KR101335343B1 - Apparatus and method for memory management - Google Patents

Abstract

In memory management, data read and write processing is performed by applying a stripe method to a memory including a plurality of disks. Determining whether to perform read processing according to RAID (Redundant Array of Inexpensive Disks) method, and if the target disk is inaccessible, the RAID method uses stripe data and parity data obtained from the target disk and other striped disks. Therefore, the request data is calculated and read processing is performed.

Description

Memory management apparatus and method {APPARATUS AND METHOD FOR MEMORY MANAGEMENT}

The present invention relates to a memory management apparatus and method using a RAID (Redundant Array of Inexpensive Disks) scheme.

Flash memory is a storage device that is rapidly increasing in use in recent years because of the advantages of high speed and stability, light weight and low power consumption.

In particular, workstations that require large amounts of electricity can save a lot of power while using flash memory, compared with the very high power consumption of conventional hard disks.

However, general flash memory has different units of read / write operation (page unit) and delete operation (block unit), and EBW (erase-before), when data is to be stored, must be deleted beforehand. -write) property.

Various flash translation layers (FTLs) or flash file systems (FFSs) have been developed to prevent performance degradation of write operations due to the characteristics of flash memory.

In this technique, when new data is updated in flash memory, instead of deleting the block to which the existing page belongs, new data to be updated after marking the invalidate information that the data of the page is no longer valid is deleted. Record on another page. As such, an invalidate page is then collected by a garbage collection operation.

At this time, a valid page is copied to a new block and a block erase operation is performed for garbage collection operations. If a page access request is received while such a garbage collection operation is being performed, the performance of flash memory is severely degraded. In particular, when a read request occurs during a garbage collection operation, the response time of the read request is slowed because the read request must be performed after the garbage collection operation is completed.

In this regard, Korean Patent Laid-Open Publication No. 10-2008-0085574 (name of the invention: an apparatus and method for garbage collection of a nonvolatile memory) discloses a conventional technology for garbage collection of a memory. In particular, the "Technology to which the invention pertains and the prior art in the field" of the published patent publication describes the definition of garbage collection related to the technical field of the present invention and problems occurring when garbage collection is performed.

In order to solve the above-described problems of the prior art, it is an object of the present invention to provide a memory management apparatus and method capable of efficiently reading and processing data in a memory.

Memory management apparatus according to an aspect of the present invention for achieving the above technical problem, the data read and write process by applying a stripe (stripe) method to a memory including a plurality of disks, the plurality of disks A data input / output control unit configured to determine whether to perform read processing according to a RAID (Redundant Array of Inexpensive Disks) scheme according to whether the read-requested data is accessible to the destination disk; And a request data operation processor configured to calculate the requested data according to the RAID method using stripe data and parity data obtained from other disks striped with the target disk when access to the target disk is impossible. It includes.

According to another aspect of the present invention, there is provided a memory management method comprising: receiving a read request for data that has been processed by applying a stripe method to a memory including a plurality of disks; Determining whether the read request data is accessible to the destination disk in which the read request data is stored; If the state of the destination disk is inaccessible, obtaining stripe data and parity data from the other disks striped with the destination disk; And calculating the requested data according to the RAID method using the stripe data and the parity data.

According to any one of the above-described problem solving means of the present invention, the disk in which the requested data is stored is inaccessible during the read processing of data stored in a stripe manner on a memory including a plurality of disks Even if it is (eg, garbage collection processing state), there is an effect that a read process can be performed quickly by calculating the requested data using other stripe data and parity bits.

In addition, in performing garbage collection of a plurality of disks in which data is stored in a stripe method, memory processing performance may be efficiently improved by determining the garbage collection processing order for each disk based on the size of the available storage area of each disk. .

1 is a block diagram illustrating a configuration of a memory management apparatus according to an embodiment of the present invention.
2 is a diagram for describing a memory striping method according to an embodiment of the present invention.
3 is a view for explaining a data operation processing method applying a RAID method according to an embodiment of the present invention.
4 is a flowchart illustrating a memory management method according to an embodiment of 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 is a block diagram illustrating a configuration of a memory management apparatus according to an embodiment of the present invention.

2 is a diagram for describing a memory striping method according to an embodiment of the present invention, and FIG. 3 is a diagram for describing a data operation processing method using a RAID method according to an embodiment of the present invention.

First, as shown in FIG. 1, the memory management apparatus 100 according to an exemplary embodiment of the present invention may include a data input / output controller 110, a memory 120, a garbage collection processor 130, and a request data operation processor 140. ).

The data input / output controller 110 applies a stripe method to a memory 120 including a plurality of disks to perform data read and write processes.

Specifically, as shown in FIG. 2, the data input / output control unit 110 converts input data (indicated as 'D0' to 'D19' in FIG. 2) into a plurality of disks (Disk0, Disk1, Disk2, and Disk3 in FIG. 2). It stores sequentially divided on the memory 120 including. The data input / output control unit 110 distributes and records parity data (indicated as 'P0' through 'P4' in FIG. 2) on each disk for stability of memory management.

As described above, a method of dividing and storing data on a plurality of disks is called a RAID (Redundant Array of Inexpensive Disks) method, and data sharing one parity bit is called a stripe.

For example, as shown in FIG. 2, data D0, D1, D2, D3, and parity data P0 form one stripe on the memory 120 having five disks, and P0 is D0 to D3. Stores parity information for.

For reference, in the RAID method, parity data is redundant data stored for each stripe for stability of input data. As such, the parity data may be calculated by an XOR (eXclusive OR) operation of data stored on each disk for each stripe (hereinafter referred to as 'stripe data').

In this case, when the data input / output controller 110 receives a read request for data stored on the memory 120, the data input / output controller 110 may access a disk (hereinafter, referred to as an “object disk”) in which the read request data is stored among the plurality of disks. Check the status.

When the target disk is in an accessible state, the data input / output controller 110 reads and provides data stored on the corresponding block on the target disk.

On the other hand, when the destination disk is in an inaccessible state, the data input / output controller 110 determines read processing according to a RAID method using the destination disk and other disks striped.

As such, when it is determined that the RAID type read processing is performed on the target disk, the data input / output controller 110 requests the request data operation processing unit 140 for arithmetic processing on the data for which the read processing is requested.

Meanwhile, the data input / output controller 110 according to an embodiment of the present invention may determine the accessible state according to the garbage collection processing state for each of the plurality of disks in the memory 120. For reference, garbage collection is a procedure of deleting data in a storage area (eg, a block area storage area) which is not available for each disk and converting the data into a usable storage area.

In this case, the data input / output controller 110 receives garbage collection processing information for each disk from the garbage collection processing unit 130 and determines whether the corresponding disk is accessible based on the received garbage collection processing information. For reference, the garbage collection processing information includes one of a garbage collection permission request and a garbage collection processing state for a specific disk.

In addition, when the data input / output controller 110 receives a garbage collection permission request for a specific disk from the garbage collection processing unit 130, the data input / output controller 110 permits garbage collection processing for the corresponding disk according to a preset procedure.

Specifically, when the data input / output controller 110 receives a garbage collection permission request for a plurality of disks, the data input / output controller 110 may permit the garbage collection process sequentially in the order in which the permission request is received. In this case, the data input / output unit 110 may permit the garbage collection process for the next disk after the garbage collection process for one disk is completed.

In addition, the data input / output unit 110 displays a garbage collection processing state (meaning that garbage collection is currently being performed on the corresponding disk) for a plurality of disks that are permitted by the garbage collection processing unit 130 by a specific condition. Receives and manages the access status for each disk.

The garbage collection processing unit 130 performs garbage collection processing on each of the plurality of disks, and transmits garbage collection processing information according to the state of a usable block for each disk to the data input / output control unit 110.

In this case, the garbage collection processing unit 130 may generate garbage collection processing information according to a plurality of steps based on the size of available blocks (hereinafter, referred to as 'usable blocks') remaining for each disk.

In detail, the garbage collection processor 130 requests the garbage collection permission from the data input / output controller 110 when the size of the usable block of the disk falls within a preset first reference size and second reference size range. The garbage collection processing unit 130 performs garbage collection processing permitted from the data input / output control unit 110 in response to the permission request. For reference, the first and second reference sizes are usable block sizes without difficulty in system data processing performance and memory management.

In addition, if the size of the usable block of the disk is less than the second reference size, the garbage collection processing unit 130 may perform the garbage collection processing for the corresponding disk without performing a garbage collection permission request to the data input / output controller 110. have. In this case, the garbage collection processing unit 130 may perform garbage collection processing according to its own permission on at least one disk. The garbage collection processing unit 130 transmits a garbage collection processing state for the disk on which the garbage collection processing is performed based on its own permission to the data input / output control unit 110. For reference, less than the second reference size is a usable block size that may adversely affect system data processing performance and memory management.

When the request data operation processing unit 140 receives an operation processing request for data (hereinafter, referred to as 'request data') for which read processing is requested from the data input / output control unit 110, the request data operation processing unit 140 may calculate a request data by applying a RAID method. Perform arithmetic processing.

In this case, the request data operation processor 140 obtains the stripe data and the parity data from the target disk on which the request data is stored and other striped disks. The request data operation processor 140 calculates the request data by performing an XOR (eXclusive OR) operation on the obtained stripe data and parity data.

Specifically, as shown in (a) of FIG. 3, when only the target disks among the disks striped with each other are in an inaccessible state (that is, garbage collection processing state), the request data operation processing unit 140 performs the rest except the target disk. The data and parity data of the striping disks are processed by XOR (eXclusive OR) operation.

FIG. 3A illustrates a case in which arithmetic processing for D2 data stored in Disk 2 is requested when Disk 2 of Stripe 0 is in a garbage collection (GC) processing state. At this time, the request data operation processing unit 130 calculates D2 data by performing XOR operation as shown in Equation 1 below.

&Quot; (1) "

In this way, the calculated D2 data is transmitted to the host (that is, the subject that requested to read the D2 data from an external device or module on the system) through the data input / output controller 110.

On the other hand, as shown in (b) of FIG. 3, the request data operation processing unit 140, if two or more disks including the target disk among the disks striped together is in an inaccessible state (that is, garbage collection processing state), The request data is calculated after the state of the disk or the state of all the striped disks except the destination disk becomes accessible.

 FIG. 3B illustrates a case in which operation processing for D2 data stored in Disk 2 is requested when Disk 0, Disk 2, and Disk 3 of Stripe 0 are in a garbage collection (GC) processing state.

At this time, the request data operation processing unit 130 calculates the request data using data stored in the disk which becomes the first accessible state (that is, the state in which garbage collection processing is completed) among the target disk or all remaining striping disks.

Specifically, when the garbage collection (GC) processing of the target disk (Disk 2) is completed first, the request data operation processing unit 130 extracts the data stored in the target disk (Disk 2) as request data and outputs the data input / output controller 110. )

In addition, the request data operation processing unit 130, if the garbage collection processing of the striping disk (Disk 0 and Disk 3) other than the target disk (Disk 2) is completed before the garbage collection processing of the target disk (Disk 2), the remaining striping disk Data and parity data are obtained and processed as in Equation 1 above. The request data operation processing unit 130 provides the data input / output control unit 110 with the request data calculated through the operation processing.

Hereinafter, a memory management method according to an exemplary embodiment of the present invention will be described in detail with reference to FIG. 4.

4 is a flowchart illustrating a memory management method according to an embodiment of the present invention.

As shown in FIG. 4, when a read request for data written to a memory is generated according to a striping method (S410), it is determined whether the state of the target disk on which the requested data is stored is accessible (see FIG. 4). S420).

At this time, in an embodiment of the present invention, the accessible state of the destination disk may be determined according to whether it is garbage collection processing.

That is, the garbage collection processing state of the target disk is checked to determine that the target disk is inaccessible when the garbage collection processing is performed.

Then, if the target disk is in an inaccessible state, it is determined whether the target disk and other striped disks striped are accessible (S430).

Then, when all of the other stripe disks are in an accessible state (that is, not performing garbage collection processing), stripe data and parity bits are obtained from the other stripe disks (S440).

On the other hand, when at least one of the other stripe disks is in an inaccessible state, the process returns to step S420 to repeatedly determine the state of each disk until either one of the target disk or all other stripe disks becomes accessible. do.

At this time, when two or more disks including the target disk among the striped disks are in an inaccessible state, data stored in the first disk or in all other striped disks that are first accessible are acquired.

Next, the read data is calculated using the obtained stripe data and the parity bit, and the requested data is calculated (S450).

In this case, the operation processing is an XOR operation between the obtained stripe data and the parity bit.

Thereafter, the calculated request data is transmitted to a data read request subject (ie, a host) (S460).

On the other hand, if it is determined that the access to the target disk is possible as a result of the determination in step S420, the target disk is accessed to perform read processing (S470), and the read-processed request data is subject to the data read request. (Ie, host) (S460).

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.

110: data input and output control unit 120: memory
130: garbage collection processing unit 140: request data operation processing unit

Claims (11)

In the memory management apparatus,
A data read and write process is performed by applying a stripe method to a memory including a plurality of disks, and among the plurality of disks, a RAID is allocated depending on whether the read-requested data of the plurality of disks is accessible to the target disk. A data input / output controller for determining whether to perform read processing according to a RAID (Redundant Array of Inexpensive Disks) scheme; And
A request data operation processor configured to calculate the requested data according to the RAID method using stripe data and parity data obtained from other disks striped with the target disk when access to the target disk is impossible Memory management device comprising. The method of claim 1,
Garbage collection processing for each of the plurality of disks, and a garbage collection for transmitting one of a garbage collection permission request and a garbage collection processing state according to the state of the usable block for each disk to the data input / output controller. Including a processing unit,
The accessibility of the destination disk is determined according to the garbage collection processing state. 3. The method of claim 2,
The garbage collection processing unit,
The garbage collection permission is requested for a disk belonging to a range in which the usable block size is less than or equal to the first reference size and less than the first reference size, and the permission according to the request is received to perform the garbage collection process. and,
Perform garbage collection processing on the disk having the usable block size less than the second reference size, and provide a garbage collection processing status for at least one disk on which the garbage collection processing is performed,
And no garbage collection processing is performed on the disk having the usable block size exceeding the first reference size. 3. The method according to claim 1 or 2,
The request data operation processing unit,
When two or more disks including the target disk among the striped disks are in an inaccessible state,
And calculating the requested data after the state of the target disk or the states of all the striped disks except the target disk become accessible. 5. The method of claim 4,
The request data operation processing unit,
And calculating the requested data by using the data stored in the disk which becomes the first accessible state of the destination disk or all the remaining striping disks. The method of claim 1,
The request data operation processing unit,
And calculating the requested data through an XOR (eXclusive OR) operation on the stripe data and the parity data. A memory management method comprising:
(a) receiving a read request for data written by applying a stripe method to a memory including a plurality of disks;
determining whether the read request data is accessible to the target disk in which the read request data is stored;
(c) when the state of the target disk is inaccessible, obtaining stripe data and parity data from other disks striped with the target disk; And
(d) calculating the requested data according to the RAID method using the stripe data and the parity data. The method of claim 7, wherein
In the step (b)
Checking a garbage collection processing state for the target disk, and determining that the target disk is in an inaccessible state when the target disk is in the garbage collection processing. The method of claim 8,
Before the step (b)
Requesting a garbage collection permission for a disk belonging to a range in which a usable block size is less than a first reference size or less than a first reference size for each of the plurality of disks, and requesting garbage collection permission and receiving the permission according to the request; Performing a process;
Providing garbage collection processing status for at least one disk for which the garbage collection processing is performed, the garbage collection processing being performed on the disk having the usable block size less than the second reference size for each of the plurality of disks; And
And not performing garbage collection processing on the disk having the usable block size exceeding the first reference size for each of the plurality of disks. 9. The method according to claim 7 or 8,
In the step (c)
Determining whether an accessible state of stripe disks composed of the destination disk and the other disks is accessible; And
When two or more disks including the target disk among the stripe disks are in an inaccessible state, acquiring data stored in the disk which becomes the first accessible state among all of the target disks or the other disks; Way. The method of claim 7, wherein
In the step (d)
And calculating the requested data through an eXclusive OR (XOR) operation on the stripe data and the parity data.

Images