CN102163175B - Hybrid address mapping method based on locality analysis - Google Patents
Hybrid address mapping method based on locality analysis Download PDFInfo
- Publication number
- CN102163175B CN102163175B CN 201110105491 CN201110105491A CN102163175B CN 102163175 B CN102163175 B CN 102163175B CN 201110105491 CN201110105491 CN 201110105491 CN 201110105491 A CN201110105491 A CN 201110105491A CN 102163175 B CN102163175 B CN 102163175B
- Authority
- CN
- China
- Prior art keywords
- data
- log
- piece
- sequential
- random
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a hybrid address mapping method based on locality analysis, which improves the service efficiency of sequential diary blocks of a flash memory through three sequential diary block recycling modes and reduces influence of a random I/O (Input/Output) request on system performance through space locality analysis and time locality analysis of a write request. And simultaneously, the address mapping method reduces cost on operations of recycling flash memory garbage, avoids unnecessary block erase operations and improves overall performance of the system.
Description
Technical field
The present invention relates to flash memory address mapping techniques field, be specifically related to a kind of combined address mapping method based on local addresses analysis.
Background technology
The nonvolatile memory flash memory has the advantages such as read or write speed is fast, storage density is high, energy consumption is low, volume is little, lightweight, shock resistance is strong, is widely used in the middle-size and small-size storage productss such as music player, digital camera, palm PC.And the trend that replaces traditional magnetic hard disk will be arranged in future.
Important operating feature of flash memory be exactly can not be on the spot new data more.To frequently the wiping and write operation of flash memory, flash data upgrades the mode that has adopted the strange land to upgrade when reducing new data more, and it is invalid that original data are become, and simultaneously new data is written in the storage area of free time.Although this update mode has reduced the frequency of wiping of piece, caused the difference of physical address corresponding to the same logical address of priority.For compatible with existing file system, flash memory is mapped as block device, thereby needs Address Mapping to come the corresponding relation of management logic address and physical address.The address mapping is converted into physical address with the logical address of request of access, guarantees the smooth execution of read-write operation.
Existing flash memory address mapping method is divided into page address mapping, block address mapping and combined address mapping according to the mapping unit-sized.The mapping mode of page address mapping is flexible, and the mapping effect is better, but has taken larger memory space.The block address mapping is subject to the restriction of fixed page offset address, and the mapping effect is relatively poor, has but reduced the demand of memory headroom.The advantage of the combined address mapping is compatible page address mapping and block address mapping.Log type mixes the renewal with daily record form store data that the sub-fraction storage space is adopted in mapping, stores according to the page address mapping mode.Log blocks is equivalent to the storage cache of data block, so data block and corresponding relation between log blocks can be divided into the link to each other mapping and comprehensively shining upon of single mapping, group.The log type combined address mapping method of mapping has improved the service efficiency of log blocks greatly comprehensively, has reduced the frequent degree that piece is wiped, but the way of recycling that comprehensively merges has but increased the complexity of garbage reclamation.The temporal locality of write request and spatial locality are that the regularity of the logical address of write request changes, and the performance of address mapping method is seriously influenced.
Summary of the invention
The deficiency that exists in order to overcome above-mentioned prior art, the object of the present invention is to provide a kind of combined address mapping method based on local addresses analysis, improved the service efficiency of flash memory sequential log piece by the improvement of three kinds of sequential log piece ways of recycling, and spatial locality analysis and temporal locality analysis by write request, improve the service efficiency of log blocks, reduce the expense of garbage reclamation operation, reduce the piece erase operation, improve the overall performance of system.
In order to achieve the above object, the technical solution adopted in the present invention is:
A kind of combined address mapping method based on local addresses analysis, when in the flash memory during data writing, feature according to the time local characteristics of operating load, it is the update times of data, with the daily record of flash memory to be written more new data be divided into cold data and dsc data, dsc data directly is assigned to the dsc data storage area, namely in the dsc data log blocks of flash memory; And cold data allocations is to cold data storage areas, be in the cold data logging piece of flash memory, specifically will divide into order request and random request to the write request of cold data according to the temporal locality principle, the cold data of order request are out left in as the sequential update data separating in the sequential log piece of cold data logging piece, deposit in the cold data of random request in the random log blocks of cold data logging piece as random data; If the data in sequential log piece and the random log blocks are updated again, just the data that again are updated in sequential log piece and the random log blocks are dumped in the dsc data log blocks as dsc data; After the free space of sequential log piece, random log blocks and dsc data log blocks is used to complete, by with flash memory in the union operation of corresponding data block, reclaim idle log store space for next Data Update.
It is that the mode of cold data and dsc data is that data separation is upgraded in described daily record with flash memory to be written, and when update times during greater than 2 times, the daily record of corresponding flash memory to be written more new data is dsc data; When update times during less than or equal to 2 times, the daily record of corresponding flash memory to be written more new data is cold data.
The spatial locality analysis of described data is according to size distinction order request and random request to the write request of cold data: when the size of write request is order request during greater than 4KB; When the size of write request is random request during less than or equal to 4KB.
The total storage space volume size of described dsc data log blocks and cold data logging piece is fixing, is 3% of whole flash memory space; And the storage space volume of sequential log piece is less than or equal to 1/4 of the total storage space volume of dsc data log blocks and cold data logging piece; The size of dsc data log blocks storage space can also dynamically be adjusted by operating load size different and the at random storage space of log blocks simultaneously, but can not surpass predefined scope.
The mode that described dsc data log blocks reclaims idle log store space is replacement policy or empties strategy, replacement policy or empty strategy with existing dsc data log blocks as random log blocks, and distribute new idle log blocks to the dsc data log blocks.
The union operation of corresponding data block has three kinds of improved procedures in described sequential log piece and the flash memory: the sequential log block gap is write, the sequential log piece upgrades to write and do not contain homepage with the sequential log piece and write; The sequential log block gap is write when referring to write for sequential log block gap free page, delay is to the recovery of this sequential log piece, and in the reclaimer operation that carries out subsequently, the data that this sequential log block gap free page writes need not be from random log blocks or data block querying replicated moving in the sequential log piece; The sequential log piece upgrades to be write when referring to that the data of storing in the sequential log piece are upgraded again, with the Data Migration that again upgrades in the dsc data log blocks, when merging the data block of this sequential log piece and correspondence, the disarmed state of this page or leaf keeps in the sequential log piece, adopt the mode that will not reclaim and leave accordingly the dsc data of depositing in the sequential log piece in, still leave in the dsc data piece; The sequential log piece does not contain homepage and refers to be the mode of the extra allocation order log blocks of this order request when the logic request address of order request and logical block boundary first address during at a distance of the certain deviation amount.
Improve the service efficiency of flash memory sequential log piece by the improvement of three kinds of sequential log piece ways of recycling, and passed through spatial locality analysis and the temporal locality analysis of write request, reduced the impact of stochastic pattern I/O request on system performance.This address mapping method has reduced the expense of flash memory garbage reclamation operation, has avoided unnecessary piece erase operation, has improved the overall performance of system.
Description of drawings
Accompanying drawing is the structural representation of address mapping method of the present invention.
Embodiment
The present invention will be described in more detail below in conjunction with accompanying drawing.
As shown in drawings, combined address mapping method based on local addresses analysis, when flash memory is sent write request, feature according to the time local characteristics of operating load, it is the update times of data, it is cold data and dsc data that data separation is upgraded in the daily record of flash memory to be written, and dsc data directly is assigned to the dsc data storage area, namely in the dsc data log blocks of flash memory; And cold data allocations is to cold data storage areas, be in the cold data logging piece of flash memory, specifically will divide into order request and random request to the write request of cold data according to the temporal locality principle, the cold data of order request are out left in as the sequential update data separating in the sequential log piece of cold data logging piece, deposit in the cold data of random request in the random log blocks of cold data logging piece as random data; If the data in sequential log piece and the random log blocks are updated again, just the data that again are updated in sequential log piece and the random log blocks are dumped in the dsc data log blocks as dsc data; After the free space of sequential log piece, random log blocks and dsc data log blocks is used to complete, by with flash memory in the union operation of corresponding data block, reclaim idle log store space for next Data Update.It is that the mode of cold data and dsc data is that data separation is upgraded in described daily record with flash memory to be written, and when update times during greater than 2 times, the daily record of corresponding flash memory to be written more new data is dsc data; When update times during less than or equal to 2 times, the daily record of corresponding flash memory to be written more new data is cold data.The spatial locality analysis of described data is according to size distinction order request and random request to the write request of cold data: when the size of write request is order request during greater than 4KB; When the size of write request is random request during less than or equal to 4KB.The total storage space volume size of described dsc data log blocks and cold data logging piece is fixing, is 3% of whole flash memory space; And the storage space volume of sequential log piece is less than or equal to 1/4 of the total storage space volume of dsc data log blocks and cold data logging piece; Simultaneously the size of dsc data log blocks storage space is can also be by operating load different and dynamically adjust according to the size of the storage space of random log blocks, but can not surpass certain threshold range.The mode that described dsc data log blocks reclaims idle log store space is replacement policy or empties strategy, replacement policy or empty strategy with existing dsc data log blocks as random log blocks, and distribute new idle log blocks to the dsc data log blocks.The union operation of corresponding data block has three kinds of improved procedures in described sequential log piece and the flash memory: the sequential log block gap is write, the sequential log piece upgrades to write and do not contain homepage with the sequential log piece and write; The sequential log block gap is write when referring to write for sequential log block gap free page, delay is to the recovery of this sequential log piece, and the data that this interval writes do not need to copy from random log blocks or data block when the sequential log piece merges to move in the random log blocks.And traditional union operation is in the situation that run into the sequential log block gap and write and just carry out immediately part and merge and reclaim this sequential log piece, ignored this log blocks writing in the renewal process subsequently, the possibility that the page at interval is filled again, and might be fully written by this log blocks, can carry out exchange merges, in the manner, this kind situation is not reclaimed immediately, but postpone the recovery of this sequential log piece, in reclaimer operation subsequently, this page data does not need querying replicated moving in the sequential log piece from random log blocks or data block, has effectively solved this problem; The sequential log piece upgrades to be write when referring to that the data of storing in the sequential log piece are upgraded again, with the Data Migration that again upgrades in the dsc data log blocks, and when merging the data block of this sequential log piece and correspondence, adopt the mode that will not reclaim for the dsc data of depositing in the sequential log piece, still leave in the dsc data piece.Traditional mode concerns for the sequential update that guarantees sequential log piece and data block in merging process, the sequential log piece is reclaimed immediately when Data Update is write, do not use completely in the clearance spaces of log blocks like this and wipe this log blocks in the situation, reduced the block space utilization rate, move in the dsc data log blocks as dsc data by the data that will again upgrade in the manner, thereby postponed the recovery of sequential log piece, improved the service efficiency of log blocks, when the data block of merge order log blocks and correspondence, adopt the mode that will not reclaim for the dsc data of depositing in the sequential log piece, still leave in the dsc data piece, reduced like this migration that copies of valid data; When the sequential log piece does not contain homepage and refers to that border first address when the logic request address of order request and logical block is at a distance of the certain deviation amount, be added to the mode of this order request allocation order log blocks.Traditional mode can't be asked the allocation order log blocks for this, processes thereby this order request is used as random request, has added in the manner to be the function of this request allocation order log blocks, has improved the processing power of order request in the Address Mapping.
Improve the service efficiency of flash memory sequential log piece by the improvement of three kinds of sequential log piece ways of recycling, and passed through spatial locality analysis and the temporal locality analysis of write request, reduced the impact of stochastic pattern I/O request on system performance.This address mapping method has reduced the expense of flash memory garbage reclamation operation, has avoided unnecessary piece erase operation, has improved the overall performance of system.
Claims (4)
1. combined address mapping method based on local addresses analysis, it is characterized in that: when in the flash memory during data writing, feature according to the time local characteristics of operating load, it is the update times of data, it is cold data and dsc data that data separation is upgraded in the daily record of flash memory to be written, dsc data directly is assigned to the dsc data storage area, namely in the dsc data log blocks of flash memory; And cold data allocations is to cold data storage areas, be in the cold data logging piece of flash memory, specifically will divide into order request and random request to the write request of cold data according to the temporal locality principle, the cold data of order request are out left in as the sequential update data separating in the sequential log piece of cold data logging piece, deposit in the cold data of random request in the random log blocks of cold data logging piece as random data; If the data in sequential log piece and the random log blocks are updated again, just the data that again are updated in sequential log piece and the random log blocks are dumped in the dsc data log blocks as dsc data; After the free space of sequential log piece, random log blocks and dsc data log blocks is used to complete, by with flash memory in the union operation of corresponding data block, reclaim idle log store space for next Data Update;
It is that the mode of cold data and dsc data is that data separation is upgraded in described daily record with flash memory to be written, and when update times during greater than 2 times, the daily record of corresponding flash memory to be written more new data is dsc data; When update times during less than or equal to 2 times, the daily record of corresponding flash memory to be written more new data is cold data;
The spatial locality analysis of described data is according to size distinction order request and random request to the write request of cold data: when the size of write request is order request during greater than 4KB; When the size of write request is random request during less than or equal to 4KB.
2. the combined address mapping method based on local addresses analysis according to claim 1 is characterized in that: the total storage space volume size of described dsc data log blocks and cold data logging piece is fixing, is 3% of whole flash memory space; And the storage space volume of sequential log piece is less than or equal to 1/4 of the total storage space volume of dsc data log blocks and cold data logging piece; Simultaneously the size of dsc data log blocks storage space is can also be by operating load different and dynamically adjust according to the size of the storage space of random log blocks, and is no more than certain threshold range.
3. the combined address mapping method based on local addresses analysis according to claim 1, it is characterized in that: the mode that described dsc data log blocks reclaims idle log store space is replacement policy or empties strategy, replacement policy or empty strategy with existing dsc data log blocks as random log blocks, and distribute new idle log blocks to the dsc data log blocks.
4. the combined address mapping method based on local addresses analysis according to claim 1 is characterized in that: the union operation of corresponding data block has three kinds of improved procedures in described sequential log piece and the flash memory: the sequential log block gap is write, the sequential log piece upgrades to write and do not contain homepage with the sequential log piece and write; The sequential log block gap is write when referring to write for sequential log block gap free page, by the recovery of delay to this sequential log piece, and these sequential log block gap free page data do not need querying replicated moving in the sequential log piece from random log blocks and data block in the reclaimer operation subsequently; The sequential log piece upgrades to be write when referring to that the data of storing in the sequential log piece are upgraded again, with the Data Migration that again upgrades in the dsc data log blocks, when merging the data block of this sequential log piece and correspondence, adopt the mode that will not reclaim for the dsc data of depositing in the sequential log piece, still leave in the dsc data piece; The sequential log piece does not contain homepage and refers to be the mode of the extra allocation order log blocks of this order request when the logic request address of order request and logical block boundary first address during at a distance of the certain deviation amount.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110105491 CN102163175B (en) | 2011-04-26 | 2011-04-26 | Hybrid address mapping method based on locality analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110105491 CN102163175B (en) | 2011-04-26 | 2011-04-26 | Hybrid address mapping method based on locality analysis |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102163175A CN102163175A (en) | 2011-08-24 |
CN102163175B true CN102163175B (en) | 2013-02-27 |
Family
ID=44464411
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110105491 Expired - Fee Related CN102163175B (en) | 2011-04-26 | 2011-04-26 | Hybrid address mapping method based on locality analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102163175B (en) |
Families Citing this family (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102385624B (en) * | 2011-10-25 | 2014-01-01 | 曙光信息产业(北京)有限公司 | DFS (distributed file system)-oriented log data organization method |
CN102521144B (en) * | 2011-12-22 | 2015-03-04 | 清华大学 | Flash translation layer system |
CN103034586B (en) * | 2012-11-30 | 2016-03-23 | 记忆科技(深圳)有限公司 | By method and the system thereof of flash translation layer (FTL) identification upper layer application |
CN103049312A (en) * | 2012-12-07 | 2013-04-17 | 记忆科技(深圳)有限公司 | Input/output (IO) optimizing processing method and system thereof |
CN103036994B (en) * | 2012-12-18 | 2015-08-19 | 曙光信息产业(北京)有限公司 | Realize the cloud storage system of load balancing |
CN103279530B (en) * | 2013-05-31 | 2016-07-27 | 上海携程商务有限公司 | The query composition buffer memory method for building up of time series data and system, the query composition caching method of time series data and system |
CN103365788B (en) * | 2013-08-06 | 2016-01-13 | 山东大学 | The adaptive local rubbish recovering method that real-time flash memory conversion layer uses |
CN103488583B (en) * | 2013-09-09 | 2016-08-10 | 华中科技大学 | The solid-state disk implementation method that a kind of high-performance is highly reliable |
CN103677670A (en) * | 2013-12-11 | 2014-03-26 | 华为技术有限公司 | Method and device for reading data |
CN107168891B (en) * | 2014-07-23 | 2020-08-14 | 华为技术有限公司 | I/O feature identification method and device |
CN106201906B (en) * | 2016-07-11 | 2019-01-18 | 浪潮(北京)电子信息产业有限公司 | A kind of cold and hot data separation method and system for flash memory |
CN106598768B (en) * | 2016-11-28 | 2020-02-14 | 华为技术有限公司 | Method and device for processing write request and data center |
CN108733577B (en) * | 2017-04-21 | 2021-10-22 | 群联电子股份有限公司 | Memory management method, memory control circuit unit and memory storage device |
CN108563586B (en) * | 2018-04-12 | 2021-11-02 | 华中科技大学 | Method for separating garbage recovery data and user data in solid-state disk |
CN109376094A (en) * | 2018-09-27 | 2019-02-22 | 郑州云海信息技术有限公司 | A kind of method of the cold and hot Data Migration of garbage reclamation in storage system |
CN111124254B (en) * | 2018-10-30 | 2023-09-29 | 伊姆西Ip控股有限责任公司 | Method, electronic device and program product for scheduling memory space reclamation requests |
CN111143313B (en) * | 2019-12-25 | 2023-07-25 | 成都三零嘉微电子有限公司 | Method for improving log block read-write performance of hybrid mapping algorithm |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101937319A (en) * | 2009-06-29 | 2011-01-05 | 联发科技股份有限公司 | Memory systems and mapping methods thereof |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101028929B1 (en) * | 2008-12-31 | 2011-04-12 | 성균관대학교산학협력단 | Methods of Distributing Log Block Associativity in Real-time System And Flash Memory Device Performing the Same |
-
2011
- 2011-04-26 CN CN 201110105491 patent/CN102163175B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101937319A (en) * | 2009-06-29 | 2011-01-05 | 联发科技股份有限公司 | Memory systems and mapping methods thereof |
Also Published As
Publication number | Publication date |
---|---|
CN102163175A (en) | 2011-08-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102163175B (en) | Hybrid address mapping method based on locality analysis | |
CN102981963B (en) | A kind of implementation method of flash translation layer (FTL) of solid-state disk | |
CN103049397B (en) | A kind of solid state hard disc inner buffer management method based on phase transition storage and system | |
CN102768645B (en) | The solid state hard disc forecasting method of hybrid cache and solid-state hard disk SSD | |
CN106547703B (en) | A kind of FTL optimization method based on block group structure | |
CN103136121B (en) | Cache management method for solid-state disc | |
US9690694B2 (en) | Apparatus, system, and method for an address translation layer | |
CN101169751B (en) | System with flash memory device and data recovery method thereof | |
CN102508788B (en) | SSD (solid state drive) and SSD garbage collection method and device | |
CN103365788B (en) | The adaptive local rubbish recovering method that real-time flash memory conversion layer uses | |
CN105335098A (en) | Storage-class memory based method for improving performance of log file system | |
CN104166634A (en) | Management method of mapping table caches in solid-state disk system | |
WO2014074449A2 (en) | Wear leveling in flash memory devices with trim commands | |
CN106528438A (en) | Segmented junk recovery method for solid-state storage device | |
CN102841851A (en) | Flash memory management method and flash memory device | |
CN102521160A (en) | Write buffer detector, addressing method of written data and parallel channel write method | |
CN106815152A (en) | A kind of method for optimizing page level flash translation layer (FTL) | |
CN102819494B (en) | Optimization method when a kind of flash memory is sequentially written in | |
CN103744798A (en) | Garbage recycling method of solid state discs | |
CN107797772A (en) | A kind of garbage retrieving system and method based on flash media | |
CN102306124A (en) | Method for implementing hardware driver layer of Nand Flash chip | |
KR101180288B1 (en) | Method for managing the read and write cache in the system comprising hybrid memory and ssd | |
CN101833514B (en) | Method for realizing virtual internal memory in nand type flash memory medium based on micromemory system | |
CN102521161B (en) | Data caching method, device and server | |
CN102880553B (en) | The reading/writing method of the outer FLASH file system of a kind of sheet based on MCU |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130227 Termination date: 20150426 |
|
EXPY | Termination of patent right or utility model |