory chip # 0 recorded before the garbage collection, and the data 2 'will be recorded on the same memory chip # 1 on which the data 2 has been recorded before the garbage collection.

5, when the data segment B is recorded in the memory device 10, for example, the memory chips # 0, # 1, # 2, and # 3, the memory control device 100 stores the data segment B The duplicated data segment B 'can be further recorded in the memory device 10. [

As described above, in the present invention, data scrubbing is performed at the time of reading data (data segment) by the garbage collection in the memory device 10 configured by a flash-based device such as an SSD, and data defect detection By performing the garbage collection reflecting the result, it is possible to minimize the IO overhead due to data scrubbing, optimize both the effect of garbage collection and the effect of data scrubbing, thereby improving the overall performance of the storage system.

On the other hand, the defect judgment unit 120 judges whether a data segment for which the garbage collection is not performed in the memory device 10 so that a certain period of time elapses is read from the memory device 10, .

Data (data segment) which is hardly overwritten after being written once by the memory device 10 is referred to as cold data (data segment), and data (data segment) with a large overwrite Quot; hot data " (data segment).

Here, although the garbage collection is frequently performed by selecting the hot data segment as the sorting target, the garbage collection is not performed for the cold data segment because it is not necessary to perform the garbage collection.

Therefore, for a cold data segment with little change in data validity, it is desirable to perform data scrubbing when a predetermined period of time has been reached.

The memory control apparatus 100 according to the present invention will be described in detail with reference to the memory control apparatus 100 according to the first embodiment of the present invention with reference to the data segment in which the garbage collection is not performed in the memory device 10, It is preferable to periodically perform data scrubbing to read data from the memory 10 and determine whether the data is defective.

Here, depending on the validity of the data making up the victim data segment, there may be an empty GC Gathering Queue among the GC Gathering Queues # 0 to # 3, and there may be a full GC Gathering Queue. It is acceptable to leave the data segment buffer (not shown) in an incomplete state (for example, not full state) due to an empty GC Gathering Queue before a full GC gathering queue occurs.

However, if a GC Gathering Queue that is full of GC Gathering Queues # 0 to # 3 occurs, the memory control device 100 will update the data in the full GC Gathering Queue instead of the empty GC Gathering Queue if there is an empty GC Gathering Queue It is preferable to record the vacated portion of the data segment buffer (not shown).

As described above, according to the memory control device of the present invention, data scrubbing is performed based on a storage system, which is performed for garbage collection and data reliability enhancement, which are essentially performed in a flash-based device such as an SSD, Not only improves performance but also minimizes IO overhead due to data scrubbing.

Hereinafter, an operation method of the memory control device according to the preferred embodiment of the present invention will be described with reference to FIG. For convenience of explanation, the reference numerals of FIG. 1 and FIG. 2 will be described.

The operation method of the memory control device 100 according to the present invention selects data (or data segment) to be rearranged in the memory device 10 when a garbage collection execution event (aka GC event) occurs (S100) S110). Here, the selected data (or data segment) is referred to as victim data (or a victim data segment).

Hereinafter, for convenience of explanation, the data segment A shown in FIG. 4 is selected as a victim data segment.

Accordingly, the operation method of the memory control device 100 according to the present invention reads the data segment A from the selected memory device 10 (S120).

Then, the operation method of the memory control device 100 according to the present invention selects data that is valid among two or more data of the data segment A, that is, data 1, 2, 3, and 4 (S130). For example, the operation method of the memory control device 100 according to the present invention can select data 1 and 2 as valid data and data 3 and 4 as invalid data.

Meanwhile, the operation method of the memory control device 100 according to the present invention performs data scrubbing to determine whether data is defective in the V data segment A while reading the victim data segment, that is, the data segment A in step S120 .

More specifically, in the operation method of the memory control device 100 according to the present invention, when two or more data constituting the data segment A, that is, data 1, 2, 3, and 4, are sequentially read in step S120, By calculating a hash value for the data, a hash value for each of the data 1, 2, 3, and 4 is calculated (S140).

That is, in the method of operating the memory control device 100 according to the present invention, in the process of sequentially reading the data 1, 2, 3, and 4 in the data segment A, hash.

The operation method of the memory control device 100 according to the present invention checks the hash values H_1, H_2, H_3, and H_4 of the data 1, 2, 3, and 4 from the hash value H_A in the data segment A S150).

The operation method of the memory control device 100 according to the present invention is a method of operating the memory control device 100 according to the first embodiment of the present invention in which the calculated hash value hash is divided into hash values H_1, H_2, H_3, and H_4) (S160).

For example, in the method of operating the memory control device 100 according to the present invention, the hash value (hash) calculated for the data 1 is the same as the hash value (H_1) of the data 1 in the data segment A, The hash value (hash) calculated for data 2 is the same as the hash value (H_2) of data 2 in data segment A and the hash value If the hash value (hash) calculated for the data 4 is different from the hash value (H_4) of the data 4 in the data segment A, it can be judged as a data defect for the data 2 and 4.

If the data 2 and 4 are found to be defective, the method of operation of the memory control device 100 according to the present invention determines whether the data 2 and 4 are defective or the data 2 in the data segment A , And the hash values (H_2, H_4) of 4 are defective.

For example, the method of operation of memory control device 100 according to the present invention identifies the original or replicated data segment of data segment A. In this case, since the data segment A is the original, the replicated data segment A 'can be confirmed.

Thus, the operation method of the memory control device 100 according to the present invention is such that, as described above, for the data 2 and 4 determined as a data defect, the hash value hash calculated as described above is stored in the data segment A ' (H1, H4) of the data 2 and 4 in the data segment A, which is not a data defect, is judged as a data defect if it is different can do.

Hereinafter, for the sake of convenience of explanation, the operation method of the memory control device 100 according to the present invention will be described as a case where the data is finally determined as a data defect in step S170 (S180).

Meanwhile, when the operation of the memory control device 100 according to the present invention is completed, if it is determined that the data is defective, the garbage collection is performed on the victim data segment, i.e., the data segment A, (After S190).

More specifically, the operation method of the memory control device 100 according to the present invention checks whether there is specific data determined as a data defect among the valid data selected in step S130, for example, data 1 and 2 (S190).

For example, in the operation method of the memory control device 100 according to the present invention, if data 1, 2, 3, and 4 of data 1 and 2 are determined to be data defects in steps S140 to S180, , It is possible to identify the specific data, that is, the data 2, which is judged to be a data defect in the data 2 (S190 Yes).

In this case, the operation method of the memory control device 100 according to the present invention is a method of recovering specific data determined as a data defect, that is, data 2, and then a new data segment including valid data, that is, data 1 and recovered data 2, And can be recorded in the memory device 10.

More specifically, the method of operation of the memory control apparatus 100 according to the present invention is a method of checking the replicated data segment A 'of the data segment A and comparing the data 2 with the corresponding data 2 corresponding to the data 2 of the replicated data segment A' Quot ;, so that the data 2 can be recovered (S200). In this case, the recovered data 2 means data 2 '.

Thereafter, the method of operation of the memory control apparatus 100 according to the present invention configures a new data segment, such as data segment B of FIG. 5, containing valid data, i. E., Data 1 and recovered data 2 (S210).

Here, the new data segment B is configured such that the above-mentioned valid data 1, 2 (= 2 ') are recorded in the same memory chip that was recorded before performing the garbage collection among a plurality of memory chips in the memory device 10 .

More specifically, the operation method of the memory control apparatus 100 according to the present invention can hold GC gathering queues # 0 to # 3 as shown in FIG. This corresponds to four SSDs, that is, four memory chips # 0 to # 3, in which the above-described data segment is recorded.

5, for the valid data, that is, the data 1 and the recovered data 2 (= data 2 '), the data 1 is the original data (= Data 2 ') is stored in the memory chip, for example, the memory chip # 1 corresponding to the memory chip # 0 in which the data 1 has been recorded, and the recovered data 2 And stores it in the GC Gathering Queue # 1 corresponding to the GC Gathering Queue # 1.

Accordingly, the operation method of the memory control device 100 according to the present invention is to write the data 1 from the GC gathering queue # 0 to the memory location corresponding to the memory chip # 0 in the data segment buffer (not shown) Data 2 'is fetched from the GC Gathering Queue # 1 and is written to the memory location corresponding to the memory chip # 1 in the data segment buffer (not shown), and the remaining data in the data segment buffer (not shown) For example, data 5, 6, 7).

Accordingly, when the data segment buffer (not shown) is in the full state, the operation method of the memory control device 100 according to the present invention is such that the hash value is stored for each data recorded in the data segment buffer And a new data segment B including two or more pieces of data sequentially recorded in a data segment buffer (not shown), that is, data 1, 2 ', 5, 6, 7 and a hash value H_B calculated for each piece of data can do.

The operation method of the memory control device 100 according to the present invention then distributes data segment B across memory chips # 0, # 1, # 2, and # 3, as shown in FIG.

At this time, the data 1 will be recorded on the same memory chip # 0 recorded before the garbage collection, and the data 2 'will be recorded on the same memory chip # 1 on which the data 2 has been recorded before the garbage collection.

As described above, the operation method of the memory control device according to the present invention is based on the storage system based on the data scrubbing performed to improve garbage collection and data reliability, which are essentially performed in a flash-based device such as an SSD, Not only improves the overall performance of the system but also minimizes the IO overhead due to data scrubbing.

The method of operating the memory controller according to an exemplary embodiment of the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer readable medium. The computer-readable medium may include program instructions, data files, data structures, and the like, alone or in combination. The program instructions recorded on the medium may be those specially designed and configured for the present invention or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

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 invention as defined by the appended claims.

According to the memory control device and the operation method of the memory control device according to the present invention, data scrubbing performed for garbage collection and data reliability enhancement, which are essentially performed in a flash-based device such as an SSD, It is possible to minimize the IO overhead due to data scrubbing as well as to improve the performance of the device. This is an invention that is industrially applicable because it is practically possible to carry out clearly.

10: Memory device
100: memory control device
110: Data read unit 120: Defect determination unit
130:
200: File system / application

Claims (11)

A data read unit for reading a specific data segment recorded in the memory device when a garbage collection performing event for the memory device occurs;
A defect determination unit for determining whether a data defect is present in the specific data segment during the reading of the specific data segment; And
When the specific data determined to be a data defect among the valid data is confirmed by reflecting the result of determining whether the data is defective in performing garbage collection in which valid data for the specific data segment is recorded in another memory location in the memory device, And a memory control unit for recovering and recording the specific data. The method according to claim 1,
The memory device includes a plurality of flash-based memory chips,
Wherein the memory chip is an SSD (Solid State Drive). The method according to claim 1,
Wherein the specific data segment comprises two or more pieces of data distributedly recorded in a plurality of memory chips in the memory device and a hash value calculated in the distributed recording for each of the two or more pieces of data. The method of claim 3,
The defect-
The hash value for each of the two or more data being sequentially read by the data read unit is calculated and whether or not the calculated hash value is different from the hash value in the specific data segment is determined for each of the two or more data, And judges that the data is defective. The method according to claim 1,
Wherein the specific data segment includes two or more data distributedly recorded in a plurality of memory chips in the memory device,
The memory control unit,
If valid data is selected from among the two or more data and specific data determined as a data defect among valid data is identified by reflecting a result of determining whether the data is defective, the specific data is restored, and a new And a data segment is formed and recorded in the memory device. 6. The method of claim 5,
The memory device records the same original or duplicate data segment as the specific data segment,
The memory control unit,
And replaces the specific data with corresponding data corresponding to the specific data of the original or replicated data segment to recover the specific data. 6. The method of claim 5,
Wherein the new data segment comprises:
Wherein the valid data is recorded in a memory chip that has been written before performing the garbage collection among a plurality of memory chips in the memory device. The method according to claim 1,
The defect-
Wherein the memory device reads from the memory device a data segment for which the garbage collection is not performed so that a predetermined time elapses, and judges only whether data is defective or not. A data read step of reading a specific data segment recorded in the memory device when a garbage collection execution event for the memory device occurs;
Determining whether a data defect is present in the specific data segment during the reading of the specific data segment; And
When the specific data determined to be a data defect among the valid data is confirmed by reflecting the result of determining whether the data is defective in performing garbage collection in which valid data for the specific data segment is recorded in another memory location in the memory device, And a garbage collection control step of recovering and recording the specific data. 10. The method of claim 9,
Wherein the specific data segment includes two or more data distributedly recorded in a plurality of memory chips in the memory device,
The garbage collection control step includes:
If valid data is selected from among the two or more data and specific data determined as a data defect among valid data is identified by reflecting a result of determining whether the data is defective, the specific data is restored, and a new And a data segment is formed and recorded in the memory device. 11. The method of claim 10,
The memory device records the same original or duplicate data segment as the specific data segment,
The garbage collection control step includes:
And replaces the specific data with corresponding data corresponding to the specific data of the original or replicated data segment to recover the specific data.

Images