CN103019966B - A data storage system including backup body memory and its management method - Google Patents

Abstract

The invention relates to a data storage system and a management method thereof. The data storage system comprises N storage devices, a backup memory and a controller, wherein N is a natural number. Each storage device includes a respective write cache. The backup memory retains the data stored therein once the data storage system is subjected to a power interruption. The controller is configured to perform a predetermined operation on a plurality of data transmitted from an access request unit to generate a plurality of write data, transmit the write data to the write caches of the N storage devices, copy the write data into the backup memory, and mark the write data copied into the backup memory as "valid" in response to the N storage devices transmitting a write acknowledge message.

Description

A data storage system including backup body memory and its management method

technical field

The present invention relates to a data storage system (data storage system) and its management method (managing method), in particular to a data storage system conforming to a disk array (redundant array of independent drives, RAID) architecture. And especially, the data storage system of the present invention includes backup memory, so as to prevent the data to be written from disappearing when the data storage system suffers from power interruption, and is protected.

Background technique

As users store more and more data, data storage systems conforming to the RAID architecture (also known as RAID systems) have been widely used to store large amounts of data. Disk array (RAID) system can provide high availability (high availability), high performance (high performance) or large capacity (high volume) data storage space to the host (host).

The different technologies utilized by disk array systems are called RAID levels, which are also distinguished: RAID 0, RAID 1, RAID2, RAID 3, RAID 4, RAID 5, and RAID 6. Each RAID level technology has its own advantages and disadvantages.

The composition of the existing disk array system includes a disk array controller (RAID controller) and a disk array composed of a plurality of disk drives (disk drives). A disk array controller is connected to each disk drive, and the disk array controller defines the disk array as one or more of RAID 0, RAID 1, RAID 2, RAID 3, RAID 4, RAID 5, or RAID 6 Logical disk drive (logical disk drive) to achieve.

Please refer to FIG. 1 , the structure of an existing data storage system 1 conforming to a RAID structure is shown in FIG. 1 . The data storage system 1 includes a plurality of storage devices 12 and a controller 10 .

In a specific embodiment, the plurality of storage devices 12 are standardized into a disk array architecture, such as RAID 0, RAID 1, RAID 2, RAID 3, RAID 4, RAID 5, or RAID 6.

The controller 10 is capable of generating (reconstructing) redundant data identical to the data to be read. For example, under the RAID5 architecture, the controller 10 generates redundant data through an exclusive OR (XOR) operation.

In practical applications, each storage device 12 can be a tape drive (tape drive), a disk drive (disk drive), a memory device (memory drive), an optical storage recording device (optical storage drive), or in A sector corresponding to a single read-write head in the same disk drive, or other equivalent storage devices.

Also shown in FIG. 1 , the controller 10 is connected to each storage device 12 . Fig. 1 also shows an access request unit (application I/O request unit) 2. The access request unit 2 is connected to the controller 10 via a transmission interface 14 . In practical applications, the access request unit 2 can be a network computer, a minicomputer, a mainframe, a notebook computer, or any electronic device that needs to read data in the data storage system 1, such as a mobile phone, a personal digital assistant (personaldigital assistant, PDA), digital video equipment, digital music player, etc.

When the access request unit 2 is an independent electronic device, it can pass through a storage area network (SAN), a local area network (LAN), a serial advanced technology (serial ATA, SATA) interface, a fiber channel (FC), Transmission interfaces such as Small Computer Standard Interface (SCSI), or input/output (I/O) interfaces such as PCI Express are connected to the data storage system 1 . In addition, when the access request unit 2 is a special integrated circuit component, or other equivalent devices capable of sending I/O read requests, it can send data read requests to the device according to commands (or requests) from other devices. The controller 10 reads the data in the storage devices 12 through the controller 10 .

The controller 10 and the storage devices 12 in the data storage system 1 can be placed in a single box at the same time, or can be separated independently and placed in different boxes. In practice, the controller 10 can communicate with the controller through transmission interfaces such as Fiber Channel (FC), Small Computer System Interface (SCSI), Serial Attached SCSI (SAS), Serial ATA (SATA), Parallel ATA (PATA, or IDE). These storage devices 12 are connected. If these storage devices 12 are disk drives, each storage device 12 can be a disk drive with various interface types such as FC, SAS, SCSI, SATA, and PATA. The controller 10 can be a disk array controller, or a controller with a data storage system capable of generating redundant data.

Also shown in FIG. 1 , each storage device 12 includes a respective write cache (write cache) 122 . Each storage device 12 is divided into multiple data blocks. From the point of view of fault tolerance, it can be divided into two types: user data segment and parity data segment. The user data section stores general user data. The parity data segment stores a redundant set of parity data for back-calculation of user data when there is a need for fault tolerance. Corresponding user data segments and co-located data segments existing in different storage devices form a storage stripe (stripe), wherein the data in the co-located data segment is mutually exclusive or ( XOR) operation results. It should be declared that those skilled in the art understand that the calculation of the data in the parity data segment, in addition to the exclusive OR (XOR) operation, also includes various parity (parity) operations or similar As long as there is the following relationship: the data in any data section in the same storage partition can be calculated from the data in its corresponding data section.

The user end wants to write data into the data storage system 1 , and the controller 10 receives multiple pieces of data transmitted by the access request unit 2 via the transmission interface 14 . Then, the controller 10 performs a predetermined operation (for example, exclusive OR (XOR) operation) on the received data to generate a plurality of writing data. Then, the controller 10 transmits the write data to the write caches 122 of the plurality of storage devices 12 . After the write data is stored in the write caches 122 of the storage devices 12 , the storage devices 12 reply to the controller 10 with a write confirmation message. At this time, the controller 10 assumes that the write data has been written into the plurality of storage devices 12 and does not protect the data. However, not all of the write data stored in the write caches 122 of the plurality of storage devices 12 have been written into the recording media of the plurality of storage devices 12 themselves. Once the data storage system 1 suffers from an unexpected power interruption, the written data still stored in the write caches 122 of the plurality of storage devices 12 will disappear.

Contents of the invention

Therefore, the present invention provides a data storage system and a management method thereof. And especially, the data storage system of the present invention includes backup memory, so as to prevent the data to be written from disappearing when the data storage system suffers from power interruption, and is protected.

A data storage system according to a preferred embodiment of the present invention includes N storage devices, backup memory and a controller, wherein N is a natural number. Each storage device includes a respective write cache. Once the data storage system suffers a power outage, the backup memory still preserves the data stored therein. The controller is used to receive multiple pieces of data transmitted by an access request unit, perform a predetermined operation on the received data to generate multiple pieces of write data, and send the write data to the write caches of the N storage devices , copy the written data into the backup memory, and mark the written data copied into the backup memory as valid in response to the N storage devices sending a write confirmation message. And, when the controller judges that the storage space of the backup memory is lower than a threshold value or copying these written data to the addresses in the backup memory is duplicated, the controller issues a clear cache command (flush cache command) to the N storage devices, and in response to the N storage devices, send a clear cache completion message to clear the written data stored in the backup memory.

In a specific embodiment, when the data storage system is turned on, the controller judges whether there is written-in data marked valid in the backup memory. If the judging result is affirmative, the controller writes the valid write data in the backup memory into the N storage devices. And, after finishing writing the write-in data marked valid into the N storage devices, the controller clears the write-in data stored in the backup memory.

In a specific embodiment, the predetermined operation may be an exclusive OR (XOR) operation, a mirroring operation (mirroring operation), a no operation (NO operation), a slice operation (slice operation), a finite Galois field (Galois field, GF) essential polynomial operation, a finite field mathematical maximum distance separable code (maximum distance separable code) operation, other parity (parity) operation or similar operation techniques.

In a specific embodiment, the backup memory is a non-volatile memory or a volatile memory configured with an independent power supply.

In one embodiment, the backup memory is only electrically connected to the controller. In another embodiment, the backup memory is a write cache in the controller. In another specific embodiment, the backup memory is electrically connected between the controller and the N storage devices.

The management method according to a preferred embodiment of the present invention is directed to the management method of a data storage system. The data storage system includes multiple storage devices and a backup memory. Each storage device includes a respective write cache. When the data storage system suffers from power interruption, the backup memory still saves the data stored therein. According to the management method of the preferred embodiment of the present invention, firstly, multiple pieces of data transmitted by an access request unit are received. Then, according to the management method of the present invention, a predetermined operation is performed on the received data to generate multiple writing data. Then, according to the management method of the present invention, it is judged whether the storage space of the backup memory is lower than a threshold value, or whether the addresses for copying the written data to the backup memory are duplicated. If the judgment result is negative, the management method according to the present invention transmits the write data to the write caches of the N storage devices, and copies the write data to the backup memory. Moreover, the management method according to the present invention marks the write data copied into the backup memory as valid in response to the N storage devices sending a write confirmation message.

If the storage space of the backup memory is lower than a threshold value or the addresses for copying the written data to the backup memory are duplicated, the management method of the present invention issues a clear cache command to the N storage devices. Moreover, the management method according to the present invention clears the write-in data stored in the backup memory in response to the N storage devices sending a cache clearing completion message.

When the data storage system is turned on, it is judged according to the management method of the present invention whether there is written-in data marked valid in the backup memory. If the judgment result is affirmative, according to the management method of the present invention, write the valid write-in data in the backup memory into the N storage devices, and write the valid write-in data into the N storage devices after completion. After the device is installed, clear the written data stored in the backup memory.

The advantages and spirit of the present invention can be further understood through the following embodiments and accompanying drawings.

Description of drawings

The present invention will be further described below in conjunction with accompanying drawing:

FIG. 1 is a schematic diagram of an existing data storage system conforming to the RAID architecture.

FIG. 2 is a schematic structural diagram of a data storage system according to a preferred embodiment of the present invention.

FIG. 3 is a schematic diagram of a modification of a backup memory of the data storage system according to a preferred embodiment of the present invention.

FIG. 4 is a schematic diagram of another modification of the backup memory of the data storage system according to a preferred embodiment of the present invention.

Fig. 5 is a flowchart of a management method according to a preferred embodiment of the present invention.

Fig. 6 is another flow chart of the management method according to the preferred embodiment of the present invention.

Description of component symbols:

[Description of main component symbols]

1, 3: Data storage system 10, 30: Controller

12, 32: storage device 122, 322: write cache

14, 34: Transmission interface 36: Backup memory

2, 4: Access request unit 5: Management method

S50～S70: process steps

Detailed ways

The invention provides a data storage system and a management method thereof. And especially, according to the present invention, the data storage system includes a backup memory, so as to prevent the data to be written from disappearing and be protected when the data storage system suffers from power interruption. The following is a detailed description of the preferred specific embodiments of the present invention, so as to fully explain the characteristics, spirit, advantages and practical feasibility of the present invention.

Please refer to FIG. 2 , the architecture of the data storage system 3 according to a preferred embodiment of the present invention is shown in FIG. 2 . The data storage system 3 includes N storage devices 32 , a controller 30 and a backup memory 36 , wherein N is a natural number (1, 2, 3, . . . ).

In a specific embodiment, the N storage devices 32 are standardized into a disk array structure, for example, RAID 0, RAID 1, RAID 2, RAID 3, RAID 4, RAID 5, or RAID 6.

The controller 30 is capable of generating (reconstructing) redundant data identical to the data to be read. For example, under the RAID 5 architecture, the controller 30 generates redundant data through an exclusive OR (XOR) operation.

In a specific embodiment, each storage device 32 can be a tape drive, a disk drive, a memory device, an optical storage recording device, or a sector corresponding to a single read-write head in the same disk drive , or other equivalent storage devices.

Also shown in FIG. 2 , the controller 30 is connected to each storage device 32 . FIG. 2 also shows an access request unit 4 . The access request unit 4 is connected to the controller 30 via a transmission interface 34 . In practical applications, the access request unit 4 can be a network computer, a minicomputer, a mainframe, a notebook computer, or any electronic device that needs to read data in the data storage system 3, such as a mobile phone, a personal digital assistant , digital video equipment, digital music players, etc.

When the access request unit 4 is an independent electronic device, it can pass through a storage area network (SAN), a local area network (LAN), a serial advanced technology (serial ATA, SATA) interface, a fiber channel (FC) , Small Computer Standard Interface (SCSI) and other transmission interfaces, or PCI Express and other input/output (I/O) interfaces are connected to the data storage system 3 . In addition, when the access request unit 4 is a special integrated circuit component, or other equivalent devices capable of sending I/O read requests, it can send data read requests to the device according to commands (or requests) from other devices. The controller 30 reads the data in the storage devices 32 through the controller 30 .

The controller 30 and the storage devices 32 in the data storage system 3 can be placed in a single box at the same time, or can be separated independently and placed in different boxes. In practice, the controller 30 can communicate with the controller through transmission interfaces such as Fiber Channel (FC), Small Computer System Interface (SCSI), Serial Attached SCSI (SAS), Serial ATA (SATA), Parallel ATA (PATA, or IDE). These storage devices 32 are connected. If these storage devices 32 are disk drives, each storage device 32 may be a disk drive with various interface types such as FC, SAS, SCSI, SATA, and PATA. The controller 30 can be a disk array controller, or a controller with a data storage system capable of generating redundant data.

Also shown in FIG. 2 , each storage device 32 includes a respective write cache 322 . Each storage device 32 is divided into a plurality of data sectors. From the point of view of fault tolerance, it can be divided into two types: user data segment and parity data segment. The user data section stores general user data. The parity data segment stores a redundant set of parity data for back-calculation of user data when there is a need for fault tolerance. Corresponding user data segments and co-located data segments existing in different storage devices form a storage partition, wherein the data in the co-located data segment is executed with the data in the user data segment to perform an exclusive OR (XOR) operation The result obtained. It should be declared that those skilled in the art understand that the calculation of data in the same-bit data segment, in addition to using exclusive or (XOR) operation, for various RAID architectures, can also be a mirror operation or no operation. (NO operation), segmentation operation, finite Galois field (Galois field, GF) essential polynomial operation, finite field mathematical maximum distance separable code (maximum distance separable code) operation, other parity operation or similar operation technology, As long as the following relationship exists: the data in any data segment in the same storage partition can be calculated from the data in its corresponding data segment.

The user terminal wants to write data into the data storage system 3 , and the controller 30 receives multiple pieces of data transmitted by the access request unit 4 via the transmission interface 34 . Then, the controller 30 performs a predetermined operation (for example, exclusive OR (XOR) operation) on the received data to generate a plurality of writing data. Then, the controller 30 sends the write data to the write caches 322 of the N storage devices 32 . In particular, the controller 30 copies the written data into the backup memory 36 .

After the write data is stored in the write caches 322 of the N storage devices 32 , the N storage devices 32 reply to the controller 30 with a write confirmation message. At this time, the controller 30 marks the written data copied into the backup memory 36 as valid in response to the write confirmation message sent by the N storage devices 32 , and still protects the written data. However, not all of the write data stored in the write caches 322 of the N storage devices 32 have been written into the recording media of the N storage devices 32 themselves.

And, when the controller 30 judges that the storage space of the backup memory 36 is lower than a threshold value or copying these write data to the addresses in the backup memory 36 is duplicated, the controller 30 issues a clear cache command to the N A storage device 32. After the N storage devices 32 receive the clear cache command, they will write the write data in the write cache 322 into the recording media of the N storage devices 32 themselves. After all the writing data in the caches 322 are written into the recording media of the N storage devices 32 themselves, the N storage devices 32 reply to the controller 30 with a cache clearing completion message. At this time, the controller 30 responds to the clearing and cache completion message, and clears the written data stored in the backup memory 36 .

Especially, once the data storage system 3 suffers unexpected power interruption, the backup memory 36 still saves the data stored therein. When the data storage system 3 is turned on, the controller 30 judges whether there is valid writing data in the backup memory 36 . If the judgment result is affirmative, the controller 30 writes the valid write data in the backup memory 36 into the N storage devices 32 . Moreover, after writing the valid write-in data into the N storage devices 32 , the controller 30 clears the write-in data stored in the backup memory 36 . Thereby, the data storage system 3 according to the present invention can prevent the data to be written from disappearing when the power is interrupted, and ensure that the data to be written is written into the N storage devices 32 .

In a specific embodiment, the backup memory 36 is a non-volatile memory (such as Flash Memory), or a volatile memory configured with an independent power supply (such as a battery or an uninterruptible power supply system).

In a specific embodiment, as shown in FIG. 2 , the backup memory 36 is only electrically connected to the controller 30 .

In another embodiment, as shown in FIG. 3 , the backup memory 36 is a write cache in the controller 30 . Components marked with the same numbers as those in FIG. 2 in FIG. 3 have the same or similar functions and operations, and will not be repeated here.

In another specific embodiment, as shown in FIG. 4 , the backup memory 36 is electrically connected between the controller 30 and the N storage devices 32 . Components marked with the same numbers as those in FIG. 2 in FIG. 4 have the same or similar functions and operations, and will not be repeated here.

Please refer to FIG. 5 . FIG. 5 is a flowchart illustrating a management method 2 according to a preferred embodiment of the present invention. The management method 5 according to the present invention is a management method for the data storage system 3 shown in, for example, FIG. 2 , FIG. 3 or FIG. 4 . The structure of the data storage system 3 has been described in detail above, and will not be repeated here.

As shown in FIG. 5 , the management method 5 according to the present invention first executes step S50 to receive multiple pieces of data sent by an access request unit 4 .

Next, according to the management method 5 of the present invention, step S52 is performed to perform a predetermined operation on the received data to generate multiple write-in data.

In a specific embodiment, the predetermined operation may be an exclusive OR (XOR) operation, a mirror image operation, a none operation, a slicing operation, a finite Galois field (Galois field, GF) essential polynomial operation, a Mathematical maximum distance separable code operations for finite fields, other parity operations or similar operation techniques.

Then, according to the management method 5 of the present invention, step S54 is executed to determine whether the storage space of the backup memory 36 is lower than a threshold value.

If the judgment result of step S54 is negative, according to the management method 5 of the present invention, step S56 is executed to judge whether the addresses for copying the written data to the backup memory 36 are duplicated.

If the judgment result of step S56 is negative, according to the management method 5 of the present invention, step S58 is executed to send these write data to the write cache 322 of the N storage devices 32, and copy these write data to the backup memory within 36.

After step S58, the management method 5 according to the present invention executes step S60 to mark the written data copied into the backup memory 36 as valid in response to the N storage devices 32 sending a write confirmation message.

If the judging result of step S54 is affirmative, or the judging result of step S56 is affirmative, according to the management method 5 of the present invention, step S62 is executed to issue a clear cache command to the N storage devices 32 . At this time, after all the written data in the caches 322 are written into the recording media of the N storage devices 32 themselves, the N storage devices 32 reply a clearing and cache completion message.

After step S62, according to the management method 5 of the present invention, step S64 is executed to clear the write-in data stored in the backup memory 36 in response to the clearing cache completion message sent by the N storage devices 32 . After step S64, the procedure is completed.

Please refer to FIG. 6 , when the data storage system 3 is turned on, the management method 5 according to the present invention further executes step S66 to determine whether there is valid write-in data in the backup memory 36 .

If the judgment result of step S66 is affirmative, the management method 5 according to the present invention further executes step S68 to write the write-in data marked valid in the backup memory 36 into the N storage devices 32 . After step S68, that is, after writing the valid write-in data into the N storage devices, the management method 5 of the present invention executes step S70 to clear the write-in data stored in the backup memory 36 . Thereby, the management method 5 according to the present invention can prevent the data to be written from disappearing when the data storage system 3 suffers a power outage, and ensure that the data to be written is written into the N storage devices 32 .

Through the above detailed description of the preferred embodiments of the present invention, it is possible to clearly understand the data storage system and its management method provided by the present invention, which can prevent the data to be written from disappearing when the data storage system suffers from a power outage, and is Protect.

Through the above detailed description of the preferred embodiments, it is hoped that the features and spirit of the present invention can be described more clearly, and the scope of the present invention is not limited by the preferred embodiments disclosed above. On the contrary, the intention is to cover various changes and equivalent arrangements within the scope of the claimed patent scope of the present invention. Therefore, the scope of the claimed scope of the present invention should be interpreted in the broadest way based on the above description, so as to cover all possible changes and equivalent arrangements.

Claims (9)

1. a data storage system, becomes electrically to be coupled with an access requirement unit, comprises:

N number of storage device, optionally export a write and confirm message, each storage device comprises a respective write cache and a respective recording medium, and N is a natural number;

One archive memory, suffers power breakdown once this data storage system, and this archive memory still preserves storage data within it; And

One controller, in order to many data transmitted from this access requirement unit are performed a predetermined operation to produce many write data, transmit the write cache of these write data to this N number of storage device, copy these write data in this archive memory, and respond this write and confirm message, these write data be copied in this archive memory are denoted as " effectively "

Wherein when this archive memory storage area lower than a threshold value or copy these write data in this archive memory address repeat time, this controller issues the fast command fetch of a cleaning to this N number of storage device, make the write cache of this N number of storage device after those write data are write the recording medium of this N number of storage device completely, reply this controller one by this N number of storage device to clear up cache and complete message, this controller responds this cleaning cache and completes message and remove the write data be stored in this archive memory.

2. data storage system according to claim 1, it is characterized in that: when this data storage system is started shooting, if have sign in this archive memory " effectively " write data, this is then effectively write data and writes back this N number of storage device by this controller, and in the completed, remove the write data be stored in this archive memory.

3. data storage system according to claim 2, is characterized in that: this predetermined operation is selected from the one in the group be made up of without the mathematics maximum distance separable codes computing of computing, all point of computing, a limited Patrick Gallois field essence multinomial operation and a finite field a mutual exclusion or (XOR) computing, a mirror image computing.

4. data storage system according to claim 2, it is characterized in that: this archive memory is the volatile memory of a non-voltile memory or configuration one independent current source, this archive memory is only electrically connected to the write cache in this controller, this controller, or is connected electrically between this controller and this N number of storage device.

5. the management method for a data storage system, this data storage system becomes electrically to be coupled with an access requirement unit, and comprise N number of storage device and an archive memory, N is a natural number, each storage device comprises a respective write cache and a respective recording medium, suffer power breakdown once this data storage system, this archive memory still preserves storage data within it, and this management method comprises the following step:

A () receives many data that this access requirement unit transmits;

B data that these receive by () perform a predetermined operation to produce many write data;

C () judges that whether the storage area of this archive memory is lower than a threshold value, or judge whether copy these write data repeats to the address in this archive memory;

If step (c) judged result is yea-sayer, perform the following step:

The fast command fetch of a cleaning is issued to this N number of storage device by a controller;

The write cache of this N number of storage device, after these write data are write the recording medium of this N number of storage device completely, is replied this controller one by this N number of storage device and is cleared up cache and complete message; And

This cleaning cache being responded this N number of storage device transmission by this controller completes message, removes the write data be stored in this archive memory;

If step (c) judged result is naysayer, perform the following step:

Transmit the write cache of these write data to this N number of storage device, and copy these write data in this archive memory; And

Respond the write confirmation message that this N number of storage device transmits, these write data that will be copied in this archive memory are denoted as " effectively ".

6. management method according to claim 5, is characterized in that: when this data storage system is started shooting, and performs the following step:

Judge whether have sign in this archive memory " effectively " write data; And

If judged result is yea-sayer, writes this N number of storage device by indicating effective write data in this archive memory, and in the completed, remove the write data be stored in this archive memory.

7. management method according to claim 6, is characterized in that: this predetermined operation is selected from the one in the group be made up of without the mathematics maximum distance separable codes computing of computing, all point of computing, a limited Patrick Gallois field essence multinomial operation and a finite field a mutual exclusion or (XOR) computing, a mirror image computing.

8. management method according to claim 7, is characterized in that: this archive memory is the volatile memory of a non-voltile memory or configuration one independent current source.

9. a data storage system, becomes electrically to be coupled with an access requirement unit, comprises:

N number of storage device, optionally export a write and confirm message, each storage device comprises a respective write cache and a respective recording medium, and N is a natural number;

One archive memory, suffers power breakdown once this data storage system, and this archive memory still preserves storage data within it; And

One controller, in order to many data transmitted from this access requirement unit are performed a predetermined operation to produce many write data, copy these write data in this archive memory, and respond this write and confirm message, these write data be copied in this archive memory are denoted as " effectively "

Wherein when this archive memory storage area lower than a threshold value or copy these write data in this archive memory address repeat time, this controller issues the fast command fetch of a cleaning to this N number of storage device, make the write cache of this N number of storage device after those write data are write the recording medium of this N number of storage device completely, reply this controller one by this N number of storage device to clear up cache and complete message, this controller responds this cleaning cache and completes message and remove the write data be stored in this archive memory, when this data storage system is started shooting, if have sign in this archive memory " effectively " write data, this is then effectively write data and writes back this N number of storage device by this controller, and in the completed, removing is stored in the write data in this archive memory.