CN104281517A - Log mode based memory space management method and device - Google Patents
Log mode based memory space management method and device Download PDFInfo
- Publication number
- CN104281517A CN104281517A CN201410548611.XA CN201410548611A CN104281517A CN 104281517 A CN104281517 A CN 104281517A CN 201410548611 A CN201410548611 A CN 201410548611A CN 104281517 A CN104281517 A CN 104281517A
- Authority
- CN
- China
- Prior art keywords
- daily record
- application
- storage space
- merging
- release
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a log mode based memory space management method and device. The method applied to memory equipment includes: when applying for memory space or releasing the memory space, recording the corresponding application or release log in a second map region (SMR) of preset size, wherein the application or release log is used for recording an initial position and size information of the applied or released memory space; according to certain strategy, sequentially executing memory space combination on the application or release log and a previous combined log, and recording the combined log generated at the time in the SMR, wherein the combined log records the initial position and size of spare memory space. By the method and device, the problem that equipment performance is reduced due to release of plenty of random memory space during existing memory space management is solved, and the requirement on memory capacity of the memory equipment is low.
Description
Technical field
The present invention relates to technical field of memory, particularly relate to a kind of memory space management based on log mode and device.
Background technology
No matter be based on file system or the application based on other block device, have the requirement that two basic: 1, record data and be stored on which block of block device; 2, which block of block device is idle condition.
In principle, can be idle without which block of record block equipment, and only record data and be stored on which block.Because the state of block only has two kinds: using state or idle condition.So it is idle for can being extrapolated which block by the block used.But this means again needs to travel through all blocks used.In order to make the efficiency of application free block higher, which block most block device all can record is idle.
Current record block is idle condition or using state has mainly contained two kinds of methods, and one is the mode of bitmap, and one is the mode of b-tree.Certainly, the mode of the most frequently used still bitmap.Bitmap is that to state the N number of piece of block device with N number of bit of one section of continuous space be idle or occupied.Comparatively speaking, the resource consumption of bitmap is smaller, and for the block size of 4KB, its resource consumption is only 1/ (4096*8)=0.003% (its 8 represent a byte 8 bits).For the file system of 1GB size, the space size required for bitmap is 32KB; And for the file system of 1TB size, then the space size needed is 2MB.This size is for the internal memory of system, or acceptable, but from the angle of search efficiency, relatively just lower.And if for the file system of 1PB, bitmap need size then become 32GB, this size for most of equipment internal memory all cannot accept.This just mean need at every turn obtain bulk state time, all need from disk read, this will have a strong impact on the performance of file system.
For the problems referred to above, the method that ratio is easier to expect is: bitmap is divided into multiple little bitmap, then maps the little bitmap after division, shown in please refer to the drawing 1 with bitmap.So just can find out which little bitmap by the bitmap of most top layer is available free piece; If little bitmap is not in internal memory, then little bitmap is read out from disk.
But, so still can there are the following problems: no matter be application or release block resource, bitmap needs to upgrade.For the operation of application block resource, can control the renewal of bitmap, but the operation of release block resource is then difficult to control, trigger because releasing operation is upper layer application.Frequently carry out the release of block resource, just mean frequent updating bitmap, just mean there is a lot of read-write IO.More very, if the renewal of bitmap relates to much little bitmap, when namely the renewal of bitmap has very large randomness, system will can't bear the heavy load.
B-tree adopts the mode of extent (one section of continuous print storage space) to manage continuous print block (block).The method that the mode of Extent adopts skew to add length manages idle block space.But there is above-mentioned same problem in b-tree and bitmap.
Summary of the invention
In view of this, the invention provides a kind of memory space management based on log mode.
The method is applied to memory device, comprise: when application storage space or release storage space, corresponding application or release daily record is recorded, the reference position of this application or release log recording application or release storage space and size information in the second mapping area SMR of default size; According to certain strategy, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges, and in described SMR, record this merging daily record produced, the reference position of the storage space that described merging log recording is idle and size.
Preferably, the record aimed at day in SMR adopts the mode of circulation covering to carry out; Described according to certain strategy, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges and comprises: when the last time merge ratio that the merging daily record and current not merged application that produce or release daily record total quantity account for the total saturation of SMR content area daily record reach predetermined threshold value time, order is to described application or release daily record, and the merging daily record of last time performs the operation of storage space merging; Or at application storage space and when not meeting the storage space of size needed for this application, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges.
Preferably, order is to described application or release daily record, and the operation that the merging daily record execution storage space of last time merges comprises: order is for the not merged application of current every bar or release daily record, according to the reference position recorded in daily record and size, judge whether to merge with current node in RBTree, if can, merge; After merging by newly-generated node city in RBTree, and delete merged node; The content of this newly-generated node is: this merges reference position and the size of rear idle storage space.
Preferably, the daily record of recording in described SMR comprises: expression is the information of putspace operations or application spatial operation; Expression is the information of truly applying for the daily record of releasing operation or the daily record of union operation; The numbering Cur_journal_id of daily record, this numbering increases successively along with the increase of daily record; Described order is to described application or release daily record, and being operating as of merging daily record execution storage space merging of last time numbers according to daily record the operation that the order increased progressively performs described merging successively.
Preferably, disk storage space is divided into several regions region, and each region comprises data area DR and mapping area MR; Wherein MR comprises main mapping area PMR and the second mapping area SMR; PMR comprises head zone and have recorded the position of each SMR and the content area of next PMR position that belong to this PMR management; SMR comprises head zone and records the content area of described daily record.
Preferably, after memory device startup work, PMR information, SMR information will be read in internal memory from disk, and carry out the management of storage space based on the PMR information in internal memory and SMR information.
Preferably, if reading the maximum daily record of numbering in the SMR content area in internal memory from disk is merge daily record, when needs carry out union operation further, need again the last time merging daily record before described merging daily record, the application date book after last merging daily record, release daily record to be re-started a union operation.
Preferably, when applying for storage space, if there is the space available continuously that several meet application condition in the DR of SMR management, then select the described free space that LBA (Logical Block Addressing) lba is little.
Compared to prior art, technical scheme of the present invention solves a large amount of random storage space in existing storage space management and discharges the problem of the equipment performance reduction caused, and less demanding to the memory size of memory device.
Accompanying drawing explanation
Fig. 1 is that a kind of multilayer bitmap maps schematic diagram.
Fig. 2 is embodiment of the present invention process flow diagram.
Fig. 3 is the partition structure figure of embodiment of the present invention storage space.
Fig. 4 is embodiment of the present invention log content figure.
Fig. 5 is that embodiment of the present invention daily record merges exemplary plot.
Fig. 6 (a) ~ Fig. 6 (b) is that another daily record of the embodiment of the present invention merges exemplary plot.
Embodiment
For the technical matters mentioned in background technology, the present invention mainly provides a kind of technical scheme of carrying out storage space management based on log mode.This technical scheme solves a large amount of random storage space in existing storage space management and discharges the problem of the equipment performance reduction caused.Describe in detail below by way of specific embodiment.
Embodiment of the present invention process flow diagram shown in please refer to the drawing 2.
S21, application storage space or release storage space in, corresponding application or release daily record is recorded, the reference position of this application or release log recording application or release storage space and size information in the second mapping area SMR of default size.
S22, according to certain strategy, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges, and in described SMR, record this merging daily record produced, the reference position of the storage space that described merging log recording is idle and size.
Said method, when carrying out storage space application or release, is only the application or releasing operation daily record that record is corresponding; After meeting some requirements, (essence of union operation is: count which space in storage space and be released to carry out union operation to above-mentioned application or release daily record again, by continuous d/d space, i.e. extent, merge, and using this union operation result as merge daily record be appended in SMR region record as a new daily record).Doing so avoids the management carrying out storage space frequently, improve the performance of memory device.Separately, the reference position of this merging log recording idle storage space and size, the application operation of new storage space can judge whether satisfied application demand according to this reference position and size, if met, then directly applies for this storage space.
Please join Fig. 3 further, as an example, can split for the mode of the storage space of memory device according to Fig. 3.Storage space is divided into several regions region, and each region comprises data area DR (data region) and mapping area MR (map region); Wherein MR comprises main mapping area PMR and mapping area SMR; PMR comprises head zone (map region head) and have recorded the position (SMR Entry) of each SMR and the content area of next PMR position (Next PMR Entry) that belong to this PMR management; SMR comprises head zone (Map Region Head) and records the content area of described daily record (Map Region Journal).
What this storage space adopted is the mode that two-stage maps, and each region region can divide according to certain size, such as 128GB.Whole like this storage space will be divided into some region, and each region all comprises above-mentioned two-stage mapping structure, manages separately.
As an example, daily record Map Region Journal (MRJ) please refer to the drawing 4 of the content area record of SMR.Every bar MRJ comprises: Reserved (reserving) position of 2bit.Account for the Ops_flag of 2bit, wherein low bit represents the operation in Free up Memory or application space, such as releasing operation is 0, and application is operating as 1; This daily record of higher bit bit representation is real application/release daily record, or merges the daily record generated, such as real application/release daily record is 0, and merging the daily record generated is 1.Account for (also representing that DR max cap. is (2^28-1) * 4KB ≈ 1TB) Offset and Size of 28bit respectively, this Offset expression operates the skew (relative position of space in DR related to, in units of 4KB), Size expression operates the size related to.Cur_journal_id represents the id of current log, often remembers that this id of daily record will add once certainly.Above the reference position of the storage space of log recording represents with Offset in this example.
Above-mentioned PMR information, SMR information are stored in the internal memory of memory device; Other information are then recorded in disk space.The space taken needed for PMR information, SMR information is less, so lower to internal memory size requirements.Certainly, the information of the last accessed storage space of DR corresponding to SMR can also be preserved further in internal memory, like this when subsequent application storage space, if this storage space meets the requirement of application, directly wherein draw get space from this.
Because the size of the SMR content regions for log presets, and application operation, releasing operation, daily record that union operation is corresponding are ever-increasing, so adopt the mode of circulation covering here by log recording in SMR.
In order to manage storage space accurately, the merging for daily record is needed to take certain strategy.Strategy one: when the last time merge ratio that the merging daily record and current not merged application that produce or release daily record total quantity account for the total saturation of SMR content area daily record reach predetermined threshold value time, order is to described application or release daily record, and the merging daily record of last time performs the operation of storage space merging; Strategy two: at application storage space and when not meeting the storage space of size needed for this application, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges.Here be citing two examples, the present embodiment does not get rid of the strategy of other storage space managements.
The example that please refer to the drawing 5 provides, in the first row, " MRJ 45 " represents the 45th article of daily record, and " 1 " in the first row represents that the 45th article of daily record merges the daily record generated; The lexical or textual analysis that other each row are identical.As can be seen from this example, the daily record before the 45th article of daily record through merging treatment, and defines two merging daily records 45 and 46, and these 45 and 46 merge daily records is last merging daily records merging generation; Afterwards, newly-generated again some real (unconsolidated) Operation Logs: 47,48,49,50,51,52.The merging daily record that such last time merging produces and current not merged daily record are following daily record: 45 ~ 52, and totally 8, total saturation of SMR content area daily record is here 10, so its accounting is 80%.When predetermined threshold value in above-mentioned tactful one is 78%, then current needs carries out union operation to daily record 45 ~ 52.The situation that this consolidation strategy can avoid effective daily record unprocessed and capped in time occurs.
Strategy two is when applying for storage space, when space size to application is greater than the space size of current idle, whether the size that the mode that then daily record can be adopted to merge looks to merge free space corresponding to the new daily record that obtains meets the requirement of application, if, then application operation will be successful, directly the offset value in the space after merging be fed back to upper strata.
When the present embodiment carries out the union operation of storage space for daily record, RBTree can be relied on and carry out this union operation.Such as the not merged application of every bar or release daily record, according to this application or the reference position and the size that discharge log recording space, judge whether to merge with current node in RBTree, if can, merge; After merging by newly-generated node city in RBTree, and delete merged node.The content of this newly-generated node is: the reference position and the size that merge rear idle storage space.Each node of RBTree all records reference position and the size information of free space, and index value is described reference position.The above-mentioned principle that can merge is, the application/Free up Memory recorded in the free-space information of RBTree interior joint record and daily record overlaps, partially overlap, continuously.
Give some instances below above-mentioned union operation is described:
Example 1: have the node that represents free space in RBTree: offset value is 32, size value is 16K; Suppose every 4K, offset value increase by 1.Current needs carries out union operation to the release daily record of offset value to be 36, size be 4K.According to above-mentioned offset value and size value, judge that current log can merge with this node, the offset value that the new node after merging is corresponding is 32, size value is 20.So be inserted in RBTree according to its index value 32 by this new node, and the node of to be 32, size value by original offset value be 16K is deleted from RBTree.And in SMR, generating a merging daily record, the content of this merging daily record is: releasing operation, and offset value is 32, size value is 20.
The node that one represents free space is had: offset value is 32, size value is 16K in example 2, RBTree; Suppose every 4K, offset value increase by 1.Current needs carries out union operation to the release daily record of offset value to be 31, size value be 4KB.According to above-mentioned offset value and size value, judge that current log can merge with this node, the offset value that the new node after merging is corresponding is 31, size value is 20.So be inserted in RBTree according to its index value 31 by this new node, and the node of to be 32, size value by original offset value be 16K is deleted from RBTree.And in SMR, generating a merging daily record, the content of this merging daily record is: releasing operation, and offset value is 31, size value is 20.
The node that one represents free space is had: offset value is 32, size value is 16K in example 3, RBTree; Suppose every 4K, offset value increase by 1.Current needs carries out union operation to the release daily record of offset value to be 32, size value be 16KB.Directly ignore this release daily record, do not need to do any process.
The node that one represents free space is had: offset value is 32, size value is 16K in example 4, RBTree; Suppose every 4K, offset value increase by 1.It is current that to need offset value be 34, size value is 4KB, application date book merge.Now, need original node to split into two, and by original node deletion.Its offset of newly-generated node is respectively 32,35, size value and is respectively 8K, 4K.Now need in SMR, to generate two further and merge daily record.
The node that one represents free space is had: offset value is 32, size value is 16K in example 5, RBTree; Suppose every 4K, offset value increase by 1.Current needs merges the application date book of offset value to be 35, size value be 4KB.Now, delete origin node, the offset value of newly-generated node is 32, size value is 12K; In fact the size value of origin node is only needed directly to change 12K into.The merging daily record that this new node is corresponding is generated further in SMR.
The node that one represents free space is had: offset value is 32, size value is 16K in example 6, RBTree; Suppose every 4K, offset value increase by 1.Current needs merges the application date book of offset value to be 32, size value be 4KB.Now, delete origin node, the offset value of newly-generated node is 33, size value is 12K.The merging daily record that this new node is corresponding is generated further in SMR.
Here there is a kind of situation to need to illustrate, when memory device starts, first need the PMR information in disk, SMR information to read in internal memory.Now may there is the situation shown in Fig. 6 (a) or Fig. 6 (b).In Fig. 6 (a), because the MRJ that numbering journal_id is maximum is real application or release daily record, so now only need to be that 09,10,11 3 MRJ participates in daily record merging by journal_id to carry out union operation; And in Fig. 6 (b), because MRJ that numbering journal_id is maximum merges daily record, so now to carry out union operation, then need by journal_id be 06,07,08,09 these four daily records participate in merging.Wherein journal_id is the daily record of 06,07 is the last merging daily record produced; Journal_id be 08,09 daily record be last merge daily record produce after newly-generated real application/release daily record.If the result merged and current journal_id are that the result of 10,11 daily records is consistent, then do not need to carry out any operation.If the result merged is except when front journal_id is outside the result of 10,11 daily records, also comprise other amalgamation results, then increase the daily record of these other amalgamation results further, this situation is likely that system closedown or other situation cause the merging of last time not complete completely.
In addition, a bit needing in addition to illustrate, when applying for storage space, except the size of free space continuous in the DR considering SMR management, also needing to consider this free space as far as possible in the position that LBA (Logical Block Addressing) lba is little.This is mainly because for existing hard disk, magnetic track corresponding to little lba is in the outside of hard disk, when hard disc speed of rotation is definite value, there is higher linear velocity in outside, also just means and reads and writes speed (outside magnetic track and inner side magnetic track mean speed are than being 2:1) faster.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within the scope of protection of the invention.
Claims (8)
1. based on a memory space management for log mode, the method is applied to memory device, it is characterized in that, the method comprises:
When application storage space or release storage space, in the second mapping area SMR of default size, record corresponding application or release daily record, the reference position of this application or release log recording application or release storage space and size information;
According to certain strategy, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges, and in described SMR, record this merging daily record produced, the reference position of the storage space that described merging log recording is idle and size.
2. the method for claim 1, is characterized in that, the record aimed at day in SMR adopts the mode of circulation covering to carry out;
Described according to certain strategy, order is to described application or release daily record, and the operation that the merging daily record execution storage space of last time merges comprises:
When the last time merge ratio that the merging daily record and current not merged application that produce or release daily record total quantity account for the total saturation of SMR content area daily record reach predetermined threshold value time, order is to described application or release daily record, and the merging daily record of last time performs the operation of storage space merging; Or,
At application storage space and when not meeting the storage space of size needed for this application, order is to described application or release daily record, and the merging daily record of last time performs the operation that storage space merges.
3. the method for claim 1, it is characterized in that, order is to described application or release daily record, and the operation that the merging daily record execution storage space of last time merges comprises: order is for the not merged application of current every bar or release daily record, according to the reference position recorded in daily record and size, judge whether to merge with current node in RBTree, if can, merge; After merging by newly-generated node city in RBTree, and delete merged node; The content of this newly-generated node is: this merges reference position and the size of rear idle storage space.
4. the method for claim 1, is characterized in that, the daily record of recording in described SMR comprises: expression is the information of putspace operations or application spatial operation; Expression is the information of truly applying for the daily record of releasing operation or the daily record of union operation; The numbering Cur_journal_id of daily record, this numbering increases successively along with the increase of daily record;
Described order is to described application or release daily record, and being operating as of merging daily record execution storage space merging of last time numbers according to daily record the operation that the order increased progressively performs described merging successively.
5. the method for claim 1, is characterized in that, the method also comprises, and disk storage space is divided into several regions region, and each region comprises data area DR and mapping area MR; Wherein MR comprises main mapping area PMR and the second mapping area SMR; PMR comprises head zone and have recorded the position of each SMR and the content area of next PMR position that belong to this PMR management; SMR comprises head zone and records the content area of described daily record.
6. method as claimed in claim 5, is characterized in that, after described memory device startup work, will read PMR information, SMR information in internal memory, and carry out the management of storage space based on the PMR information in internal memory and SMR information from disk.
7. method as claimed in claim 6, it is characterized in that, if reading the maximum daily record of numbering in the SMR content area in internal memory from disk is merge daily record, when needs carry out union operation further, need again the last time merging daily record before described merging daily record, the application date book after last merging daily record, release daily record to be re-started a union operation.
8. the method for claim 1, is characterized in that, when applying for storage space, if there is the space available continuously that several meet application condition in the DR of SMR management, then selects the described free space that LBA (Logical Block Addressing) lba is little.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410548611.XA CN104281517B (en) | 2014-10-16 | 2014-10-16 | Log mode based memory space management method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410548611.XA CN104281517B (en) | 2014-10-16 | 2014-10-16 | Log mode based memory space management method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104281517A true CN104281517A (en) | 2015-01-14 |
CN104281517B CN104281517B (en) | 2017-05-17 |
Family
ID=52256415
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410548611.XA Active CN104281517B (en) | 2014-10-16 | 2014-10-16 | Log mode based memory space management method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104281517B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107229664A (en) * | 2016-03-25 | 2017-10-03 | Hgst荷兰公司 | Indirect data structure for managing file system metadata |
CN109284295A (en) * | 2018-10-17 | 2019-01-29 | 郑州云海信息技术有限公司 | A kind of data-optimized method and device |
CN112198818A (en) * | 2019-07-08 | 2021-01-08 | 浙江宇视科技有限公司 | Control method, device and equipment of stepping type driving structure and storage medium |
CN112306825A (en) * | 2019-07-31 | 2021-02-02 | 中科寒武纪科技股份有限公司 | Memory operation recording method and device and computer equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20040193803A1 (en) * | 2003-03-28 | 2004-09-30 | Kazuhiko Mogi | Cache management method for storage device |
CN1549616A (en) * | 2003-05-20 | 2004-11-24 | 深圳市中兴通讯股份有限公司南京分公 | Self-adaption information buffer storing method used for multi-sample |
US20070136479A1 (en) * | 2001-09-17 | 2007-06-14 | Miller Michael J | System for automated device-to-device transfer system |
CN101286177A (en) * | 2008-05-30 | 2008-10-15 | 中兴通讯股份有限公司 | Method and device for allocating space to file in file allocation table |
CN101567001A (en) * | 2009-05-22 | 2009-10-28 | 清华大学 | Method for managing metadata file layout of parallel file system |
CN102024034A (en) * | 2010-11-26 | 2011-04-20 | 中国科学院声学研究所 | Fragment processing method for high-definition media-oriented embedded file system |
-
2014
- 2014-10-16 CN CN201410548611.XA patent/CN104281517B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070136479A1 (en) * | 2001-09-17 | 2007-06-14 | Miller Michael J | System for automated device-to-device transfer system |
US20040193803A1 (en) * | 2003-03-28 | 2004-09-30 | Kazuhiko Mogi | Cache management method for storage device |
CN1549616A (en) * | 2003-05-20 | 2004-11-24 | 深圳市中兴通讯股份有限公司南京分公 | Self-adaption information buffer storing method used for multi-sample |
CN101286177A (en) * | 2008-05-30 | 2008-10-15 | 中兴通讯股份有限公司 | Method and device for allocating space to file in file allocation table |
CN101567001A (en) * | 2009-05-22 | 2009-10-28 | 清华大学 | Method for managing metadata file layout of parallel file system |
CN102024034A (en) * | 2010-11-26 | 2011-04-20 | 中国科学院声学研究所 | Fragment processing method for high-definition media-oriented embedded file system |
Non-Patent Citations (2)
Title |
---|
岳丽华 等: "基于分离日志的闪存数据库***存储管理方法", 《中国科学技术大学学报》 * |
钟宝荣 等: "内存数据库中空闲页面管理的方法研究", 《计算机工程与设计》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107229664A (en) * | 2016-03-25 | 2017-10-03 | Hgst荷兰公司 | Indirect data structure for managing file system metadata |
CN109284295A (en) * | 2018-10-17 | 2019-01-29 | 郑州云海信息技术有限公司 | A kind of data-optimized method and device |
CN109284295B (en) * | 2018-10-17 | 2021-09-17 | 郑州云海信息技术有限公司 | Data optimization method and device |
CN112198818A (en) * | 2019-07-08 | 2021-01-08 | 浙江宇视科技有限公司 | Control method, device and equipment of stepping type driving structure and storage medium |
CN112198818B (en) * | 2019-07-08 | 2022-05-13 | 浙江宇视科技有限公司 | Control method, device and equipment of stepping type driving structure and storage medium |
CN112306825A (en) * | 2019-07-31 | 2021-02-02 | 中科寒武纪科技股份有限公司 | Memory operation recording method and device and computer equipment |
CN112306825B (en) * | 2019-07-31 | 2024-02-27 | 中科寒武纪科技股份有限公司 | Memory operation recording method and device and computer equipment |
Also Published As
Publication number | Publication date |
---|---|
CN104281517B (en) | 2017-05-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9298384B2 (en) | Method and device for storing data in a flash memory using address mapping for supporting various block sizes | |
CN105589812B (en) | Disk fragments method for sorting, device and host | |
CN105574104B (en) | A kind of LogStructure storage system and its method for writing data based on ObjectStore | |
CN105573681B (en) | Method and system for establishing RAID in SSD | |
CN104699417B (en) | Data storage device and data access method thereof | |
CN105930282B (en) | A kind of data cache method for NAND FLASH | |
CN104423894B (en) | Data memory device and method for controlling flash memory | |
CN104484283B (en) | A kind of method for reducing solid state disk write amplification | |
CN107728937B (en) | Key value pair persistent storage method and system using nonvolatile memory medium | |
US20140082265A1 (en) | Data storage device and flash memory control method thereof | |
CN103838853B (en) | Mixed file system based on different storage media | |
CN104866428A (en) | Data access method and data access device | |
CN103440207A (en) | Caching method and caching device | |
CN103324578A (en) | NAND flash memory device and random writing method thereof | |
CN104238962A (en) | Method and device for writing data into cache | |
CN104268088A (en) | Vehicle DVR (Digital Video Recorder) hard disk data storage method | |
CN103559027A (en) | Design method of separate-storage type key-value storage system | |
KR101374065B1 (en) | Data Distinguish Method and Apparatus Using Algorithm for Chip-Level-Parallel Flash Memory | |
CN106815152A (en) | A kind of method for optimizing page level flash translation layer (FTL) | |
CN104281517A (en) | Log mode based memory space management method and device | |
CN101169761A (en) | Large capacity cache implement method and storage system | |
CN102033793A (en) | Snapshot method and solid-state hard disk | |
CN108628542A (en) | A kind of Piece file mergence method and controller | |
CN105260139B (en) | A kind of disk management method and system | |
CN103729149A (en) | Data storage method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |