WO2019127212A1 - 一种数据写入的方法及固态硬盘阵列 - Google Patents

一种数据写入的方法及固态硬盘阵列 Download PDF

Info

Publication number
WO2019127212A1
WO2019127212A1 PCT/CN2017/119390 CN2017119390W WO2019127212A1 WO 2019127212 A1 WO2019127212 A1 WO 2019127212A1 CN 2017119390 W CN2017119390 W CN 2017119390W WO 2019127212 A1 WO2019127212 A1 WO 2019127212A1
Authority
WO
WIPO (PCT)
Prior art keywords
write
controller
ftl
refresh
data block
Prior art date
Application number
PCT/CN2017/119390
Other languages
English (en)
French (fr)
Inventor
吴登奔
徐晓忻
王俊捷
Original Assignee
华为技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为技术有限公司 filed Critical 华为技术有限公司
Priority to PCT/CN2017/119390 priority Critical patent/WO2019127212A1/zh
Priority to EP17936484.9A priority patent/EP3726364B1/en
Priority to CN201780040471.7A priority patent/CN110462577B/zh
Publication of WO2019127212A1 publication Critical patent/WO2019127212A1/zh
Priority to US16/914,375 priority patent/US11243701B2/en

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
    • G06F3/0614—Improving the reliability of storage systems
    • G06F3/0619—Improving the reliability of storage systems in relation to data integrity, e.g. data losses, bit errors
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0629—Configuration or reconfiguration of storage systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0638—Organizing or formatting or addressing of data
    • G06F3/064—Management of blocks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0655—Vertical data movement, i.e. input-output transfer; data movement between one or more hosts and one or more storage devices
    • G06F3/0659—Command handling arrangements, e.g. command buffers, queues, command scheduling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601—Interfaces specially adapted for storage systems
    • G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
    • G06F3/0671—In-line storage system
    • G06F3/0683—Plurality of storage devices
    • G06F3/0688—Non-volatile semiconductor memory arrays

Definitions

  • the present application relates to the field of information technology, and in particular, to a data writing method and a solid state hard disk array.
  • Redundant Array of Independent Disks is a redundant array of independent disks. It combines multiple independent disks into different arrays to form a disk array. The effect of the data provided by the individual disks is used to enhance the entire disk. Array performance. RAID technology cuts data into segments, each of which is striped, stored on each disk of the disk array to provide data redundancy. With the parity check, when any disk in the disk array fails and a certain piece of data is lost, the system can still read the complete data, and when the data is reconstructed, the lost piece of data is calculated and then stored in the disk.
  • a disk array applying RAID technology needs to ensure the atomicity of a data write operation when writing data to be written divided into strips, that is, after a write operation of writing data to be written to the disk array, Writes to all disks in the disk array need to succeed or fail. If some of the disk write data is successful, and the other disk fails to write data, the data in the disk array will be inconsistent, resulting in the data redundancy provided by the RAID.
  • the embodiment of the invention provides a data writing method and a solid-state hard disk array, which can realize the atomicity of a write operation of each data block when n data blocks are written into n solid-state hard disks.
  • the present invention provides a data writing method, which is applied to a solid state disk array.
  • the method includes: a controller of the solid state disk array acquires data to be written, and the data to be written includes n data blocks. .
  • the controller generates n write commands, wherein the n write commands are in one-to-one correspondence with the n data blocks.
  • the controller sends the n write commands to the n solid state drives in the SSD array, wherein the n write commands are in one-to-one correspondence with the n SSDs, and each write command is used to request a corresponding solid state
  • the hard disk stores the data block corresponding to the write command.
  • the controller determines whether n write success responses corresponding to the n write commands are received.
  • the controller triggers a Flash Translation Layer (FTL) refresh of the n data blocks, where the FTL refresh of each data block includes: A mapping relationship between the logical address of each data block and the current physical address of the data block is established in the FTL list.
  • FTL Flash Translation Layer
  • Separating the data storage from the FTL list refresh ensures the atomicity of the write operation of the array controller or each solid state hard disk after the data storage to the idle physical storage space is interrupted. At the same time, the extra IO overhead caused by the successful write response in this method is relatively small, and will not have a great impact on the speed of data writing.
  • n write success responses corresponding to the n write commands are received, before the controller triggers the FTL refresh of the n data blocks,
  • the controller generates n write logs, and stores the n write logs in a non-volatile memory.
  • the n write logs are in one-to-one correspondence with the n data blocks, and each write log records index information of the corresponding data block.
  • each SSD in the SSD array maintains an FTL list of the node, and an FTL list of the local node of each SSD It is used to record the mapping relationship between the logical address and the physical address of the data block saved by each solid state hard disk.
  • the controller triggers the FTL refresh of the n data blocks, and specifically includes: the controller generates n refresh commands, wherein the n refresh commands are in one-to-one correspondence with the n data blocks; the controller refreshes the n data blocks The commands are respectively sent to the n SSDs, where each refresh command is used to request the FTL refresh of the corresponding SSD by the corresponding SSD.
  • This implementation mode provides a method for the FTL refresh of the controller trigger data block when each solid state hard disk manages the FTL, which complements the integrity of the solution.
  • the method further includes: receiving, by the controller, a refresh success message sent by any one of the n solid state disks The controller deletes the write log of the data block corresponding to the SSD that sent the refresh success message.
  • This implementation provides a method for processing a write log, which avoids the need to refresh all data blocks when FTL refresh is triggered again, and improves the efficiency of writing data to be written.
  • the controller maintains a global FTL list, where the global FTL list is used to record a logical address of the data block saved on the SSD array
  • the triggering the FTL refresh of the n data blocks by the controller includes: the controller establishing the mapping in the global FTL list according to the mapping relationship between the logical address stored in each data block of the n data blocks and the current physical address.
  • This implementation mode provides a method for the controller to trigger the FTL refresh of the data block when the controller manages the FTL, which complements the integrity of the solution.
  • the controller completes an FTL refresh of a data block, and deletes a write log corresponding to the data block that is FTL refreshed.
  • This implementation provides a method for processing a write log, which avoids the need to refresh all data blocks when FTL refresh is triggered again, and improves the efficiency of writing data to be written.
  • the controller determines The state of the stored write log in the non-volatile memory, if the write log stored in the non-volatile memory of the controller is in the first state, the controller initiates the write operation of the write of the data to be written again.
  • the controller determines a state of the stored write log in the nonvolatile memory, if the write log stored in the nonvolatile memory of the controller is in the second state, triggered again according to the write log stored in the nonvolatile memory of the controller FTL refresh.
  • the controller triggers the FTL refresh again according to the write log in the nonvolatile memory of the controller, where the controller acquires The index information of each write log in the non-volatile memory is obtained according to the index information of each write log, and the mapping relationship between the logical address of the data block corresponding to each write log and the current physical address is obtained.
  • the mapping relationship between the logical address of the data block corresponding to each write log and the current physical address is established in the FTL list.
  • the controller acquires index information of each write log in the non-volatile memory, and sends the obtained index information of each write log to the corresponding SSD, so that the corresponding SSD is obtained according to the obtained index.
  • mapping between the logical address of the data block corresponding to each write log and the current physical address in the information acquisition cache, and establishing the logical address of the data block corresponding to each write log and the current physical address in the FTL list of the node The mapping relationship between.
  • the controller determines whether to complete the FTL refresh of one data block, and deletes the write log corresponding to the data block that is completed by the FTL refresh.
  • the present invention provides a controller for a solid state disk array, the controller including a data acquisition unit, a write command generation unit, a write command transmission unit, a determination unit, and a refresh unit.
  • the data receiving unit is configured to acquire data to be written, where the data to be written includes n data blocks;
  • the write command generating unit is configured to generate n write commands, wherein the n write commands and the n data blocks are
  • the write command sending unit is configured to separately send the n write commands to the n solid state hard disks in the SSD array, wherein the n write commands are in one-to-one correspondence with the n solid state disks, each write The command is used to request the corresponding SSD to store the data block corresponding to the write command;
  • the determining unit is configured to determine whether to receive the n write success responses corresponding to the n SSDs; if it is determined that the n corresponding write commands are received, a write success response, the refresh unit is configured to perform a Flash Translation Layer (FTL)
  • Separating the data storage from the FTL list refresh ensures the atomicity of the write operation of the array controller or each solid state hard disk after the data storage to the idle physical storage space is interrupted. At the same time, the extra IO overhead caused by the successful write response in this method is relatively small, and will not have a great impact on the speed of data writing.
  • the controller further includes a log storage unit, if the determining unit determines that n write success responses corresponding to the n write commands are received, in the refresh unit Before performing the FTL refresh of the n data blocks, the log storage unit is configured to generate n write logs, and save the n write logs, the n write logs are in one-to-one correspondence with the n data blocks, and each write log Record the index information of the corresponding data block.
  • each SSD in the SSD array maintains an FTL list of the node, and an FTL list of the local node of each SSD It is used to record the mapping relationship between the logical address and the physical address of the data block saved by each solid state hard disk.
  • the refreshing unit is configured to trigger the FTL refresh of the n data blocks, and specifically includes: generating n refresh commands, wherein the n refresh commands are in one-to-one correspondence with the n data blocks; and the n refresh commands are respectively sent to The n SSDs, wherein each refresh command is used to request an FTL refresh of the corresponding SSD by the corresponding SSD.
  • the controller further includes a deleting unit, configured to: receive any one of the n solid state hard disks and send the SSD The refresh success message deletes the write log of the data block corresponding to the SSD that sent the refresh success message.
  • the controller further includes a maintenance unit, where the maintenance unit is configured to maintain a global FTL list, where the global FTL list is used to record the solid state
  • the refreshing unit is configured to trigger the FTL refresh of the n data blocks, specifically: the refreshing unit is configured to notify the maintenance unit to establish the global FTL list according to a logical address of each data block of the n data blocks.
  • the controller further includes a deleting unit, configured to: complete FTL refresh of a data block, and delete the FTL refresh. Write log of the data block.
  • the The determining unit is further configured to determine a state of the stored write log in the non-volatile memory of the controller. If the write log stored in the nonvolatile memory of the controller is in the first state, the write command transmitting unit is further configured to initiate a write operation of the data to be written again.
  • the determining unit further Used to determine the state of the stored write log in the controller's non-volatile memory. If the write log stored in the nonvolatile memory of the controller is in the second state, the refresh unit is further configured to trigger the FTL refresh again according to the write log stored in the nonvolatile memory of the controller.
  • the refreshing unit is configured to re-trigger the FTL refresh according to the write log in the non-volatile memory, specifically: acquiring the non-easy The index information of each write log in the memory is obtained, and according to the index information of each write log, the mapping relationship between the logical address of the data block corresponding to each write log and the current physical address is obtained, in the global FTL list.
  • the determining unit is further configured to determine whether to complete FTL refresh of a data block, where the deleting unit is further used to confirm the completion of the determining unit After the FTL of a data block is refreshed, the write log corresponding to the data block that has been FTL refreshed is deleted.
  • the present invention provides a controller for a solid state disk array, the controller comprising a processor and a memory, the processor for executing computer instructions stored in the memory to implement any one of the first aspects .
  • the present invention also provides a computer program product and a non-transitory computer readable storage medium, wherein the computer program product and the non-transitory computer readable storage medium comprise computer instructions, and the processor executes the computer instructions to implement the present invention Various methods in the first aspect.
  • the invention provides a solid state disk array comprising a controller and n solid state disks.
  • the controller is configured to: acquire data to be written, where the data to be written includes n data blocks; generate n write commands, wherein the n write commands are in one-to-one correspondence with the n data blocks; the n write commands Sended to the n SSDs respectively, wherein the n write commands are in one-to-one correspondence with the n SSDs.
  • Each of the n solid state disks is configured to: store a data block corresponding to the write command according to the corresponding write command; and store a data block corresponding to the write command, and send a write success response to the controller.
  • the controller is further configured to: determine whether the n write success responses corresponding to the n write commands are received; if it is determined that the n write success responses corresponding to the n write commands are received, triggering the flash conversion of the n data blocks
  • a Flash Translation Layer (FTL) refresh wherein the FTL refresh of each data block includes: establishing a mapping relationship between a logical address of each data block and a current physical address of the data block in the FTL list.
  • FTL Flash Translation Layer
  • the controller is further configured to: if it is determined that the n write success responses corresponding to the n write commands are received, triggering the FTL of the n data blocks Before the refresh, generate n write logs, and store the n write logs into the non-volatile memory of the controller.
  • the n write logs are in one-to-one correspondence with the n data blocks, and each write log record corresponds to the data block. Index information.
  • each SSD in the SSD array is further configured to: maintain an FTL list of the node, and each of the SSDs The FTL list of the node is used to record the mapping relationship between the logical address and the physical address of the data block saved by each SSD.
  • the controller is configured to trigger the FTL refresh of the n data blocks, and specifically includes: generating n refresh commands, wherein the n refresh commands are in one-to-one correspondence with the n data blocks; respectively, sending the n refresh commands to the n solid state drives;
  • Each SSD in the SSD array is further configured to: receive a refresh command sent by the controller, and establish a logical address between the logical address of each data block and a current physical address of the data block in an FTL list of the node. Mapping relations.
  • each SSD is further configured to: after completing the FTL refresh of the stored data block, send a refresh to the controller.
  • the controller is further configured to: receive a refresh success message sent by any one of the n solid state disks, and delete a write log of the data block corresponding to the solid state disk that sends the refresh success message.
  • the controller is further configured to: maintain a global FTL list, where the global FTL list is used to record data saved on the SSD array The mapping between the logical address of the block and the physical address.
  • the controller is configured to trigger the FTL refresh of the n data blocks, and specifically includes: establishing, according to a mapping relationship between a logical address stored by each data block of the n data blocks and a current physical address, in the global FTL list. A mapping relationship between a logical address of each data block of n data blocks and a current physical address of each data block of the n data blocks.
  • the controller is further configured to: complete FTL refresh of a data block, and delete a write log corresponding to the data block that is FTL refreshed. .
  • the controller is further configured to: the SSD array and/or the controller re-on After the power is determined, the state of the stored write log in the nonvolatile memory of the controller is determined, and if the write log stored in the nonvolatile memory of the controller is in the first state, the write operation of the data to be written is initiated again. .
  • the seventh implementation manner of the fifth aspect after the SSD and/or the controller is powered on, determine the controller The state of the write log stored in the nonvolatile memory, if the write log stored in the nonvolatile memory of the controller is in the second state, triggered again according to the write log stored in the nonvolatile memory of the controller FTL refresh.
  • the controller is configured to trigger the FTL refresh again according to the write log in the non-volatile memory of the controller, specifically: acquiring The index information of each write log in the non-volatile memory is obtained according to the index information of each write log, and the mapping relationship between the logical address of the data block corresponding to each write log and the current physical address is obtained. Establishing a mapping relationship between a logical address of the data block corresponding to each write log and the current physical address in the FTL list; or
  • a mapping relationship between a logical address of the data block corresponding to the write log and the current physical address, and a mapping relationship between the logical address of the data block corresponding to each write log and the current physical address is established in the FTL list of the local node.
  • the controller is further configured to: after determining that the FTL of one data block is refreshed again, deleting the data block that has completed the FTL refresh again Corresponding write log.
  • FIG. 1 is a schematic diagram of a solid state disk array structure using a RAID technology
  • FIG. 2 is an architectural diagram of a solid state disk array when FTL management is implemented in each solid state drive
  • FIG. 3 is a flow chart of writing data to be written when FTL management is implemented in an array controller
  • FIG. 5 is a schematic diagram of an array controller according to an embodiment of the present invention.
  • FIG. 6 is a schematic diagram of another array controller according to an embodiment of the present invention.
  • FIG. 7 is a schematic diagram of a solid state disk array according to an embodiment of the present invention.
  • the disks that make up the disk array can be flash-based solid state drives.
  • This embodiment provides a method and apparatus for data writing of a flash-based solid state hard disk array.
  • the disks that make up the disk array can be flash-based solid state drives. Since the memory bit of the flash memory can only be written from 0 to 1, the memory bit needs to be erased first when it is written from 1 to 0. Therefore, when the memory space in which the data is stored in the flash memory needs to be written twice, the first need to wipe In addition to this storage space. Moreover, since the write operation of the flash memory is performed in units of pages, and the erase operation is performed in units of blocks, if a second write operation is performed on a page in the flash memory, the block where the page is located needs to be erased, thereby reducing the entire IO operation. Efficiency, but also brings a lot of problems in storage space management.
  • the software layer is added to the flash-based SSD.
  • the in-place update can be implemented, that is, the storage space for the secondary write data written by the upper layer to see the SSD is still the original address space, and the FTL emulates the flash memory as a virtual storage with a logical address.
  • the virtual storage space of the virtual storage device is mapped to the physical storage space of the flash memory, the logical address is mapped to the physical address, and the input/output (I/O) operations of the data are performed by the logical address.
  • FTL implements address management of SSDs. It maintains the mapping between logical addresses and physical addresses of data stored in SSDs through FTL lists, and also manages the free address space of SSDs.
  • the data write operation method of the solid state hard disk that introduces the FTL is as follows: When a data needs to be written into a logical address of the solid state hard disk, the FTL allocates a free physical storage space for the data to which the data is not written. After the data is stored in the free physical storage space, the FTL refresh operation is performed. In the FTL list, the mapping between the logical address and the physical address of the free physical storage space is established, and the physical address of the logical address mapping is modified to the current physical address of the free physical storage space, and the data is written. Subsequent IO operations can be addressed to the current physical address based on the logical address of the data write.
  • the array controller 200 includes an array memory 101 and an array storage medium 102.
  • the array controller 200 is respectively connected to the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n in the solid state disk array.
  • the array controller 200 obtains data to be written, and the array processor 101 divides the data to be written into n data blocks, and stores each data block to the corresponding solid state hard disk 110a and the solid state hard disk 110b. , ... solid state drive 110n.
  • the array memory 101 is a volatile memory. After the array controller 200 is powered off, the array memory 101 loses the data stored thereon.
  • the array storage medium 102 is a non-volatile memory for storing data that needs to be retained after the array controller 200 is powered off.
  • the array controller 200 may be a RAID card that is independent of the SSD array, or an array manager of the SSD array. This embodiment does not limit this.
  • the array controller 100 is an array manager of a solid state disk array
  • the controller, the memory and the storage medium of the array manager are used to implement the array in this embodiment in addition to the functions implemented by the array manager in the prior art.
  • the functions implemented by the memory 101 and the array storage medium 102 are used to implement the array in this embodiment in addition to the functions implemented by the array manager in the prior art.
  • the SSD array When the SSD array is written to the data to be written, if the array controller and/or each SSD are powered off when the corresponding SSDs are stored in the SSDs, the SSD will be successfully written and partially solid.
  • the write operation of the hard disk fails, and the atomicity of the data write operation cannot be guaranteed. As a result, the data in each strip of the SSD array is inconsistent, and the data redundancy provided by the RAID technology fails.
  • the atomicity of the write operation of each solid state hard disk in the SSD array is ensured based on the feature that the SSD refreshes the FTL list after the SSD writes the data, and the method provided in this embodiment writes the SSD.
  • the operation of writing data at the time of data is separated from the operation of refreshing the FTL list.
  • the array controller 100 performs data segmentation on the data to be written, and divides the data to be written into n data blocks.
  • the number of data segments to be written is the same as the number of solid state disks in the SSD array, and each data block is written.
  • the algorithm for dividing the data to be written into data blocks is not limited.
  • the n data blocks are directly divided into n blocks by the RAID controller according to a certain algorithm, and may be data to be entered. One-ninth of the number may also contain a verification block containing a verification code.
  • the array controller 100 generates a write command for each data block and sends it to the corresponding solid state hard disk 110a, solid state hard disk 110b, ...
  • each data block has a one-to-one correspondence with each solid state hard disk.
  • the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n receives a write command, and stores each data block into an idle physical storage space in each solid state hard disk. After the data blocks are stored in the free physical storage space, the SSDs do not perform the FTL refresh, but return a write success response to the array controller 100, respectively. After receiving the n write success responses returned by the solid state hard disk 110a, the solid state hard disk 110b, and the solid state hard disk 110n, the array controller triggers the FTL refresh operation to complete the writing of the data to be written.
  • the FTL refresh establishes a mapping relationship between the logical address of each data block and the current physical address in the FTL list according to the current physical address of the cached data block.
  • the current physical address that is, the physical address of the free physical storage space that each data block stores after n data blocks are stored to n solid state disks.
  • Separating the data storage from the FTL list refresh ensures the atomicity of the write operation of the array controller or each solid state hard disk after the data storage to the idle physical storage space is interrupted. If the array controller 100 is powered off when sending a write command to each solid state hard disk, or one or some of the n solid state hard disks are powered off when writing the data block to the idle physical storage space, obviously, the array control is performed at this time. If the device 100 does not receive n write success messages, the FTL list refresh operation will not be performed. At this time, the n solid state hard disks all fail to write the data block, thereby ensuring the atomicity of the write operation. At the same time, the extra IO overhead caused by the successful write response in this method is relatively small, and will not have a great impact on the speed of data writing.
  • the array controller 100 or each solid state hard disk is powered off during the FTL refresh, the FTL refresh of some solid state disks is completed, and the FTL refresh of the remaining solid state disks is not completed. Therefore, some solid state disks are successfully written.
  • the data block while some solid state drives are not successfully written to the corresponding data block, causing inconsistency in write operations.
  • the atomicity of the write operation in this case can be achieved by writing a log.
  • the array controller 100 After confirming that n write success responses have been received, the array controller 100 generates a write log for each data block and performs storage on the array storage medium 102 before performing the FTL refresh.
  • the array controller 100 stores n write logs in the array storage medium 102, and n write logs are in one-to-one correspondence with n data blocks, and each write log records index information of the corresponding data block.
  • the index information is used to identify the data block. If the array controller 100 or each solid state hard disk is powered off during the FTL refresh, after the power is turned back on, the array controller 100 can obtain the stored write log in the array storage medium 102.
  • the index information finds a logical address of the data block storage corresponding to the write log stored in the array storage medium 102 according to the index information.
  • the index information may also be referred to as address information and the like, which is not limited by the embodiment of the present invention.
  • the atomic method of ensuring the write operation according to the write log is as follows:
  • the array storage medium 102 is a nonvolatile memory
  • the n write logs stored in the array storage medium 102 are not lost after the power is turned off, and then the FTL refresh needs to be performed again according to the index information in the n write logs. data block.
  • the array controller 100 again triggers the FTL refresh of the data block to be FTL refreshed.
  • the array controller 100 obtains a mapping relationship between the logical address of the cached data block and the current physical address according to the index information in the log, and establishes a logical address and a current physical address of the data block corresponding to each write log in the FTL list.
  • mapping relationship between the logical address of the data block and the current physical address is cached in the non-volatile memory before the power is turned off.
  • the array controller 100 and/or the solid state disk array interrupts the process of writing data to be written, after the power is turned back on, the array controller 100 determines the state of the write log stored in the array storage medium 102, according to Write the status of the log and perform the corresponding operations to ensure the atomicity of the write operation.
  • the array controller 100 and/or the SSD array is powered off before starting the FTL refresh, the array controller 100 has not started the storage of the FTL refresh and the write log, and at this time, there is no write log in the array storage medium 102, or The stored write log contains an incomplete write log, and the write operation of each data block is unsuccessful. Therefore, the array controller 100 rewrites the write operation that initiates the data to be written. If the array controller 100 and/or the SSD array are interrupted during the refresh process of each data block FTL, at this time, the array storage medium 102 stores a write log, and some partial data blocks may have completed the FTL refresh. Therefore, the array controller 100 triggers the FTL refresh again based on the write log.
  • the first state refers to the absence of a write log in the array storage medium 102, or the write log stored in the storage medium 103 contains an incomplete write log.
  • the second state means that the write log is stored in the array storage medium 102, and the stored write log is complete. If the array controller 100 determines that the write log stored in the array storage medium 102 is in the first state after re-powering, the array controller 100 rewrites the write operation that initiates the data to be written; the write log is in the second state, and the array controller 100 triggers the FTL refresh again based on the write log. Further, after the array controller 100 triggers the FTL refresh of one data block, the write log corresponding to the data block is deleted.
  • the array controller 100 and/or each solid state hard disk is powered off when each data block FTL is refreshed, the write log corresponding to the data block that has completed the FTL refresh has been deleted, and the write log stored in the array storage medium 102 corresponds to A data block that has not been refreshed by FTL. After power-on, the array controller 100 only needs to trigger an FTL refresh of the data block that has not been FTL refreshed.
  • the array controller 100 and/or the SSD array are powered down again during some FTL refreshes of the data blocks.
  • the array controller 100 can still trigger the FTL refresh again by the data block that has not completed the FTL refresh according to the write log stored in the array storage medium 102.
  • the array controller 100 deletes the write log corresponding to the data block every time the data block completes the FTL refresh again, and then powers off and re-powers only after the power is turned off again. It is necessary to trigger an FTL refresh of a data block that has not been FTL refreshed.
  • FTL has two implementations in a solid-state disk array that uses RAID technology.
  • the first mode is shown in FIG. 1, and FTL management is implemented in the array controller 100.
  • the FTL function module on the array controller 100 manages the logical addresses and physical addresses of all the solid state disks in the SSD array, including the mapping relationship between the physical addresses and logical addresses saved by the data blocks on each SSD, and the SSDs. The physical address of the free physical storage space, etc.
  • the array controller 100 maintains a global FTL list, and the global FTL list records the mapping relationship between the logical address and the physical address of the data block stored on the SSD array.
  • the process of the array controller 100 writing the data to be written to the SSD array is as follows: First, the array controller 100 divides the data to be written into n data blocks, and Each data block is assigned a logical address in a solid state hard disk and a corresponding solid state hard disk; the FTL function module in the array controller 100 allocates a physical address of a free physical storage space in the corresponding solid state hard disk for each logical address, and the array control
  • the device 100 generates n data write commands including n physical addresses, and sends n data write commands to the corresponding solid state hard disk 110a, solid state hard disk 110b, ... solid state hard disk 110n; finally, in n solid state hard disks.
  • the FTL function module in the controller 100 modifies the global FTL list, and establishes the logical address and the current physical address written in the global FTL list.
  • the mapping relationship, wherein the current physical address written is the physical address corresponding to each logical address allocated by the FTL function module.
  • each of the solid state drives has a hard disk controller 111 and a hard disk memory 112.
  • the hard disk memory is used to store intermediate data during the operation of the hard disk controller.
  • the hard disk controller 111a, the hard disk controller 111b, ... the hard disk controller 111n realize FTL management of the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n, respectively.
  • Each SSD has an FTL list. Each FTL list is located in the corresponding SSD. Each FTL list maintains the mapping between logical addresses and physical addresses in the SSD.
  • each solid state hard disk controller the address management of the idle physical storage space of each solid state hard disk is also implemented in each hard disk controller.
  • the process of the array controller 100 writing the data to be written to the SSD array is as follows: the segment of the data to be written and the allocation of the corresponding logical address are still implemented by the array controller 100; After n data blocks and n logical addresses in the corresponding n solid state hard disks, the controller 100 generates a data write command including a corresponding logical address for each data block, and sends a n data write command to the next data block.
  • each hard disk controller After each solid state hard disk receives the corresponding data write command, each hard disk controller writes the corresponding data block to the logical address according to the logical address in the data write command.
  • Pointing to the virtual storage space specifically, the hard disk controller of each solid state hard disk first allocates a free physical address on the solid state hard disk to the logical address corresponding to the corresponding data block, and writes the database to the physical storage space pointed to by the free physical address.
  • the FTL refresh of the corresponding data block is performed, that is, each hard disk controller is in its own
  • the mapping between the logical address corresponding to the data block and the current physical address is established in the FTL list, where the current physical address is the free physical address allocated by each hard disk controller for the logical address.
  • FTL management can be implemented in the array controller 100 or in each solid state hard disk in the SSD array.
  • the data writing method to be written corresponding to the two implementation manners is slightly different. The following describes the flow of data writing to be written corresponding to the two implementation manners.
  • the array controller 100 When the FTL management is implemented in the array controller 100, the array controller 100 maintains a global FTL list, and the global FTL list records the mapping relationship between the logical addresses and physical addresses of the data blocks stored on the SSD array. As shown in FIG. 3, the process of writing data to be written is as follows:
  • the array controller 100 divides the data to be written into n data blocks, and generates a data write command including a physical address for each data block.
  • the array controller 100 will write the data to be written to the SSD array by data segmentation.
  • the array controller 100 divides n data blocks into one-to-one correspondence with n solid-state hard disks while dividing the data to be written into n data blocks.
  • the array controller 100 allocates a solid state hard disk for each data block, and determines a logical address of the solid state hard disk corresponding to each data block.
  • the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n respectively correspond to the logical address Addr1 and the logical address Addr2. ... logical address Addrn.
  • the global FTL list stores n logical addresses Addri. Before the data to be written is written, the n logical addresses Addri may have a mapped initial physical address in the global FTL list, or may not have a mapped original physical address.
  • the n solid-state hard disks write the data block to the physical storage space pointed to by the current physical address addri according to the current physical address addri allocated by the array controller 100.
  • the mapping relationship between the n logical addresses and the n current physical addresses is cached in the nonvolatile memory, and is used again after the power is turned off. FTL refresh of the data block.
  • mapping relationship between the logical address Addr, the current address addri and the logical address Addri and the current address addri corresponding to each data block is the intermediate data in the data writing process to be written, and is cached in the array controller 100.
  • the array controller 100 sends the generated n data write commands to the corresponding n SSDs.
  • the array controller 100 respectively sends each data write command to the corresponding solid state hard disk.
  • Each data write command includes a current address addri corresponding to a data block, that is, a data write command including the physical address addr1 and data including the physical address addr2.
  • the write command, ... the data write command including the physical address addrn is sent to the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n, respectively.
  • Each data write command triggers the corresponding solid state hard disk 110 to store the corresponding data block into the free physical storage space pointed by the corresponding current address addri.
  • each solid state hard disk 110i acquires a corresponding data block according to the data write command, and The current address addri corresponding to the data block is obtained in a data write command, where i is any value in 1-n. After storing the corresponding data to the free physical storage space pointed to by the physical address, each solid state drive returns a write success response to the array controller 100.
  • the array controller 100 receives a write success message returned by each solid state drive.
  • the array controller 100 needs to perform FTL refresh after confirming that all solid state disks complete the storage of the corresponding data block. Therefore, the array controller 100 needs to receive the write success message returned by each solid state hard disk, and does not perform the write log storage and the data block FTL refresh trigger before confirming that all n write success messages are received.
  • the array controller 100 After confirming that n write success messages are received, the array controller 100 stores n write logs.
  • the array control 100 stores n write logs into the array storage medium 102.
  • the n write logs are in one-to-one correspondence with the n data blocks, and each write log records the index information of the corresponding data block. After power-on, according to the index information, the array controller 100 can find the mapping relationship between the logical address Addri corresponding to the data block that has not completed the FTL refresh and the current physical address addri.
  • the array controller 100 After storing the n write logs, the array controller 100 triggers the FTL refresh of the data block.
  • the array controller 100 After confirming that n write logs are stored, the array controller 100 triggers the FTL refresh of the data block, that is, the mapping relationship between the logical address Addri and the current physical address addri is established in the global FTL list. Specifically, the array controller 100 reads the cached logical address Addr1, the logical address Addr2, the logical address Addrn, and the mapping relationship between the n logical addresses Addr and the n physical addresses addr in the array memory 101, and n logicalities. The mapping relationship between the address Addri and the n physical addresses addri is stored in the global FTL list. If the original mapping relationship between the logical address Addri and the original physical address is in the global FTL list, the original mapping relationship is deleted. When the FTL refresh of the data block is triggered, the array controller 100 deletes the write log corresponding to the data block that completes the FTL refresh every time the FTL refresh of one data block is completed.
  • the array manager 100 determines the state of the write log stored in the array storage medium 102, according to the array storage medium 102.
  • the different states of the stored write log the array manager 100 performs different steps to ensure the atomicity of the write operations to each data block.
  • the array controller 100 If the write log stored in the array storage medium 102 is in the first state, indicating that the array controller 100 has not started the storage of the FTL refresh and the write log when the power is off, the array controller 100 rewrites the write operation that initiates the data to be written; The write log stored in the array storage medium 102 is in a second state, indicating that n write logs have been generated when the power is off, and the array controller 100 triggers the storage in the array storage medium 102 again according to the write log stored in the array storage medium 102. Write the FTL refresh of the data block corresponding to the log.
  • the array controller 100 acquires the index information of the stored write log from the write log stored in the array storage medium 102, according to the index information and the logical address cached in the nonvolatile memory.
  • the mapping relationship between the Addri and the current physical address addri is used to establish a mapping relationship between the logical address Addri of the data block corresponding to the write log stored in the array storage medium 102 and the current physical address addri in the global FTL list.
  • each SSD has a list of FTLs.
  • the process of writing data to be written is as follows:
  • the array controller 200 divides the data to be written into n data blocks, and generates a data write command including a logical address Addri for each data block.
  • the array controller 200 maps the data to be written into n data blocks one by one with n solid hard disks.
  • the array controller 200 allocates a solid state hard disk for each data block, and determines a logical address Addri of the solid state hard disk corresponding to each data block.
  • the solid state hard disk 110a, the solid state hard disk 110b, ... the solid state hard disk 110n respectively correspond to a logical address Addr1 and a logical address Addr2. , ... logical address Addrn.
  • step 301 since each solid state hard disk performs its own FTL management, the operation of assigning the current physical address addri corresponding to the logical address Addri to the data block is performed by the hard disk controller 111 of each solid state hard disk.
  • the n data write commands generated by the array controller include n logical addresses Addri, and each data write command corresponds to a logical address Addri of one data block.
  • the array controller 200 sends the generated n data write commands to the corresponding n SSDs.
  • each solid state hard disk is obtained.
  • the hard disk controller 111 first uses the FTL management function to assign the current physical address addri to the data block to be written. Then, the hard disk controller 111 of each solid state hard disk stores the corresponding data block to the free physical storage space pointed by the current physical address addri.
  • the method for storing the corresponding data block by each solid state disk in this step is different from that in step 302.
  • each SSD After receiving the corresponding write command, each SSD obtains the logical address Addri from the write command, and uses the FTL management function to allocate the current physical address addri to the data block to be written, and stores the corresponding data block in the current data block.
  • the physical address addri points to the free physical storage space in the corresponding solid state hard disk, where i is any value in 1-n.
  • each solid state drive After storing the corresponding data to the free physical storage space pointed to by the current physical address addri, each solid state drive returns a write success message to the array controller 200.
  • each solid state drive After each solid state drive allocates the current physical address addri to the data block, it also caches the mapping relationship between the logical address Addri and the current physical address addri in the non-volatile memory for FTL refresh of the data block after power-off.
  • the array controller 200 receives a write success message returned by each solid state hard disk.
  • Step 403 is the same as step 303.
  • the array controller 200 does not perform the storage of the write log and the trigger of the data block FTL refresh.
  • the array controller 200 confirms that n write logs are stored after receiving n write success messages.
  • each write log records index information of the corresponding data block.
  • the array controller 200 can find the mapping relationship between the logical address Addri corresponding to the data block that has not completed the FTL refresh and the current physical address addri.
  • the array controller 200 confirms that the FTL refresh of the data block is triggered after storing n write logs.
  • step 405 is different from step 403.
  • the specific methods for triggering the FTL refresh of each data block include:
  • the array controller 200 After confirming that n write logs are stored, the array controller 200 sends a refresh command to the n solid state hard disks.
  • Each SSD receives a refresh command, and reads the mapping relationship between the logical address Addri cached on each solid state drive and the current physical address addri from the hard disk memory 112 of each hard disk controller, and establishes corresponding data in the FTL list of each solid state drive.
  • the solid state hard disk After a solid state hard disk completes the FTL refresh of the corresponding data block, the solid state hard disk returns a refresh success message to the array controller 200, and the array controller 200 deletes the write log corresponding to the data block that completes the FTL refresh.
  • the array manager 100 determines the state of the write log stored in the array storage medium 102, according to the array storage medium 102.
  • the different states of the stored write log the array manager 100 performs different steps to ensure the atomicity of the write operations to each data block.
  • the array controller 100 If the write log stored in the array storage medium 102 is in the first state, indicating that the array controller 100 has not started the storage of the FTL refresh and the write log when the power is off, the array controller 100 rewrites the write operation that initiates the data to be written; The write log stored in the array storage medium 102 is in a second state, indicating that n write logs have been generated when the power is off, and the array controller 100 triggers the storage in the array storage medium 102 again according to the write log stored in the array storage medium 102. Write the FTL refresh of the data block corresponding to the log.
  • the array controller 100 obtains the index information of the stored write log from the write log stored in the array storage medium 102, and the data block corresponding to the write log stored in the array storage medium 102.
  • the SSD is sent a refresh command; the SSD that receives the refresh command performs FTL refresh on the corresponding data block, that is, according to the mapping relationship between the logical address Addri and the current physical address addri cached in the nonvolatile memory, in the FTL list. Establishing a mapping between the logical address Addri of the data block corresponding to the write log stored in the array storage medium 102 and the current physical address addri relationship.
  • embodiments of the present invention also provide a controller 500 for a solid state disk array.
  • the controller 500 is configured to write the data to be written divided into n data blocks into n solid state disks in the solid state disk array, and ensure the atomicity of the write operation of the n data blocks.
  • the controller includes a data acquisition unit 501, a write command generation unit 502, a write command transmission unit 503, a determination unit 504, and a refresh unit 505.
  • the data receiving unit 501 is configured to acquire data to be written including n data blocks.
  • the write command generating unit 502 is configured to generate n write commands corresponding to n data blocks one by one after the data receiving unit 501 receives the data to be written.
  • the write command sending unit 503 is configured to separately send n write commands generated by the write command generating unit to n solid state hard disks in the solid state disk array, wherein n write commands are in one-to-one correspondence with n solid state disks, and each write command And configured to request the corresponding SSD to store the data block corresponding to the write command.
  • the determining unit 504 is configured to determine whether to receive n write success responses corresponding to the n SSDs.
  • the refreshing unit 505 is configured to trigger FTL refresh of the n data blocks, where the FTL refresh of each data block includes: establishing in the FTL list A mapping relationship between the logical address Addri of each of the n data blocks and the current physical address addri of the data block.
  • the controller 500 further includes a log storage unit 506 and a nonvolatile memory 507.
  • the log storage unit 506 is configured to generate n write logs, and n The write log is stored in the non-volatile memory 507, wherein n write logs are in one-to-one correspondence with n data blocks, and each write log records index information of the corresponding data block.
  • the controller 500 also includes a deletion unit 508. After the refresh unit 505 completes the FTL refresh of a data block, the deleting unit 508 is configured to delete the write log corresponding to the data block that completes the FTL refresh.
  • the refresh unit 505 for the FTL refresh of the n data blocks is specifically configured to: generate n refreshes corresponding to the n data blocks one by one. And sending, to the n solid-state hard disks, the refresh command is used to request the FTL refresh of the corresponding data block by the corresponding solid state hard disk.
  • the determining unit 504 is further configured to determine the state of the stored write log in the non-volatile memory 507. If the write log stored in the nonvolatile memory 507 is in the first state, the write command transmitting unit 503 is further configured to initiate a write operation of the data to be written again.
  • the refresh unit 505 is further configured to trigger the FTL refresh again according to the write log stored in the nonvolatile memory.
  • the deleting unit 508 is further configured to delete the write log corresponding to the data block that is completed or FTL refreshed again.
  • the refreshing unit 505 is used for the FTL to refresh again, and specifically includes: acquiring index information of each write log in the non-volatile memory 507, according to the index information of each of the write logs and the non-volatile memory of the n solid-state hard disks.
  • mapping between the logical address Addri of the data block corresponding to each write log and the current physical address addri is stored, and the logical address Addri of the data block corresponding to each write log and the current physical address addri are established in the FTL list of each solid state drive. The mapping relationship between them.
  • the refresh unit 505 for the FTL refresh of the n data blocks is specifically configured to: establish in the global FTL list according to the logical address of each data block of the n data blocks. A mapping relationship between the logical address Addri of each data block of n data blocks and the current physical address addri of each data block of n data blocks.
  • the determining unit 504 is further configured to determine the state of the stored write log in the non-volatile memory 507. If the write log stored in the nonvolatile memory 507 is in the first state, the write command transmitting unit 503 is further configured to initiate a write operation of the data to be written again.
  • the refresh unit 505 is further configured to trigger the FTL refresh again according to the write log stored in the nonvolatile memory.
  • the deleting unit 508 is further configured to delete the write log corresponding to the data block that has been completed or FTL refreshed again.
  • the refreshing unit 505 is configured to obtain the index information of each write log in the non-volatile memory 507, according to the index information of each write log and the non-volatile of the n solid-state hard disks.
  • the mapping between the logical address of the data block corresponding to each write log and the current physical address is stored in the memory, and the logical address of the data block corresponding to each write log is established between the global FTL list and the current physical address. Mapping relationship.
  • Each unit in the controller 500 can be implemented by a corresponding hardware chip. In another implementation, one or more of the units may be integrated on a single hardware chip. In another implementation, the various units in controller 500 can also be implemented by a processor executing computer instructions on a memory. This embodiment of the present invention does not limit this.
  • FIG. 6 is a schematic structural diagram of a controller 600 according to an embodiment of the present invention.
  • the controller 600 includes a processor 601 that is coupled to the controller memory 602.
  • the processor 601 can be a central processing unit (CPU), an image processing unit (GPU), a field programmable gate array (FPGA), or a digital signal processor (English: digital signal processor) , DSP) and other computing logic or a combination of any of the above computing logic.
  • the processor 601 can be a single core processor or a multi-core processor.
  • the bus 603 is used to transfer information between the components of the controller 600.
  • the bus 603 can use a wired connection or a wireless connection. This application does not limit this.
  • the bus 603 is also connected to a communication interface 604.
  • Communication interface 604 utilizes a transceiver, such as but not limited to a transceiver, to interface with external device 605, which may be interconnected to the network by wire or wireless.
  • the external device 605 includes a solid state disk array and a nonvolatile memory for buffering intermediate data during operation of the controller 600.
  • the method of the embodiment of the present invention may be completed/supported by the processor 601 executing software code in the controller memory 602.
  • FIG. 6 is merely an example of a controller 600 that may include more or fewer components than those shown in FIG. 6, or have different component configurations. Meanwhile, the various components shown in FIG. 6 can be implemented in hardware, software, or a combination of hardware and software.
  • Embodiments of the present invention also provide a solid state disk array. As shown in FIG. 7, the solid state disk array 700 includes a controller 710 and n solid state disks.
  • Controller 710 is used to:
  • n write commands are in one-to-one correspondence with n data blocks
  • N write commands are respectively sent to the n solid state hard disks, wherein n write commands are in one-to-one correspondence with n solid state hard disks.
  • Each solid state drive in n solid state drives is used to:
  • a write success response is sent to the controller 710. If it is determined that the n write success responses corresponding to the n write commands are received, the controller 710 performs FTL refresh of the n data blocks.
  • the controller 710 is also used to:
  • n write logs are generated, and n write logs are stored in the controller's nonvolatile memory before performing FTL refresh of n data blocks.
  • the n write logs are in one-to-one correspondence with the n data blocks, and each write log records index information of the corresponding data block.
  • the controller 710 is configured to perform the FTL refresh of the n data blocks, including:
  • n refresh commands wherein the n refresh commands are in one-to-one correspondence with the n data blocks;
  • Each SSD in the SSD array is further configured to: receive a refresh command sent by the controller, perform FTL refresh of the stored data block, that is, establish a logical address and data of each data block in an FTL list of the node. The mapping relationship between the current physical addresses of the blocks; after the FTL refresh of the stored data blocks is successful, a refresh success message is sent to the controller 710.
  • the controller 710 is further configured to: receive a refresh success message sent by any one of the n solid state disks, and delete a write log of the data block corresponding to the solid state disk that sends the refresh success message.
  • the controller 710 When the controller 710 is further configured to maintain the global FTL list, the controller is configured to perform FTL refresh of the n data blocks, specifically: according to a logical address of each data block of the n data blocks, in the global FTL list.
  • the mapping relationship between the logical address of each data block of n data blocks and the physical address of each data block of n data blocks is established; the FTL refresh of one data block is completed, and the data block corresponding to the FTL refresh is deleted. Write the log.
  • the controller 710 is further configured to determine a state of the stored write log in the non-volatile memory of the controller 710 after the SSD array and/or the controller is powered back up, if the controller 710 is in the non-volatile memory
  • the stored write log is in a first state, and the write operation of the data to be written is initiated again; after the SSD array and/or the controller is powered back on, determining storage in the nonvolatile memory of the controller 710
  • the state of the write log if the write log stored in the nonvolatile memory of the controller 710 is in the second state, the FTL is refreshed again according to the write log stored in the nonvolatile memory of the controller 710; determining to complete one data block After the FTL is refreshed again, the write log corresponding to the data block that has been FTL refreshed again is deleted.
  • the FTL refreshing specifically includes: acquiring index information of each write log in the non-volatile memory of the controller 710, and acquiring the controller 710 or the non-volatile memory of the n solid-state hard disks according to the index information of each of the write logs.
  • the mapping between the logical address of the data block corresponding to each write log and the current physical address is stored, and the mapping between the logical address of the data block corresponding to each write log and the current physical address is established in the FTL list. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Security & Cryptography (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)
  • Read Only Memory (AREA)

Abstract

一种数据写入的方法和固态硬盘阵列（700）。该固态硬盘阵列基于RAID系统，包括n个固态硬盘（110n），向该固态硬盘阵列（700）写入待写入数据时，将该待写入数据分为n个数据块，与n个固态硬盘（110n）一一对应。n个数据块全部存储至对应的固态硬盘（110n）后，再进行n个数据块的FTL刷新，完成待写入数据的写入。当固态硬盘阵列（700）在存储n个数据块时断电，由于尚未进行数据块的FTL刷新，n个数据块都未成功写入固态硬盘阵列（700），保证了n个数据块的写操作的原子性。

Description

一种数据写入的方法及固态硬盘阵列 技术领域
本申请涉及信息技术领域，尤其涉及一种数据写入的方法及固态硬盘阵列。
背景技术
磁盘阵列技术(Redundant Array of Independent Disks，RAID)即独立磁盘冗余阵列，它把多块独立的磁盘按不同方式组合起来形成一个磁盘阵列，利用个别磁盘提供的数据所产生加成效果提升整个磁盘阵列效能。RAID技术将数据切割成若干区段，每个区段为一个条带，分别存储在磁盘阵列的各个磁盘上，以提供数据冗余。利用同位检查，当磁盘阵列中任意一个磁盘发生故障丢失某段数据时，系统仍可读出完整的数据，并在数据重构时，将丢失的一段数据经计算后重新存入磁盘中。
应用RAID技术的磁盘阵列在将划分为条带的待写入数据写入各磁盘时，需要保证数据写操作的原子性，即，在将待写入数据写入磁盘阵列的一次写操作之后，对磁盘阵列内所有的磁盘的写操作需要都成功，或者都失败。若出现部分磁盘写入数据成功，而其余磁盘写入数据失败的情况，则会导致磁盘阵列内各条带数据不一致，导致RAID提供的数据冗余失效。
然而，当磁盘阵列的各磁盘进行并行的写操作时，若系统发生断电，会导致部分磁盘的写操作成功，而部分磁盘的写操作失败，无法保证数据写操作的原子性，导致磁盘阵列内各条带数据不一致，RAID技术提供的数据冗余失效。
发明内容
本发明实施例提供一种数据写入的方法以及固态硬盘阵列，可以实现n个数据块写入n个固态硬盘时，各数据块的写操作的原子性。
第一方面，本发明提供一种数据写入的方法，该方法应用于固态硬盘阵列，该方法包括：该固态硬盘阵列的控制器获取待写入数据，该待写入数据包括n个数据块。该控制器生成n个写命令，其中，该n个写命令与该n个数据块一一对应。该控制器将该n个写命令分别发送到该固态硬盘阵列中的n个固态硬盘，其中，该n个写命令与该n个固态硬盘一一对应，每个写命令用于请求对应的固态硬盘存储该写命令对应的数据块。该控制器确定是否接收到该n个写命令对应的n个写成功响应。如果确定接收到该n个写命令对应的n个写成功响应，该控制器触发该n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立该每个数据块的逻辑地址与该数据块的当前物理地址之间的映射关系。
将数据存储和FTL列表刷新分离，保证了阵列控制器或各固态硬盘在数据存储至空闲物理存储空间的过程中断电后写操作的原子性。同时，此方法中写成功响应所带来的额外的IO开销比较小，不会对数据写入的速度带来很大的影响。
结合第一方面，在第一方面的第一种实现方式中，如果确定接收到该n个写命令对 应的n个写成功响应，在该控制器触发该n个数据块的FTL刷新之前，该控制器生成n个写日志，将该n个写日志存入非易失存储器，该n个写日志与该n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
当阵列控制器100或各固态硬盘在进行FTL刷新时发生断电部分固态硬盘的FTL刷新完成，而其余固态硬盘的FTL刷新未完成，因此，部分固态硬盘成功写入对应的数据块，而部分固态硬盘未成功写入对应的数据块，造成写操作的不一致。写日志解决了这一不一致。
结合第一方面的第一种实现方式，在第一方面的第二种实现方式中，该固态硬盘阵列中的每个固态硬盘维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录该每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系。该控制器触发该n个数据块的FTL刷新，具体包括：该控制器生成n个刷新命令，其中，该n个刷新命令与该n个数据块一一对应；该控制器将该n个刷新命令分别发送到该n个固态硬盘，其中，每个刷新命令用于请求对应的固态硬盘对对应的数据块进行的FTL刷新。
本实现方式给出了各固态硬盘管理FTL时控制器触发数据块的FTL刷新的方法，补充了方案的完整性。
结合第一方面的第二种实现方式，在第一方面的第三种实现方式中，该方法还包括：该控制器接收该n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，该控制器删除发送该刷新成功消息的固态硬盘对应的数据块的写日志。
本实现方式提供了处理写日志的一种方法，避免了再次触发FTL刷新时，需要刷新全部数据块的情况，提高了待写入数据写入的效率。
结合第一方面的第一种实现方式，在第一方面的第四种实现方式中，该控制器维护全局FTL列表，该全局FTL列表用于记录该固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系。该控制器触发该n个数据块的FTL刷新包括：该控制器根据缓存的该n个数据块的每个数据块存储的逻辑地址与当前物理地址的映射关系，在该全局FTL列表中建立该n个数据块的每个数据块的逻辑地址与该n个数据块的每个数据块的当前物理地址之间的映射关系。
本实现方式给出了控制器管理FTL时控制器触发数据块的FTL刷新的方法，补充了方案的完整性。
结合第一方面的第四种实现方式，在第一方面的第五种实现方式中，该控制器完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
本实现方式提供了处理写日志的一种方法，避免了再次触发FTL刷新时，需要刷新全部数据块的情况，提高了待写入数据写入的效率。
结合第一方面的第一种至第六种中的任一实现方式，在第一方面的第七种实现方式中，该固态硬盘阵列和/或该控制器重新上电之后，该控制器确定该非易失存储器中的存储的写日志的状态，如果该控制器的非易失存储器中存储的写日志为第一状态，该控制器再次发起该待写入数据写的写操作。
结合第一方面的第一种至第六种中的任一实现方式，在第一方面的第七种实现方式中，该固态硬盘阵列和/或该控制器重新上电之后，该控制器确定该非易失存储器中的 存储的写日志的状态，如果该控制器的非易失存储器中存储的写日志为第二状态，根据该控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
结合第一方面的第七种实现方式，在第一方面的第八种实现方式中，该控制器根据该控制器的非易失存储器中的写日志，再次触发FTL刷新包括：该控制器获取该非易失存储器中的每个写日志的索引信息，根据该每个写日志的索引信息，获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系。或者，该控制器获取该非易失存储器中的每个写日志的索引信息，将获取的该每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系。
结合第一方面的第八种实现方式，在第一方面的第九种实现方式中，该控制器确定是否完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
第二方面，本发明提供一种固态硬盘阵列的控制器，该控制器包括数据获取单元、写命令生成单元、写命令发送单元、判断单元和刷新单元。该数据接收单元用于获取待写入数据，该待写入数据包括n个数据块；该写命令生成单元用于生成n个写命令，其中，该n个写命令与该n个数据块一一对应；该写命令发送单元用于将该n个写命令分别发送到该固态硬盘阵列中的n个固态硬盘，其中，该n个写命令与该n个固态硬盘一一对应，每个写命令用于请求对应的固态硬盘存储该写命令对应的数据块；该判断单元用于确定是否接收该n个固态硬盘对应的n个写成功响应；如果确定接收到该n个写命令对应的n个写成功响应，该刷新单元用于，进行该n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立该每个数据块的逻辑地址与该数据块的当前物理地址之间的映射关系。
将数据存储和FTL列表刷新分离，保证了阵列控制器或各固态硬盘在数据存储至空闲物理存储空间的过程中断电后写操作的原子性。同时，此方法中写成功响应所带来的额外的IO开销比较小，不会对数据写入的速度带来很大的影响。
结合第二方面，在第二方面的第一种实现方式中，该控制器还包含日志存储单元，如果该判断单元确定接收到该n个写命令对应的n个写成功响应，在该刷新单元进行该n个数据块的FTL刷新之前，该日志存储单元用于，生成n个写日志，保存该n个写日志，该n个写日志与该n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
当阵列控制器100或各固态硬盘在进行FTL刷新时发生断电部分固态硬盘的FTL刷新完成，而其余固态硬盘的FTL刷新未完成，因此，部分固态硬盘成功写入对应的数据块，而部分固态硬盘未成功写入对应的数据块，造成写操作的不一致。写日志解决了这一不一致。
结合第二方面的第一种实现方式，在第二方面的第二种实现方式中，该固态硬盘阵 列中的每个固态硬盘维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录该每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系。该刷新单元用于触发该n个数据块的FTL刷新，具体包括：生成n个刷新命令，其中，该n个刷新命令与该n个数据块一一对应；将该n个刷新命令分别发送到该n个固态硬盘，其中，每个刷新命令用于请求对应的固态硬盘对对应的数据块进行的FTL刷新。
结合第二方面的第二种实现方式，在第二方面的第三种实现方式中，该控制器还包括删除单元，该删除单元用于：接收该n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，删除发送该刷新成功消息的固态硬盘对应的数据块的写日志。
结合第二方面的第一种实现方式，在第二方面的第四种实现方式中，该控制器还包括维护单元，该维护单元用于维护全局FTL列表，该全局FTL列表用于记录该固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系。该刷新单元用于触发该n个数据块的FTL刷新，具体包括：该刷新单元用于根据该n个数据块的每个数据块的逻辑地址，通知该维护单元在该全局FTL列表中建立该n个数据块的每个数据块的逻辑地址与该n个数据块的每个数据块的物理地址之间的映射关系。
结合第二方面的第四种实现方式，在第二方面的第五种实现方式中，该控制器还包括删除单元，该删除单元用于：完成一个数据块的FTL刷新，删除完成FTL刷新的数据块的写日志。
结合第二方面的第一种至第五种中的任一实现方式，在第二方面的第六种实现方式中，该固态硬盘阵列重新上电和/或该控制器重新上电之后，该判断单元还用于，确定该控制器的非易失存储器中的存储的写日志的状态。如果该控制器的非易失存储器中存储的写日志为第一状态，该写命令发送单元还用于再次发起该待写入数据的写操作。
结合第二方面的第一种至第六种中的任一实现方式，在第二方面的第七种实现方式中，该固态硬盘阵列和/或该控制器重新上电之后，该判断单元还用于，确定该控制器的非易失存储器中的存储的写日志的状态。如果该控制器的非易失存储器中存储的写日志为第二状态，该刷新单元还用于，根据该控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
结合第二方面的第七种实现方式，在第二方面的第八种实现方式中，该刷新单元用于根据该非易失存储器中的写日志，再次触发FTL刷新具体包括：获取该非易失存储器中的每个写日志的索引信息，根据该每个写日志的索引信息，获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系；或者，获取该非易失存储器中的每个写日志的索引信息，将获取的该每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系。
结合第二方面的第八种实现方式，在第二方面的第九种实现方式中，该判断单元还用于确定是否完成一个数据块的FTL刷新；该删除单元还用于该判断单元确认完成一个数据块的FTL刷新后，删除完成了FTL刷新的数据块对应的写日志。
第三方面，本发明提供了一种固态硬盘阵列的控制器，该控制器包括处理器和存储器，该处理器用于运行该存储器中存储的计算机指令来实现第一方面中的任一种实现方式。
第四方面，本发明还提供了计算机程序产品和非易失性计算机可读存储介质，其中计算机程序产品和非易失性计算机可读存储介质中包含计算机指令，处理器执行计算机指令实现本发明第一方面中的各种方法。
第五方面，本发明一种固态硬盘阵列，该固态硬盘阵列包含控制器和n个固态硬盘。该控制器用于：获取待写入数据，该待写入数据包括n个数据块；生成n个写命令，其中，该n个写命令与该n个数据块一一对应；该n个写命令分别发送到该n个固态硬盘，其中，该n个写命令与该n个固态硬盘一一对应。该n个固态硬盘中的每个固态硬盘用于：根据对应的写命令，存储写命令对应的数据块；存储写命令对应的数据块后，向该控制器发送写成功响应。该控制器还用于：确定是否接收到该n个写命令对应的n个写成功响应；如果确定接收到该n个写命令对应的n个写成功响应，触发该n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立该每个数据块的逻辑地址与该数据块的当前物理地址之间的映射关系。
结合第五方面，在第五方面的第一种实现方式中，该控制器还用于：如果确定接收到该n个写命令对应的n个写成功响应，在触发该n个数据块的FTL刷新之前，生成n个写日志，将该n个写日志存入该控制器的非易失存储器，该n个写日志与该n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
结合第五方面的第一种实现方式，在第五方面的第二种实现方式中，该固态硬盘阵列中的每个固态硬盘还用于：维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录该每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系。该控制器用于触发该n个数据块的FTL刷新，具体包括：生成n个刷新命令，其中，该n个刷新命令与该n个数据块一一对应；将该n个刷新命令分别发送到该n个固态硬盘；。该固态硬盘阵列中的每个固态硬盘还用于：接收该控制器发送的刷新命令，在本节点的FTL列表中建立该每个数据块的逻辑地址与该数据块的当前物理地址之间的映射关系。
结合第五方面的第二种实现方式，在第五方面的第三种实现方式中，每个固态硬盘还用于：在完成存储的数据块的FTL刷新后，向该控制器发送一个刷新成功消息。该控制器还用于：接收该n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，删除该发送刷新成功消息的固态硬盘对应的数据块的写日志。
结合第五方面的第一种实现方式，在第五方面的第四种实现方式中，该控制器还用于：维护全局FTL列表，该全局FTL列表用于记录该固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系。该控制器用于触发该n个数据块的FTL刷新，具体包括：根据缓存的该n个数据块的每个数据块存储的逻辑地址与当前物理地址的映射关系，在该全局FTL列表中建立该n个数据块的每个数据块的逻辑地址与该n个数据块的每个数据块的当前物理地址之间的映射关系。
结合第五方面的第四种实现方式，在第五方面的第五种实现方式中，该控制器还用于：完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
结合第五方面的第一种至第六种中的任一实现方式，在第五方面的第六种实现方式中，该控制器还用于：该固态硬盘阵列和/或该控制器重新上电之后，确定该控制器的非易失存储器中的存储的写日志的状态，如果该控制器的非易失存储器中存储的写日志为第一状态，再次发起该待写入数据的写操作。
结合第五方面的第一种至第六种中的任一实现方式，在第五方面的第七种实现方式中，该固态硬盘阵列和/或该控制器重新上电之后，确定该控制器的非易失存储器中的存储的写日志的状态，如果该控制器的非易失存储器中存储的写日志为第二状态，根据该控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
结合第五方面的第七种实现方式，在第五方面的第八种实现方式中，该控制器用于根据该控制器的非易失存储器中的写日志，再次触发FTL刷新，具体包括：获取该非易失存储器中的每个写日志的索引信息，根据该每个写日志的索引信息，获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系；或者，
获取该非易失存储器中的每个写日志的索引信息，将获取的该每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的该每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立该每个写日志对应的数据块的逻辑地址与该当前物理地址之间的映射关系。
结合第五方面的第八种实现方式，在第五方面的第九种实现方式中，该控制器还用于：确定完成一个数据块的FTL再次刷新后，删除完成了再次FTL刷新的数据块对应的写日志。
附图说明
图1为应用RAID技术的固态硬盘阵列架构图；
图2为FTL管理在各固态硬盘中实现时的固态硬盘阵列的架构图；
图3为FTL管理在阵列控制器中实现时，待写入数据写入的流程图；
图4为FTL管理在各固态硬盘中实现时，待写入数据写入的流程图；
图5为本发明实施例中一种阵列控制器的示意图；
图6为本发明实施例中另一种阵列控制器的示意图；
图7为本发明实施例中一种固态硬盘阵列的示意图。
具体实施方式
组成磁盘阵列的磁盘可以是基于闪存的固态硬盘。本实施例提供一种基于闪存的固态硬盘阵列的数据写入的方法和装置。
组成磁盘阵列的磁盘可以是基于闪存的固态硬盘。由于闪存的存储位只能由0写到1，存储位从1写到0时需要首先擦除该数据位，因此，闪存中存储有数据的存储空间需要二次写入数据时，首先需要擦除该存储空间。又由于闪存的写操作是以页为单位进行，而擦除操作以块为单位进行，若对闪存中某页面进行二次写操作，需要擦 除该页面所在的块，降低了整个IO操作的效率，同时也带来了很多存储空间管理上的问题。因此，在基于闪存的固态硬盘中加入软件层，即闪存转换层(Flash Translator Layer，FTL)。为了使数据写入固态硬盘时能够实现原地更新，即，使得从上层看到固态硬盘的二次写数据写入的存储空间仍为原地址空间，FTL将闪存模拟为具有逻辑地址的虚拟存储设备，虚拟存储设备的虚拟存储空间映射到闪存的物理存储空间，逻辑地址映射到物理地址，数据的输入输出(Input/Output，I/O)操作都通过逻辑地址来进行。FTL实现固态硬盘的地址管理，它通过FTL列表来维护存储在固态硬盘中的数据的逻辑地址与物理地址的映射关系，同时也管理固态硬盘的空闲地址空间。
引入FTL的固态硬盘的数据写操作方法如下：需把数据写入固态硬盘的某逻辑地址时，FTL为数据分配一块未写入数据的空闲物理存储空间。数据存储至该空闲物理存储空间后，进行FTL刷新的操作。FTL刷新即在FTL列表中，建立逻辑地址与该空闲物理存储空间的物理地址的映射关系，将该逻辑地址映射的物理地址修改为该空闲物理存储空间的当前物理地址，完成数据的写入，使得后续的IO操作能够根据数据写入的逻辑地址，寻址到该当前物理地址。
应用RAID技术的固态硬盘阵列在实现上需要一个阵列控制器，如图1所示，阵列控制器200包括阵列内存101和阵列存储介质102。阵列控制器200分别与固态硬盘阵列中的固态硬盘110a、固态硬盘110b、……固态硬盘110n相连接。数据写入固态硬盘阵列时，阵列控制器200获得待写入数据，阵列处理器101将待写入数据分割为n个数据块，并将各数据块存储至对应的固态硬盘110a、固态硬盘110b、……固态硬盘110n。阵列内存101为易失性存储器，阵列控制器200断电后，阵列内存101即丢失其上存储的数据。阵列存储介质102为非易失性存储器，用于存储阵列控制器200断电后仍需保留的数据。阵列控制器200可以是独立于固态硬盘阵列的RAID卡，也可以是固态硬盘阵列的阵列管理器，本实施例对此不做限制。当阵列控制器100是固态硬盘阵列的阵列管理器，该阵列管理器的控制器、内存和存储介质除了实现现有技术中阵列管理器实现的功能之外，还用于实现本实施例中阵列内存101和阵列存储介质102实现的功能。
固态硬盘阵列在待写入数据写入时，若阵列控制器和/或各固态硬盘，在各固态硬盘存储对应的数据块时发生断电，会导致部分固态硬盘的写操作成功，而部分固态硬盘的写操作失败，无法保证数据写操作的原子性，导致固态硬盘阵列内各条带数据不一致，RAID技术提供的数据冗余失效。
为了解决这一问题，保证固态硬盘阵列中的各固态硬盘的写操作的原子性，基于固态硬盘在写数据时先写入数据后刷新FTL列表的特性，本实施例提供的方法将固态硬盘写数据时的写入数据的操作与刷新FTL列表的操作分离，
具体如下：
阵列控制器100对待写入数据进行数据分段，将待写入数据划分为n个数据块，待写入数据分段的数量与固态硬盘阵列中固态硬盘的数量相同，每个数据块写入一个固态硬盘。本实施例对将待写入数据划分为数据块的算法不做限制，n个数据块是由RAID控制器按照一定算法直接将代写入数据分割为n块得到，可以是都是待入数据的n分之一，也可以包含含有验证码的验证块。阵列控制器100为每个数据块生成一个写命 令，发送给对应的固态硬盘110a、固态硬盘110b、……固态硬盘110n，每个数据块与每个固态硬盘一一对应。固态硬盘110a、固态硬盘110b、……固态硬盘110n接收写命令，分别将各数据块存储至各固态硬盘中的空闲物理存储空间。各固态硬盘在数据块存储至空闲物理存储空间后，先不进行FTL刷新，而是分别向阵列控制器100返回写成功响应。阵列控制器接收到固态硬盘110a、固态硬盘110b、……固态硬盘110n返回的n个写成功响应后，再触发FTL刷新的动作，完成待写入数据的写入。FTL刷新即根据缓存的数据块的当前物理地址，在FTL列表中建立每个数据块的逻辑地址与当前物理地址的映射关系。当前物理地址即n个数据块分别存储至n个固态硬盘后，每个数据块存储的空闲物理存储空间的物理地址。
将数据存储和FTL列表刷新分离，保证了阵列控制器或各固态硬盘在数据存储至空闲物理存储空间的过程中断电后写操作的原子性。若阵列控制器100在向各固态硬盘发送写命令时断电，或n个固态硬盘中的某个或某几个在将数据块写入空闲物理存储空间时断电，显然，此时阵列控制器100没有接收到n个写成功消息，则FTL列表刷新的操作就不会进行，此时，n个固态硬盘都写入数据块失败，保证了写操作的原子性。同时，此方法中写成功响应所带来的额外的IO开销比较小，不会对数据写入的速度带来很大的影响。
进一步地，当阵列控制器100或各固态硬盘在进行FTL刷新时发生断电，此时部分固态硬盘的FTL刷新完成，而其余固态硬盘的FTL刷新未完成，因此，部分固态硬盘成功写入对应的数据块，而部分固态硬盘未成功写入对应的数据块，造成写操作的不一致。这种情况下写操作的原子性，可由写日志实现。确认收到n个写成功响应之后，进行FTL刷新之前，阵列控制器100为每个数据块生成一个写日志，并将存储于阵列存储介质102。即，阵列控制器100将n个写日志存入阵列存储介质102，n个写日志与n个数据块一一对应，每个写日志记录对应的数据块的索引信息。该索引信息用于标识数据块，若阵列控制器100或各固态硬盘在进行FTL刷新时发生断电，重新上电后，阵列控制器100可以在阵列存储介质102中获取存储的写日志中的索引信息，根据索引信息找到阵列存储介质102中存储的写日志对应的数据块存储的逻辑地址。本发明实施例中，索引信息还可以叫做地址信息等，本发明实施例对此不做限制。
若阵列控制器100或各固态硬盘在进行FTL刷新时断电，根据写日志保证写操作的原子性的方法具体如下：
由于阵列存储介质102为非易失存储器，存储于阵列存储介质102中的n个写日志在断电之后不会丢失，那么根据n个写日志中的索引信息即可得到需要再次进行FTL刷新的数据块。重新上电后，阵列控制器100再次触发需进行FTL刷新的数据块的FTL刷新。具体地，阵列控制器100根据日志中的索引信息，获取缓存的数据块的逻辑地址与当前物理地址的映射关系，在FTL列表中建立每个写日志对应的数据块的逻辑地址与当前物理地址之间的映射关系，其中数据块的逻辑地址与当前物理地址的映射关系在断电之前，即缓存在非易失存储器中。综上，若阵列控制器100和/或固态硬盘阵列在待写入数据写入的过程中断电，重新上电之后，阵列控制器100判断阵列存储介质102中存储的写日志的状态，根据写日志的状态，进行对应的操作，以保证写操作的原子性。由上述可知，若阵列控制器100和/或固态硬盘阵列在开始FTL刷新之前断 电，阵列控制器100尚未开始FTL刷新和写日志的存储，此时阵列存储介质102中不存在写日志，或者存储的写日志中包含有不完整的写日志，各数据块的写操作都没有成功，因此，阵列控制器100重写发起待写入数据的写操作。若阵列控制器100和/或固态硬盘阵列在各数据块FTL刷新的过程中断电，此时阵列存储介质102中存储有写日志，可能存在部分数据块已完成FTL刷新，因此，阵列控制器100根据写日志再次触发FTL刷新。
即，第一状态是指阵列存储介质102中不存在写日志，或存储介质103中存储的写日志中包含有不完整的写日志。第二状态是指阵列存储介质102中存储有写日志，且存储的写日志都完整。若重新上电后，阵列控制器100判断阵列存储介质102中存储的写日志为第一状态，阵列控制器100重写发起待写入数据的写操作；写日志为第二状态，阵列控制器100根据写日志再次触发FTL刷新。进一步地，阵列控制器100触发一个数据块的FTL刷新后，删除该数据块对应的写日志。也就是说，若阵列控制器100和/或各固态硬盘在各数据块FTL刷新时断电，已完成FTL刷新的数据块对应的写日志已被删除，阵列存储介质102中存储的写日志对应未FTL刷新的数据块。重新上电之后，阵列控制器100只需要触发未进行FTL刷新的数据块的FTL刷新。
在一些情况下，再次触发FTL刷新后，阵列控制器100和/或固态硬盘阵列在一些数据块进行FTL刷新的过程中再次断电。重新上电后，阵列控制器100仍可根据阵列存储介质102中存储的写日志来触发未完成FTL刷新的数据块再次进行FTL刷新。同样地，数据块在断电后进行FTL再次刷新时，每当一个数据块完成FTL再次刷新，阵列控制器100即删除该数据块对应的写日志，那么再次断电并重新上电之后，只需要触发未进行FTL刷新的数据块的FTL刷新。
FTL在应用RAID技术的固态硬盘阵列中有两种实现方式。
第一种方式如图1所示，FTL管理在阵列控制器100中实现。此时，阵列控制器100上的FTL功能模块管理固态硬盘阵列中的所有固态硬盘的逻辑地址与物理地址，包括各固态硬盘上数据块保存的物理地址与逻辑地址的映射关系、各固态硬盘上的空闲物理存储空间的物理地址等。具体地，阵列控制器100中维护有全局FTL列表，全局FTL列表记录固态硬盘阵列上保存的数据块的逻辑地址与物理地址的映射关系。当FTL在阵列控制器100中实现，此时，阵列控制器100向固态硬盘阵列写入待写入数据的过程如下：首先阵列控制器100将待写入数据分为n个数据块，并为每个数据块分配一个固态硬盘和对应固态硬盘中的逻辑地址；阵列控制器100中的FTL功能模块为每个逻辑地址在对应的固态硬盘中分配一个空闲物理存储空间的物理地址后，阵列控制器100生成包含n个物理地址的n个数据写命令，并将n个数据写命令一一下发给对应的固态硬盘110a、固态硬盘110b、……固态硬盘110n；最后，在n个固态硬盘分别接收对应的写命令、分别将数据写入物理地址指向的空闲物理存储空间后，控制器100中的FTL功能模块修改全局FTL列表，在全局FTL列表中建立逻辑地址与写入的当前物理地址的映射关系，其中，写入的当前物理地址即FTL功能模块分配的每个逻辑地址对应的物理地址。
第二种方式如图2所示，FTL管理在各固态硬盘中实现。此时，除了闪存颗粒以外， 各固态硬盘中分别具有硬盘控制器111、硬盘内存112。硬盘内存用于硬盘控制器运行过程中中间数据的存储。硬盘控制器111a、硬盘控制器111b、……硬盘控制器111n分别实现固态硬盘110a、固态硬盘110b、……固态硬盘110n的FTL管理。每个固态硬盘具有一个FTL列表，每个FTL列表位于对应的固态硬盘中，每个FTL列表中维护对应固态硬盘中逻辑地址与物理地址的映射关系。同时，各固态硬盘的空闲物理存储空间的地址管理也分别在各硬盘控制器中实现。当FTL在各固态硬盘中实现时，阵列控制器100向固态硬盘阵列写入待写入数据的过程如下：待写入数据的分段及对应逻辑地址的分配仍由阵列控制器100实现；得到n个数据块和对应的n个固态硬盘中的n个逻辑地址后，控制器100为每个数据块生成一个包含对应的逻辑地址的数据写命令，并将n个数据写命令一一下发给对应的固态硬盘110a、固态硬盘110b、……固态硬盘110n；各固态硬盘接收对应的数据写命令后，各硬盘控制器根据数据写命令中的逻辑地址，将对应的数据块写入逻辑地址指向的虚拟存储空间，具体地，每个固态硬盘的硬盘控制器首先分别为对应数据块对应的逻辑地址分配一个该固态硬盘上的空闲物理地址，将数据库写入空闲物理地址指向的物理存储空间后，根据阵列控制器100的刷新命令，进行对应数据块的FTL刷新，即每个硬盘控制器分别在各自的FTL列表中建立数据块对应的逻辑地址与当前物理地址的映射关系，其中当前物理地址即各硬盘控制器为逻辑地址分配的空闲物理地址。
如前所述，FTL管理可以在阵列控制器100中实现，也可以在固态硬盘阵列中的各固态硬盘中实现。本实施例中，两种实现方式对应的待写入数据写入方法略有不同，下文依次阐述两种实现方式对应的待写入数据写入的流程。
FTL管理在阵列控制器100中实现时，阵列控制器100中维护有全局FTL列表，全局FTL列表记录固态硬盘阵列上保存的数据块的逻辑地址与物理地址的映射关系。如图3中所示，待写入数据写入的流程如下：
301，阵列控制器100将待写入数据分为n个数据块，并为每个数据块生成一个包含物理地址的数据写命令。
阵列控制器100将通过数据分段将待写入数据写入固态硬盘阵列。阵列控制器100在将待写入数据划分为n个数据块的同时，将n个数据块与n个固态硬盘一一对应。阵列控制器100为每个数据块分配一个固态硬盘，确定每个数据块对应的固态硬盘的逻辑地址，固态硬盘110a、固态硬盘110b、……固态硬盘110n分别对应逻辑地址Addr1、逻辑地址Addr2、……逻辑地址Addrn。全局FTL列表中存储有n个逻辑地址Addri，在待写入数据写入之前，n个逻辑地址Addri在全局FTL列表中可能有映射的初始物理地址，也可能没有映射的原物理地址。
同时，由于FTL管理的功能在阵列控制器100中实现，n个固态硬盘均根据阵列控制器100分配的当前物理地址addri，将数据块写入当前物理地址addri指向的物理存储空间。阵列控制器100为n个数据块分配逻辑地址Addri和当前物理地址addri后，还将n个逻辑地址与n个当前物理地址的映射关系缓存在非易失存储器中，用于断电后再次进行数据块的FTL刷新。同时，每个数据块对应的逻辑地址Addr、当前地 址addri及逻辑地址Addri与当前地址addri的映射关系皆为待写入数据写入过程中的中间数据，缓存在阵列控制器100中。
302，阵列控制器100将生成的n个数据写命令分别发给对应的n个固态硬盘。
阵列控制器100分别将每个数据写命令发送给对应的固态硬盘，每个数据写命令包含一个数据块对应的当前地址addri，即，包含物理地址addr1的数据写命令、包含物理地址addr2的数据写命令、……包含物理地址addrn的数据写命令分别发送给固态硬盘110a、固态硬盘110b、……固态硬盘110n。每个数据写命令触发对应的固态硬盘110将对应的数据块存储至对应的当前地址addri指向的空闲物理存储空间中，具体地，每个固态硬盘110i根据数据写命令获取对应的数据块，并在数据写命令中获取该数据块对应的当前地址addri，其中i是1-n中的任意值。将对应的数据存储至物理地址指向的空闲物理存储空间后，各固态硬盘向阵列控制器100返回写成功响应。
303，阵列控制器100接收各固态硬盘返回的写成功消息。
为实现数据写入和FTL列表刷新的分离，阵列控制器100需在确认所有固态硬盘完成对应数据块的存储之后，再进行FTL刷新。因此，阵列控制器100需接收各固态硬盘返回的写成功消息，在确认收到全部n个写成功消息之前，不进行写日志的存储和数据块FTL刷新的触发。
304，阵列控制器100确认收到n个写成功消息后，存储n个写日志。
阵列控制100将n个写日志存入阵列存储介质102，n个写日志与n个数据块一一对应，每个写日志记录对应的数据块的索引信息。重新上电后，根据索引信息，阵列控制器100可以找到未完成FTL刷新的数据块对应的逻辑地址Addri与当前物理地址addri的映射关系。
305，阵列控制器100在存储n个写日志后，触发数据块的FTL刷新。
确认存储n个写日志后，阵列控制器100触发数据块的FTL刷新，即，在全局FTL列表中建立逻辑地址Addri与当前物理地址addri的映射关系。具体地，阵列控制器100在阵列内存101中读取缓存的逻辑地址Addr1、逻辑地址Addr2、……逻辑地址Addrn以及n个逻辑地址Addr与n个物理地址addr的映射关系，并将n个逻辑地址Addri与n个物理地址addri的映射关系存入全局FTL列表。若全局FTL列表中具有逻辑地址Addri与原物理地址的原映射关系，则删除原映射关系。阵列控制器100在触发数据块的FTL刷新时，每完成一个数据块的FTL刷新，删除该完成FTL刷新的数据块对应的写日志。
当阵列管理器100和/或各固态硬盘在待写入数据写入时断电，重新上电后，阵列管理器100判断阵列存储介质102中存储的写日志的状态，根据阵列存储介质102中存储的写日志的不同状态，阵列管理器100执行不同的步骤，保证待各数据块写操作的原子性。若阵列存储介质102中存储的写日志为第一状态，说明断电时，阵列控制器100尚未开始FTL刷新和写日志的存储,阵列控制器100重写发起待写入数据的写操作；若阵列存储介质102中存储的写日志为第二状态，说明断电时，n个写日志已经生成，阵列控制器100根据阵列存储介质102中存储的写日志，再次触发阵列存储介质102中存储的写日志对应的数据块的FTL刷新，具体地，阵列控制器100从阵列存储介质102中存储的写日志获取存储的写日志的索引信息，根据索引信息和缓存在非 易失存储器中的逻辑地址Addri与当前物理地址addri的映射关系，在全局FTL列表中建立阵列存储介质102中存储的写日志对应的数据块的逻辑地址Addri与当前物理地址addri的映射关系。
FTL管理在各固态硬盘中实现时，如图2所示，硬盘控制器111a、硬盘控制器111b、……硬盘控制器111n分别实现固态硬盘110a、固态硬盘110b、……固态硬盘110n的FTL管理。每个固态硬盘具有一个FTL列表。如图4所示，待写入数据写入的流程如下：
401，阵列控制器200将待写入数据分为n个数据块，并为每个数据块生成一个包含逻辑地址Addri的数据写命令。
与步骤301中相同，阵列控制器200在将待写入数据划分为n个数据块的同时，将n个数据块与n个固态硬盘一一对应。阵列控制器200为每个数据块分配一个固态硬盘，确定每个数据块对应的固态硬盘的逻辑地址Addri，固态硬盘110a、固态硬盘110b、……固态硬盘110n分别对应逻辑地址Addr1、逻辑地址Addr2、……逻辑地址Addrn。
与步骤301中不同的是，由于各固态硬盘分别进行各自的FTL管理，因此为数据块分配与逻辑地址Addri对应的当前物理地址addri这一动作由各固态硬盘的硬盘控制器111分别执行。阵列控制器生成的n个数据写命令包含n个逻辑地址Addri，每个数据写命令对应一个数据块的逻辑地址Addri。
402，阵列控制器200将生成的n个数据写命令分别发给对应的n个固态硬盘。
由于此时数据写命令中包含的是逻辑地址Addri，对比步骤302中各固态硬盘存储对应的数据块的方法，步骤402中，在接收到对应的写命令并获取逻辑地址Addri后，各固态硬盘的硬盘控制器111首先利用FTL管理功能，为待写入的数据块分配当前物理地址addri。而后，各固态硬盘的硬盘控制器111分别将对应的数据块存储至当前物理地址addri指向的空闲物理存储空间。
此步骤中各固态硬盘存储对应的数据块的方法与步骤302中不同。
每个固态硬盘在接收到对应的写命令后，从写命令中获取逻辑地址Addri后，利用FTL管理功能，为待写入的数据块分配当前物理地址addri，并将对应的数据块存储在当前物理地址addri指向对应的固态硬盘中的空闲物理存储空间，其中i是1-n中的任意值。将对应的数据存储至当前物理地址addri指向的空闲物理存储空间后，各固态硬盘向阵列控制器200返回一个写成功消息。
每个固态硬盘为数据块分配当前物理地址addri后，还将逻辑地址Addri与当前物理地址addri的映射关系缓存在非易失存储器中，用于断电后数据块的FTL刷新。
403，阵列控制器200接收各固态硬盘返回的写成功消息。
步骤403与步骤303相同，在确认收到n个写成功消息之前，阵列控制器200不进行写日志的存储和数据块FTL刷新的触发。
404，阵列控制器200确认收到n个写成功消息后，存储n个写日志。
与步骤304中相同，每个写日志记录对应的数据块的索引信息。重新上电后，根据索引信息，阵列控制器200可以找到未完成FTL刷新的数据块对应的逻辑地址Addri 与当前物理地址addri的映射关系。
405，阵列控制器200确认存储n个写日志后，触发数据块的FTL刷新。
步骤405的具体实现方式与步骤403不同。由于各固态硬盘分别进行FTL管理，因此，触发各数据块的FTL刷新的具体方法包括：
阵列控制器200确认存储n个写日志后，向n个固态硬盘发送刷新命令。
各固态硬盘接收刷新命令，从各硬盘控制器的硬盘内存112中分别读取缓存在各固态硬盘上的逻辑地址Addri与当前物理地址addri的映射关系，在各固态硬盘的FTL列表中建立对应数据块的逻辑地址Addri与当前物理地址addri的映射关系，其中i是1-n中的任意值。
当一个固态硬盘完成对应数据块的FTL刷新后，该固态硬盘向阵列控制器200返回一个刷新成功消息，阵列控制器200删除完成FTL刷新的数据块对应的写日志。
当阵列管理器100和/或各固态硬盘在待写入数据写入时断电，重新上电后，阵列管理器100判断阵列存储介质102中存储的写日志的状态，根据阵列存储介质102中存储的写日志的不同状态，阵列管理器100执行不同的步骤，保证待各数据块写操作的原子性。若阵列存储介质102中存储的写日志为第一状态，说明断电时，阵列控制器100尚未开始FTL刷新和写日志的存储,阵列控制器100重写发起待写入数据的写操作；若阵列存储介质102中存储的写日志为第二状态，说明断电时，n个写日志已经生成，阵列控制器100根据阵列存储介质102中存储的写日志，再次触发阵列存储介质102中存储的写日志对应的数据块的FTL刷新，具体地，阵列控制器100从阵列存储介质102中存储的写日志获取存储的写日志的索引信息，向阵列存储介质102中存储的写日志对应的数据块所在的固态硬盘发送刷新命令；接到刷新命令的固态硬盘对对应的数据块进行FTL刷新，即根据缓存在非易失存储器中的逻辑地址Addri与当前物理地址addri的映射关系，在FTL列表中建立阵列存储介质102中存储的写日志对应的数据块的逻辑地址Addri与当前物理地址addri的映射关系。
对应地，本发明的实施例还提供一种固态硬盘阵列的控制器500。控制器500用于将分为n个数据块的待写入数据写入固态硬盘阵列中的n个固态硬盘，且保证n个数据块的写操作的原子性。如图5所示，控制器包括数据获取单元501、写命令生成单元502、写命令发送单元503、判断单元504和刷新单元505。数据接收单元501用于获取包含n个数据块的待写入数据。写命令生成单元502用于在数据接收单元501接收待写入数据后，生成n个数据块一一对应的n个写命令。写命令发送单元503用于将写命令生成单元生成的n个写命令分别发送到固态硬盘阵列中的n个固态硬盘，其中，n个写命令与n个固态硬盘一一对应，每个写命令用于请求对应的固态硬盘存储该写命令对应的数据块。判断单元504用于确定是否接收n个固态硬盘对应的n个写成功响应。如果判断单元504确定接收到n个写命令对应的n个写成功响应，刷新单元505用于，触发n个数据块的FTL刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立n个数据块中的每个数据块的逻辑地址Addri与该数据块的当前物理地址addri之间的映射关系。
同时，控制器500还包括日志存储单元506和非易失存储器507。在判断单元504 确定接收到n个写命令对应的n个写成功响应后，刷新单元505进行所述n个数据块的FTL刷新之前，日志存储单元506用于生成n个写日志，将n个写日志存入非易失存储器507，其中，n个写日志与n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
控制器500还包括删除单元508。刷新单元505完成一数据块的FTL刷新后，删除单元508用于删除完成FTL刷新的数据块对应的写日志。
当固态硬盘阵列中的每个固态硬盘维护本节点的FTL列表，则，用于n个数据块的FTL刷新的刷新单元505，具体用于：生成与n个数据块一一对应的n个刷新命令；将n个刷新命令分别发送到所述n个固态硬盘，其中，每个刷新命令用于请求对应的固态硬盘对对应的数据块进行的FTL刷新。固态硬盘阵列重新上电和/或控制器500重新上电之后，判断单元504还用于，确定非易失存储器507中的存储的写日志的状态。如果非易失存储器507中存储的写日志为第一状态，写命令发送单元503还用于，再次发起待写入数据的写操作。如果非易失存储器507中存储的写日志为第二状态，刷新单元505还用于，根据所述非易失存储器中存储的写日志，再次触发FTL刷新。判断单元504确定一个固态硬盘完成对应数据块的FTL刷新后，删除单元508还用于，删除完成了或者再次FTL刷新的数据块对应的写日志。其中，刷新单元505用于FTL再次刷新，具体包括：获取非易失存储器507中的每个写日志的索引信息，根据所述每个写日志的索引信息和n个固态硬盘的非易失存储器中存储有每个写日志对应的数据块的逻辑地址Addri与当前物理地址addri的映射关系，在各固态硬盘的FTL列表中建立每个写日志对应的数据块的逻辑地址Addri与当前物理地址addri之间的映射关系。
当控制器500中维护全局FTL列表，则，用于n个数据块的FTL刷新的刷新单元505，具体用于：根据n个数据块的每个数据块的逻辑地址，在全局FTL列表中建立n个数据块的每个数据块的逻辑地址Addri与n个数据块的每个数据块的当前物理地址addri之间的映射关系。固态硬盘阵列重新上电和/或控制器500重新上电之后，判断单元504还用于，确定非易失存储器507中的存储的写日志的状态。如果非易失存储器507中存储的写日志为第一状态，写命令发送单元503还用于，再次发起待写入数据的写操作。如果非易失存储器507中存储的写日志为第二状态，刷新单元505还用于，根据所述非易失存储器中存储的写日志，再次触发FTL刷新。判断单元504确定完成一个数据块的FTL刷新或者FTL再次刷新后，删除单元508还用于，删除完成了或者再次FTL刷新的数据块对应的写日志。其中，刷新单元505用于再次触发FTL刷新，具体包括：获取非易失存储器507中的每个写日志的索引信息，根据所述每个写日志的索引信息和n个固态硬盘的非易失存储器中存储有每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系。
控制器500中的各单元可以分别由对应的硬件芯片实现。在另一种实现中，一个或多个单元可以集成在一个硬件芯片上。在另一种实现中，控制器500中的各单元也可以由处理器执行存储器上的计算机指令实现。本发明实施例对此不作限定。
图6为依据本发明实施例的控制器600的结构示意图。
如图6所示，控制器600包括处理器601，处理器601与控制器内存602连接。处理器601可以为中央处理器(CPU)，图像处理器(Graphics Processing Unit，GPU),现场可编程门阵列(Field Programmable Gate Array，缩写：FPGA)，或数字信号处理器(英文：digital signal processor，DSP)等计算逻辑或以上任意计算逻辑的组合。处理器601可以为单核处理器或多核处理器。总线603用于在控制器600的各部件之间传递信息，总线603可以使用有线的连接方式或采用无线的连接方式，本申请并不对此进行限定。总线603还连接有通信接口604。通信接口604使用例如但不限于收发器一类的收发装置，来实现与外部设备605的连接，通信接口605可以通过有线或者无线的形式与网络互连。外部设备605包括固态硬盘阵列和用于缓存控制器600运行过程中的中间数据的非易失存储器。本发明实施例的方法可以由处理器601执行控制器内存602中的软件代码来完成/支持。
此外，图6仅仅是一个控制器600的例子，控制器600可能包含相比于图6展示的更多或者更少的组件，或者有不同的组件配置方式。同时，图6中展示的各种组件可以用硬件、软件或者硬件与软件的结合方式实施。
本发明的实施例还提供一种固态硬盘阵列。如图7所示，固态硬盘阵列700包括控制器710和n个固态硬盘。
控制器710用于：
获取包括n个数据块的待写入数据；
生成n个写命令，其中，n个写命令与n个数据块一一对应；
将n个写命令分别发送到所述n个固态硬盘，其中，n个写命令与n个固态硬盘一一对应。
n个固态硬盘中的每个固态硬盘用于：
根据对应的写命令，存储写命令对应的数据块；
存储写命令对应的数据块后，向控制器710发送写成功响应。如果确定接收到所述n个写命令对应的n个写成功响应，控制器710进行所述n个数据块的FTL刷新。
控制器710还用于：
如果确定接收到n个写命令对应的n个写成功响应，在进行n个数据块的FTL刷新之前，生成n个写日志，将n个写日志存入所述控制器的非易失存储器，其中，n个写日志与所述n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
当固态硬盘阵列中的每个固态硬盘还用于维护本节点的FTL列表时，控制器710用于进行n个数据块的FTL刷新，具体包括：
生成n个刷新命令，其中，所述n个刷新命令与所述n个数据块一一对应；
将n个刷新命令分别发送到n个固态硬盘。
固态硬盘阵列中的每个固态硬盘还用于：接收所述控制器发送的刷新命令，进行存储的数据块的FTL刷新，即在本节点的FTL列表中建立每个数据块的逻辑地址与数据块的当前物理地址之间的映射关系；在存储的数据块的FTL刷新成功后，向控制器710发送一个刷新成功消息。
控制器710还用于：接收n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，删除发送刷新成功消息的固态硬盘对应的数据块的写日志。
当控制器710还用于维护全局FTL列表时，所述控制器用于进行所述n个数据块的FTL刷新，具体包括：根据n个数据块的每个数据块的逻辑地址，在全局FTL列表中建立n个数据块的每个数据块的逻辑地址与n个数据块的每个数据块的物理地址之间的映射关系；完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
控制器710还用于，固态硬盘阵列和/或所述控制器重新上电之后，确定控制器710的非易失存储器中的存储的写日志的状态，如果控制器710的非易失存储器中存储的写日志为第一状态，再次发起所述待写入数据的写操作；所述固态硬盘阵列和/或所述控制器重新上电之后，确定控制器710的非易失存储器中的存储的写日志的状态，如果控制器710的非易失存储器中存储的写日志为第二状态，根据控制器710的非易失存储器中存储的写日志，进行FTL再次刷新；确定完成一个数据块的FTL再次刷新后，删除完成了再次FTL刷新的数据块对应的写日志。FTL再次刷新具体包括：获取控制器710非易失存储器中的每个写日志的索引信息，根据所述每个写日志的索引信息，获取控制器710或n个固态硬盘的非易失存储器中存储的每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系。
以上所述，仅为本发明的具体实施方式，但本发明的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本发明揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本发明的保护范围之内。因此，本发明的保护范围应以所述权利要求的保护范围为准。

Claims (31)

  1. 一种数据写入的方法，其特征在于，所述方法应用于固态硬盘阵列，所述方法包括：
    所述固态硬盘阵列的控制器获取待写入数据，所述待写入数据包括n个数据块；
    所述控制器生成n个写命令，其中，所述n个写命令与所述n个数据块一一对应；
    所述控制器将所述n个写命令分别发送到所述固态硬盘阵列中的n个固态硬盘，其中，所述n个写命令与所述n个固态硬盘一一对应，每个写命令用于请求对应的固态硬盘存储所述写命令对应的数据块；
    所述控制器确定是否接收到所述n个写命令对应的n个写成功响应；
    如果确定接收到所述n个写命令对应的n个写成功响应，所述控制器触发所述n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立所述每个数据块的逻辑地址与所述数据块的当前物理地址之间的映射关系。
  2. 如权利要求1所述的方法，其特征在于，如果确定接收到所述n个写命令对应的n个写成功响应，在所述控制器触发所述n个数据块的FTL刷新之前，所述方法还包括：
    所述控制器生成n个写日志，将所述n个写日志存入非易失存储器，所述n个写日志与所述n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
  3. 如权利要求2所述的方法，其特征在于，所述固态硬盘阵列中的每个固态硬盘维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录所述每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述控制器触发所述n个数据块的FTL刷新，具体包括：
    所述控制器生成n个刷新命令，其中，所述n个刷新命令与所述n个数据块一一对应；
    所述控制器将所述n个刷新命令分别发送到所述n个固态硬盘，其中，每个刷新命令用于请求对应的固态硬盘对对应的数据块进行的FTL刷新。
  4. 如权利要求3所述的方法，其特征在于，所述方法还包括：
    所述控制器接收所述n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，所述控制器删除发送所述刷新成功消息的固态硬盘对应的数据块的写日志。
  5. 如权利要求2所述的方法，其特征在于，所述控制器维护全局FTL列表，所述全局FTL列表用于记录所述固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述控制器触发所述n个数据块的FTL刷新包括：
    所述控制器根据缓存的所述n个数据块的每个数据块存储的逻辑地址与当前物理地址的映射关系，在所述全局FTL列表中建立所述n个数据块的每个数据块的逻辑地址与所述n个数据块的每个数据块的当前物理地址之间的映射关系。
  6. 如权利要求5所述的方法，其特征在于，
    所述控制器完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日 志。
  7. 如权利要求2-6任一项所述的方法，其特征在于，所述方法还包括：
    所述固态硬盘阵列和/或所述控制器重新上电之后，所述控制器确定所述非易失存储器中的存储的写日志的状态，如果所述控制器的非易失存储器中存储的写日志为第一状态，所述控制器再次发起所述待写入数据写的写操作。
  8. 如权利要求2-7任一所述的方法，其特征在于，所述方法还包括：
    所述固态硬盘阵列和/或所述控制器重新上电之后，所述控制器确定所述非易失存储器中的存储的写日志的状态，如果所述控制器的非易失存储器中存储的写日志为第二状态，根据所述控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
  9. 如权利要求8所述的方法，其特征在于，所述控制器根据所述控制器的非易失存储器中的写日志，再次触发FTL刷新包括：
    所述控制器获取所述非易失存储器中的每个写日志的索引信息，根据所述每个写日志的索引信息，获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系；或者，
    所述控制器获取所述非易失存储器中的每个写日志的索引信息，将获取的所述每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系。
  10. 如权利要求9所述的方法，其特征在于，所述方法还包括：
    所述控制器确定是否完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
  11. 一种固态硬盘阵列的控制器，其特征在于，所述控制器包括数据获取单元、写命令生成单元、写命令发送单元、判断单元和刷新单元；
    所述数据接收单元用于获取待写入数据，所述待写入数据包括n个数据块；
    所述写命令生成单元用于生成n个写命令，其中，所述n个写命令与所述n个数据块一一对应；
    所述写命令发送单元用于将所述n个写命令分别发送到所述固态硬盘阵列中的n个固态硬盘，其中，所述n个写命令与所述n个固态硬盘一一对应，每个写命令用于请求对应的固态硬盘存储所述写命令对应的数据块；
    所述判断单元用于确定是否接收所述n个固态硬盘对应的n个写成功响应；
    如果确定接收到所述n个写命令对应的n个写成功响应，所述刷新单元用于，进行所述n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立所述每个数据块的逻辑地址与所述数据块的当前物理地址之间的映射关系。
  12. 如权利要求11所述的控制器，其特征在于，所述控制器还包含日志存储单元，
    如果所述判断单元确定接收到所述n个写命令对应的n个写成功响应，在所述刷新 单元进行所述n个数据块的FTL刷新之前，所述日志存储单元用于，
    生成n个写日志，保存所述n个写日志，所述n个写日志与所述n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
  13. 如权利要求12所述控制器，其特征在于，所述固态硬盘阵列中的每个固态硬盘维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录所述每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述刷新单元用于触发所述n个数据块的FTL刷新，具体包括：
    生成n个刷新命令，其中，所述n个刷新命令与所述n个数据块一一对应；
    将所述n个刷新命令分别发送到所述n个固态硬盘，其中，每个刷新命令用于请求对应的固态硬盘对对应的数据块进行的FTL刷新。
  14. 如权利要求13所述的控制器，其特征在于，所述控制器还包括删除单元，所述删除单元用于：
    接收所述n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，删除发送所述刷新成功消息的固态硬盘对应的数据块的写日志。
  15. 如权利要求12所述的控制器，其特征在于，所述控制器还包括维护单元，所述维护单元用于维护全局FTL列表，所述全局FTL列表用于记录所述固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述刷新单元用于触发所述n个数据块的FTL刷新，具体包括：
    所述刷新单元用于根据所述n个数据块的每个数据块的逻辑地址，通知所述维护单元在所述全局FTL列表中建立所述n个数据块的每个数据块的逻辑地址与所述n个数据块的每个数据块的物理地址之间的映射关系。
  16. 如权利要求15所述的控制器，其特征在于，所述控制器还包括删除单元，所述删除单元用于：
    所述刷新单元完成一个数据块的FTL刷新，删除完成FTL刷新的数据块的写日志。
  17. 如权利要求12-16任一所述的控制器，其特征在于，
    所述固态硬盘阵列重新上电和/或所述控制器重新上电之后，所述判断单元还用于，确定所述控制器的非易失存储器中的存储的写日志的状态；
    如果所述控制器的非易失存储器中存储的写日志为第一状态，所述写命令发送单元还用于再次发起所述待写入数据的写操作。
  18. 如权利要求12-17任一所述的控制器，其特征在于，
    所述固态硬盘阵列和/或所述控制器重新上电之后，所述判断单元还用于，确定所述控制器的非易失存储器中的存储的写日志的状态；
    如果所述控制器的非易失存储器中存储的写日志为第二状态，所述刷新单元还用于，根据所述控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
  19. 如权利要求14所述的控制器，其特征在于，所述刷新单元用于根据所述非易失存储器中的写日志，再次触发FTL刷新具体包括：
    获取所述非易失存储器中的每个写日志的索引信息，根据所述每个写日志的索引信息，获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址 之间的映射关系；或者，
    获取所述非易失存储器中的每个写日志的索引信息，将获取的所述每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系。
  20. 如权利要求17所述的控制器，其特征在于，所述判断单元还用于确定是否完成一个数据块的FTL刷新；所述删除单元还用于所述判断单元确认完成一个数据块的FTL刷新后，删除完成了FTL刷新的数据块对应的写日志。
  21. 一种固态硬盘阵列的控制器，其特征在于，所述控制器包括处理器和存储器，所述处理器用于运行所述存储器中存储的计算机指令来实现权利要求1-10中所述的方法。
  22. 一种固态硬盘阵列，其特征在于，所述固态硬盘阵列包含控制器和n个固态硬盘；
    所述控制器用于：
    获取待写入数据，所述待写入数据包括n个数据块；
    生成n个写命令，其中，所述n个写命令与所述n个数据块一一对应；
    将所述n个写命令分别发送到所述n个固态硬盘，其中，所述n个写命令与所述n个固态硬盘一一对应；
    所述n个固态硬盘中的每个固态硬盘用于：
    根据对应的写命令，存储写命令对应的数据块；
    存储写命令对应的数据块后，向所述控制器发送写成功响应；
    所述控制器还用于：
    确定是否接收到所述n个写命令对应的n个写成功响应；
    如果确定接收到所述n个写命令对应的n个写成功响应，触发所述n个数据块的闪存转换层(Flash Translation Layer，FTL)刷新，其中，每个数据块的FTL刷新包括：在FTL列表中建立所述每个数据块的逻辑地址与所述数据块的当前物理地址之间的映射关系。
  23. 如权利要求22所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    如果确定接收到所述n个写命令对应的n个写成功响应，在触发所述n个数据块的FTL刷新之前，生成n个写日志，将所述n个写日志存入所述控制器的非易失存储器，所述n个写日志与所述n个数据块一一对应，每个写日志记录对应的数据块的索引信息。
  24. 如权利要求23所述的固态硬盘阵列，其特征在于，
    所述固态硬盘阵列中的每个固态硬盘还用于：
    维护本节点的FTL列表，每个固态硬盘的本节点的FTL列表用于记录所述每个固态硬盘保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述控制器用于触发所述n个数据块的FTL刷新，具体包括：
    生成n个刷新命令，其中，所述n个刷新命令与所述n个数据块一一对应；
    将所述n个刷新命令分别发送到所述n个固态硬盘；
    所述固态硬盘阵列中的每个固态硬盘还用于：
    接收所述控制器发送的刷新命令，在本节点的FTL列表中建立所述每个数据块的逻辑地址与所述数据块的当前物理地址之间的映射关系。
  25. 如权利要求24所述的固态硬盘阵列，其特征在于，
    每个固态硬盘还用于：在完成存储的数据块的FTL刷新后，向所述控制器发送一个刷新成功消息；
    所述控制器还用于：接收所述n个固态硬盘中的任一个固态硬盘发送的刷新成功消息，删除所述发送刷新成功消息的固态硬盘对应的数据块的写日志。
  26. 如权利要求23所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    维护全局FTL列表，所述全局FTL列表用于记录所述固态硬盘阵列上保存的数据块的逻辑地址与物理地址之间的映射关系；
    所述控制器用于触发所述n个数据块的FTL刷新，具体包括：
    根据缓存的所述n个数据块的每个数据块存储的逻辑地址与当前物理地址的映射关系，在所述全局FTL列表中建立所述n个数据块的每个数据块的逻辑地址与所述n个数据块的每个数据块的当前物理地址之间的映射关系。
  27. 如权利要求26所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    完成一个数据块的FTL刷新，删除完成了FTL刷新的数据块对应的写日志。
  28. 如权利要求23-27中任一所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    所述固态硬盘阵列和/或所述控制器重新上电之后，确定所述控制器的非易失存储器中的存储的写日志的状态，如果所述控制器的非易失存储器中存储的写日志为第一状态，再次发起所述待写入数据的写操作。
  29. 如权利要求23-28中任一所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    所述固态硬盘阵列和/或所述控制器重新上电之后，确定所述控制器的非易失存储器中的存储的写日志的状态，如果所述控制器的非易失存储器中存储的写日志为第二状态，根据所述控制器的非易失存储器中存储的写日志，再次触发FTL刷新。
  30. 如权利要求29所述的固态硬盘阵列，其特征在于，所述控制器用于根据所述控制器的非易失存储器中的写日志，再次触发FTL刷新，具体包括：
    获取所述非易失存储器中的每个写日志的索引信息，根据所述每个写日志的索引信息，获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在全局FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系；或者，
    获取所述非易失存储器中的每个写日志的索引信息，将获取的所述每个写日志的索引信息发送到对应的固态硬盘，以使得对应的固态硬盘根据获取到的索引信息获取缓存的所述每个写日志对应的数据块的逻辑地址与当前物理地址的映射关系，在本节点的FTL列表中建立所述每个写日志对应的数据块的逻辑地址与所述当前物理地址之间的映射关系。
  31. 如权利要求30所述的固态硬盘阵列，其特征在于，所述控制器还用于：
    确定完成一个数据块的FTL再次刷新后，删除完成了再次FTL刷新的数据块对应的写日志。
PCT/CN2017/119390 2017-12-28 2017-12-28 一种数据写入的方法及固态硬盘阵列 WO2019127212A1 (zh)

Priority Applications (4)

Application Number Priority Date Filing Date Title
PCT/CN2017/119390 WO2019127212A1 (zh) 2017-12-28 2017-12-28 一种数据写入的方法及固态硬盘阵列
EP17936484.9A EP3726364B1 (en) 2017-12-28 2017-12-28 Data write-in method and solid-state drive array
CN201780040471.7A CN110462577B (zh) 2017-12-28 2017-12-28 一种数据写入的方法及固态硬盘阵列
US16/914,375 US11243701B2 (en) 2017-12-28 2020-06-28 Data write method and solid-state drive array

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/119390 WO2019127212A1 (zh) 2017-12-28 2017-12-28 一种数据写入的方法及固态硬盘阵列

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/914,375 Continuation US11243701B2 (en) 2017-12-28 2020-06-28 Data write method and solid-state drive array

Publications (1)

Publication Number Publication Date
WO2019127212A1 true WO2019127212A1 (zh) 2019-07-04

Family

ID=67062848

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/119390 WO2019127212A1 (zh) 2017-12-28 2017-12-28 一种数据写入的方法及固态硬盘阵列

Country Status (4)

Country Link
US (1) US11243701B2 (zh)
EP (1) EP3726364B1 (zh)
CN (1) CN110462577B (zh)
WO (1) WO2019127212A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11561871B2 (en) * 2019-12-18 2023-01-24 GRAID Technology Inc. Data transmission and protection system and method thereof

Families Citing this family (36)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019222958A1 (en) 2018-05-24 2019-11-28 Alibaba Group Holding Limited System and method for flash storage management using multiple open page stripes
WO2020000136A1 (en) 2018-06-25 2020-01-02 Alibaba Group Holding Limited System and method for managing resources of a storage device and quantifying the cost of i/o requests
US11327929B2 (en) 2018-09-17 2022-05-10 Alibaba Group Holding Limited Method and system for reduced data movement compression using in-storage computing and a customized file system
US11061735B2 (en) 2019-01-02 2021-07-13 Alibaba Group Holding Limited System and method for offloading computation to storage nodes in distributed system
US10860223B1 (en) 2019-07-18 2020-12-08 Alibaba Group Holding Limited Method and system for enhancing a distributed storage system by decoupling computation and network tasks
US11617282B2 (en) 2019-10-01 2023-03-28 Alibaba Group Holding Limited System and method for reshaping power budget of cabinet to facilitate improved deployment density of servers
US11126561B2 (en) * 2019-10-01 2021-09-21 Alibaba Group Holding Limited Method and system for organizing NAND blocks and placing data to facilitate high-throughput for random writes in a solid state drive
US11449455B2 (en) 2020-01-15 2022-09-20 Alibaba Group Holding Limited Method and system for facilitating a high-capacity object storage system with configuration agility and mixed deployment flexibility
US11379447B2 (en) 2020-02-06 2022-07-05 Alibaba Group Holding Limited Method and system for enhancing IOPS of a hard disk drive system based on storing metadata in host volatile memory and data in non-volatile memory using a shared controller
US11449386B2 (en) 2020-03-20 2022-09-20 Alibaba Group Holding Limited Method and system for optimizing persistent memory on data retention, endurance, and performance for host memory
US11385833B2 (en) 2020-04-20 2022-07-12 Alibaba Group Holding Limited Method and system for facilitating a light-weight garbage collection with a reduced utilization of resources
US11301173B2 (en) 2020-04-20 2022-04-12 Alibaba Group Holding Limited Method and system for facilitating evaluation of data access frequency and allocation of storage device resources
US11281575B2 (en) 2020-05-11 2022-03-22 Alibaba Group Holding Limited Method and system for facilitating data placement and control of physical addresses with multi-queue I/O blocks
US11461262B2 (en) 2020-05-13 2022-10-04 Alibaba Group Holding Limited Method and system for facilitating a converged computation and storage node in a distributed storage system
US11494115B2 (en) 2020-05-13 2022-11-08 Alibaba Group Holding Limited System method for facilitating memory media as file storage device based on real-time hashing by performing integrity check with a cyclical redundancy check (CRC)
US11507499B2 (en) 2020-05-19 2022-11-22 Alibaba Group Holding Limited System and method for facilitating mitigation of read/write amplification in data compression
US11556277B2 (en) 2020-05-19 2023-01-17 Alibaba Group Holding Limited System and method for facilitating improved performance in ordering key-value storage with input/output stack simplification
US11263132B2 (en) 2020-06-11 2022-03-01 Alibaba Group Holding Limited Method and system for facilitating log-structure data organization
US11422931B2 (en) 2020-06-17 2022-08-23 Alibaba Group Holding Limited Method and system for facilitating a physically isolated storage unit for multi-tenancy virtualization
US11354200B2 (en) 2020-06-17 2022-06-07 Alibaba Group Holding Limited Method and system for facilitating data recovery and version rollback in a storage device
US11354233B2 (en) 2020-07-27 2022-06-07 Alibaba Group Holding Limited Method and system for facilitating fast crash recovery in a storage device
US11372774B2 (en) 2020-08-24 2022-06-28 Alibaba Group Holding Limited Method and system for a solid state drive with on-chip memory integration
US11487465B2 (en) 2020-12-11 2022-11-01 Alibaba Group Holding Limited Method and system for a local storage engine collaborating with a solid state drive controller
US11734115B2 (en) 2020-12-28 2023-08-22 Alibaba Group Holding Limited Method and system for facilitating write latency reduction in a queue depth of one scenario
US11416365B2 (en) 2020-12-30 2022-08-16 Alibaba Group Holding Limited Method and system for open NAND block detection and correction in an open-channel SSD
US11726699B2 (en) 2021-03-30 2023-08-15 Alibaba Singapore Holding Private Limited Method and system for facilitating multi-stream sequential read performance improvement with reduced read amplification
US11461173B1 (en) 2021-04-21 2022-10-04 Alibaba Singapore Holding Private Limited Method and system for facilitating efficient data compression based on error correction code and reorganization of data placement
US11476874B1 (en) 2021-05-14 2022-10-18 Alibaba Singapore Holding Private Limited Method and system for facilitating a storage server with hybrid memory for journaling and data storage
CN113608701B (zh) * 2021-08-18 2024-09-20 合肥大唐存储科技有限公司 一种存储系统中数据管理方法和固态硬盘
JP7281515B2 (ja) * 2021-09-10 2023-05-25 株式会社日立製作所 ストレージシステム、ストレージ管理方法、ストレージ管理プログラム
CN113838487B (zh) * 2021-09-28 2023-04-25 北京信息科技大学 一种多读写头的盘状存储器、读写方法及数据保护方法
CN115878020A (zh) * 2021-09-29 2023-03-31 慧荣科技股份有限公司 编码历程信息的存取方法和计算机可读取存储介质和装置
US11860775B2 (en) 2021-09-29 2024-01-02 Silicon Motion, Inc. Method and apparatus for programming data into flash memory incorporating with dedicated acceleration hardware
US11972150B2 (en) 2021-09-29 2024-04-30 Silicon Motion, Inc. Method and non-transitory computer-readable storage medium and apparatus for programming data into flash memory through dedicated acceleration hardware
KR20240101154A (ko) * 2022-12-23 2024-07-02 삼성전자주식회사 타임아웃 방지를 위한 전자 장치 및 동작 방법
CN116627359B (zh) * 2023-07-24 2023-11-14 成都佰维存储科技有限公司 内存管理方法、装置、可读存储介质及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104794070A (zh) * 2015-04-23 2015-07-22 南京道熵信息技术有限公司 基于动态非覆盖raid技术的固态闪存写缓存系统及方法
CN105528180A (zh) * 2015-12-03 2016-04-27 浙江宇视科技有限公司 一种数据存储方法、装置及设备
CN106681848A (zh) * 2016-12-13 2017-05-17 中国科学院计算技术研究所 一种纠删码raid的数据一致性保障方法及系统
CN107111453A (zh) * 2014-12-10 2017-08-29 英特尔公司 跨多个存储装置执行原子写入操作

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6895490B1 (en) * 2001-04-09 2005-05-17 Matrix Semiconductor, Inc. Method for making a write-once memory device read compatible with a write-many file system
CN102073595A (zh) * 2011-01-24 2011-05-25 华亚微电子（上海）有限公司 防断电的损耗平衡存储方法
CN103514095B (zh) * 2012-06-18 2016-08-03 记忆科技(深圳)有限公司 一种数据库写入ssd 的方法和系统
US20150160999A1 (en) * 2012-07-13 2015-06-11 Samsung Electronics Co., Ltd. Solid state drive controller, solid state drive, data processing method of solid state drive, multi-channel solid state drive, raid controller and computer-readable recording medium having recorded therein computer program for providing sequence information to solid state drive
CN104407813B (zh) * 2014-11-20 2019-02-19 上海宝存信息科技有限公司 一种基于固态存储介质的raid系统及方法
CN106155915B (zh) * 2015-04-16 2021-01-08 中兴通讯股份有限公司 数据存储的处理方法及装置
US9652164B2 (en) * 2015-05-14 2017-05-16 Toshiba Corporation Solid-state mass storage device and method for processing forced unit access write commands
CN106354615B (zh) * 2015-07-21 2021-06-01 北京忆恒创源科技有限公司 固态硬盘日志生成方法及其装置
US11256431B1 (en) * 2017-01-13 2022-02-22 Lightbits Labs Ltd. Storage system having a field programmable gate array
CN107391391B (zh) 2017-07-19 2019-05-14 深圳大普微电子科技有限公司 在固态硬盘的ftl实现数据拷贝的方法、系统及固态硬盘

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107111453A (zh) * 2014-12-10 2017-08-29 英特尔公司 跨多个存储装置执行原子写入操作
CN104794070A (zh) * 2015-04-23 2015-07-22 南京道熵信息技术有限公司 基于动态非覆盖raid技术的固态闪存写缓存系统及方法
CN105528180A (zh) * 2015-12-03 2016-04-27 浙江宇视科技有限公司 一种数据存储方法、装置及设备
CN106681848A (zh) * 2016-12-13 2017-05-17 中国科学院计算技术研究所 一种纠删码raid的数据一致性保障方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3726364A4 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11561871B2 (en) * 2019-12-18 2023-01-24 GRAID Technology Inc. Data transmission and protection system and method thereof

Also Published As

Publication number Publication date
US20200326855A1 (en) 2020-10-15
CN110462577A (zh) 2019-11-15
EP3726364A4 (en) 2020-12-23
EP3726364B1 (en) 2022-07-27
EP3726364A1 (en) 2020-10-21
CN110462577B (zh) 2022-03-29
US11243701B2 (en) 2022-02-08

Similar Documents

Publication Publication Date Title
WO2019127212A1 (zh) 一种数据写入的方法及固态硬盘阵列
JP6224253B2 (ja) フラッシュメモリ内に記憶されたデータの推測的プリフェッチ
JP6355650B2 (ja) 不揮発性記憶デバイスのためのメモリーのリアドレシング
KR101993704B1 (ko) 플래시 메모리를 기반으로 하는 저장 장치 및 플래시 메모리를 제어하는 메모리 컨트롤러의 쓰기 메모리 블록 할당 방법
CN106354615B (zh) 固态硬盘日志生成方法及其装置
JP2019082813A (ja) メモリシステムおよび制御方法
KR20190004400A (ko) 메모리 컨트롤러의 동작 방법 및 사용자 장치의 동작 방법
JP2019082817A (ja) 計算機システムおよび制御方法
WO2018189858A1 (ja) ストレージシステム
US11928053B2 (en) System garbage collection method and method for garbage collection in solid state disk
KR101910759B1 (ko) 트랜잭션 처리 방법, 장치 및 컴퓨터 시스템
TW201939288A (zh) 資料移動方法及儲存控制器
US11029873B2 (en) Storage device with expandable logical address space and operating method thereof
US12014090B2 (en) Memory system and method of controlling nonvolatile memory and for reducing a buffer size
JP6734768B2 (ja) 二重書込みを遂行するストレージ装置を含むシステム、装置、及びその方法
US10754785B2 (en) Checkpointing for DRAM-less SSD
WO2019090493A1 (zh) 内存块回收方法和装置
Missimer et al. Partitioned real-time NAND flash storage
CN109815157B (zh) 编程命令处理方法与装置
CN113986773A (zh) 基于固态硬盘的写放大优化方法、装置及计算机设备
CN110865945B (zh) 存储设备的扩展地址空间
US20160018995A1 (en) Raid system for processing i/o requests utilizing xor commands
WO2022021280A1 (zh) 存储控制器、存储控制方法、固态硬盘及存储系统
CN114625318A (zh) 应用于固态硬盘的数据写入方法、装置、设备
TWI715116B (zh) 資料儲存裝置及其刪除命名空間之方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17936484

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2017936484

Country of ref document: EP

Effective date: 20200713