CN106527971A - Method for improving read performance of SSD during junk recovery period - Google Patents
Method for improving read performance of SSD during junk recovery period Download PDFInfo
- Publication number
- CN106527971A CN106527971A CN201610877009.XA CN201610877009A CN106527971A CN 106527971 A CN106527971 A CN 106527971A CN 201610877009 A CN201610877009 A CN 201610877009A CN 106527971 A CN106527971 A CN 106527971A
- Authority
- CN
- China
- Prior art keywords
- data
- page
- read
- raid
- read request
- 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
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
- G06F3/061—Improving I/O performance
-
- 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
-
- 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
- G06F3/0689—Disk arrays, e.g. RAID, JBOD
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)
- Techniques For Improving Reliability Of Storages (AREA)
Abstract
The invention discloses a method for improving read performance of an SSD (Solid State Drive) during a junk recovery period. The method comprises the steps that an RAID is established among channels by taking data pages as stripes; when a read request is processed, if a die in which the data page accessed by the read request is located is not being subjected to erasure operation, the read request makes a response normally at the moment; and if the die in which the data page accessed by the read request is located is being occupied by the erasure operation, the die is out operation briefly, and at the moment, data of other data pages on the same stripe is read by utilizing the RAID among the channels to perform RAID re-establishment, thereby making a response to the read request in time. According to the method, the system time delay caused by the junk recovery operation in a solid-state storage device based on an NAND Flash is shortened by utilizing the RAID technology in the SSD, so that the problem of a conflict between the normal read request of a system and the junk recovery erasure operation is solved.
Description
Technical field
The invention belongs to solid-state disk (SSD) readwrite performance control technology field, and in particular to a kind of SSD that lifted is returned in rubbish
The method of reading performance during receipts.
Background technology
In the last few years, based on NAND Flash solid state hard disc (Solid State Drive, SSD) is as performance is high, work(
Consume low, reliability it is high, shock resistance is good, noiseless many advantages, such as be widely applied.But SSD is using a period of time
Afterwards, its access performance can be decreased significantly.Garbage reclamation has significant impact to the performance of SSD, by Intel SSD DC
P3700 (800G) carries out test discovery, is 4KB for reading and writing granularity, and read-write ratio is 7:3 random mixing request.Work as Intel
When SSD is blank panel, its IOPS can reach 200,000, and readwrite bandwidth can reach 800MB/s;It is pre-buried when data are carried out to Intel SSD
After fragmentation process, its IOPS drops to 9W, and readwrite bandwidth drops to 360MB/s.Main cause is, using a period of time
Afterwards, inside SSD garbage reclamation frequent triggering, reduce the readwrite performance of SSD.
As flash chip has the characteristic for writing front erasing, strange land more New Policy when updating the data, is generally adopted, this can produce
The a large amount of invalid pages of life.For ensureing the normal use of SSD, need garbage collection operation to be carried out to invalid page, wipe the target of selection
Block is to reuse.Due to flash chip read-write granularity with erasing granularity it is different, erasing one block before, garbage reclamation one
As be related to the migration of valid data, the chip in Data Migration, residing for the source Physical Page and purpose Physical Page of data
And passage (channel) can be occupied (chip).Perform erasing operation when, although will not active channel resource, but wipe it is right
As residing wafer (die) can be occupied.Therefore, in garbage collection procedure, the readwrite performance of SSD will be a greater impact.
As shown in figure 3, a complete garbage collection operation includes three steps:1) object block to be reclaimed is selected, 2) is carried out
The migration of valid data, 3) wipes object block.When object block is wiped, whole die will be occupied, easily causes request collision,
Extraneous request can not be timely responded to.The conflict of write request can be solved by modes such as caching and headspaces, read in comparison
Conflict is more difficult, can only typically be forced to wait erasing operation to complete, and synchronous read operation in addition not only affects this passage
(channel) request, can also affect the response of each passage (channel).When comparing erasing operation and the read operation of Flash
Between, the erasing operation time is ms levels, and read time is us levels, and both are close to 100 by time ratio:1;And with SSD capacity
Increase, erasing block size further becomes big, and both time phase differences can be increasing.In sum, for read request, it is forced
Wait erasing operation complete can significantly reduce the access performance of SSD.
Because garbage reclamation is related to the migration of valid data and block erasing operation, a large amount of system resources is taken, is to the greatest extent may be used
Impact of the garbage reclamation to SSD access performances can be reduced, presently, there are the system idling-resource of making full use of carry out garbage reclamation or
The scheme of the garbage reclamation by utmost improving garbage reclamation efficiency, but in these schemes, garbage reclamation is considered all the time
One can not be split entirety, lack the consideration that garbage reclamation conflicts to SSD normal requests.For solving the above problems, prior art
In occur in that a kind of interruptable garbage reclamation mechanism, its migrate during valid data migration, often data page can in
Disconnected ongoing garbage reclamation responds normal read-write requests.The program can solve garbage reclamation and just to a certain extent
The often conflict of request, but which interrupts selection of time and has limitation, can only occur, do not account for garbage reclamation in Data Migration
During take the conflict of most long erasing operation and read-write requests.
The content of the invention
Disadvantages described above or Improvement requirement for prior art, the invention provides a kind of SSD that lifted is in the garbage reclamation phase
Between reading performance method, which passes through to build RAID in solid-state disk interchannel such that it is able to timely respond to read request, solves current
SSD during garbage reclamation read request performance it is low, to overcome present in prior art
The conflict defect of erasing operation and read-write requests.
For achieving the above object, it is proposed, according to the invention, a kind of method of lifting SSD reading performances during garbage reclamation is provided,
Including:
RAID is set up in each interchannel by band of data page;
When processing read request, if the wafer (die) that the data page that read request is accessed is located does not carry out erasing operation, this
When read request normal response;
If it is shared that the wafer (die) that the data page that read request is accessed is located just is wiped free of operation, there is short duration failure,
Now carry out RAID reconstructions using the interchannel RAID by reading the data of other data pages on same band, so as to and
When respond the read request.
Used as present invention further optimization, the RAID is rebuild i.e. by other data page XORs in RAID stripe
Mode come the data page that recovers to read required for the read request.
As present invention further optimization, if erasing operation is completed in RAID process of reconstruction is carried out, wafer (die) can
Again it is accessed, then eliminates the RAID and rebuild vestige, normal response read request.
As present invention further optimization, the RAID of each passage rebuild read operation can parallel processing, and do not affect
Normal read-write requests on the passage, rebuild read operation and complete to have sequencing.
As present invention further optimization, verification data page data is calculated or is updated in write request per treatment,
The calculating or renewal verification data page data include that whole band is write data, individual data page for the first time and writes number for the first time
Update the data according to, individual data web update data and whole band.
Used as present invention further optimization, the whole band first time data packets are included while writing number to each data page
According to verification page data does XOR by each data page data and obtains, and each data page and verification page are constituted a band finally
Write flash memory.
Used as present invention further optimization, the individual data page first time data packets are included in any bar band
Individual new data page writes data, and original checksums page data is read from flash memory DRAM first, then verifies the check number of page
New verification data is obtained according to the new data page data XOR, finally the new data page data and new verification data are write
Flash memory.
Used as present invention further optimization, the individual data web update data packets are included the individual data page
The former verification data of former data and verification page is read out from flash memory, then by the former data and new data step-by-step XOR,
Result after XOR again with former verification data step-by-step XOR, the result for obtaining be new verification data, finally by new data with it is new
Verification data is write back in flash memory.
Used as present invention further optimization, the whole band updates data packets and includes while using new data to update respectively
Data page, verification page data are obtained by the new data XOR of each data web update, finally by the new data of each data web update and
New verification page data writes back flash memory.
It is another aspect of this invention to provide that a kind of method by the built-in RAID response systems read request of interchannel is provided,
Its step is as follows:
1) RAID is set up in each interchannel by band of data page;
2) there is read request, put counter i=1;
3) judge whether the wafer (die) that the read request is located is wiped free of operation occupancy, be to go to step 4), no step turns
9);
4) judge that i, whether equal to 1, is to go to step 5), otherwise goes to step 6);
5) generate n-1 reconstruction read request to be mounted on n-1 passage respectively, read the data of remaining RAID stripe, its
Middle n is total number of channels;
6) n-i passage of parallel processing, successively existing read-write requests in treatment channel;
7) respective data blocks of same band on passage are read, the reconstruction read request on a passage, i++ is often processed;
8) judge that n-i, whether equal to 0, is to go to step 10), otherwise goes to step 3);
9) data block is directly read, the vestige that RAID rebuilds is removed, is gone to step 11);
10) according to existing n-1 data block, recover data block to be read using RAID5;
11) respond read request.
In general, by the contemplated above technical scheme of the present invention compared with prior art, with following beneficial effect
Really:
(1), in the present invention, using the built-in RAID of SSD, the die for wiping block place is regarded as of short duration in garbage collection procedure
Failure, is similar to the degraded mode of RAID, responds read request, in can effectively solve the problem that garbage reclamation by way of RAID rebuilds
Read conflict.
(2), in the present invention, RAID is built in interchannel, make full use of the concurrency inside SSD, because the data of solid-state disk
Read-write granularity is Hash memory pages, therefore its band is organized in units of page.
Description of the drawings
Fig. 1 is the multichannel SSD configuration diagrams of the embodiment of the present invention;
Fig. 2 is the internal structure schematic diagram of NAND Flash chips;
Fig. 3 is the process schematic that NAND Flash perform garbage reclamation;
Fig. 4 is the read-write requests process chart of the embodiment of the present invention;
Fig. 5 is that the SSD interchannels of the embodiment of the present invention build RAID5 operation principle schematic diagrams;
Fig. 6 be the embodiment of the present invention method in for different write requests 4 kinds calculate verification page mode schematic diagrames;
The schematic diagram for rebuilding data page is brought in the method for Fig. 7 embodiment of the present invention using RAID5 sames;
Fig. 8 is that the method for the embodiment of the present invention processes read request schematic flow sheet.
Specific embodiment
In order that the objects, technical solutions and advantages of the present invention become more apparent, it is below in conjunction with drawings and Examples, right
The present invention is further elaborated.It should be appreciated that specific embodiment described herein is only to explain the present invention, and
It is not used in the restriction present invention.
As shown in figure 1, SSD is made up of a series of flash arrays, inside SSD, there are multiple flash controllers (nfc), each sudden strain of a muscle
Memory controller controls a passage (channel), works independently between flash controller, and each passage has the passage of oneself total
Line;The multiple flash chips of carry (chip) on each passage.Fig. 2 is chip internal structure, and each flash chip is by multiple wafers
(die) constitute;Each wafer is made up of multiple packets (plane);Each packet has multiple pieces (block), and block is flash chip
The base unit of erasing, each block have multiple pages (page), and page is the base unit of flash chip read-write.
According to the method for one embodiment of the invention, as shown in figure 5, first with data page between each passage (channel)
(page) RAID being set up for band, being preferably RAID5 in the present embodiment, the port number of built-in RAID5 is preferably 5, but this
This is not limited in bright, as long as the RAID that can be verified is all suitable for this.Each band is by D0, tetra- data pages of D1, D2, D3 and P
One verification page composition, is belonging respectively to constitute 5 passages of RAID5.
For the presence of the built-in RAID of SSD, it is the uniformity for ensureing SSD data, write request per treatment is except writing which
Outside the data page of body, the verification data page of RAID stripe can be also updated.Calculating or update verification data page includes whole bar
Band is write for the first time data, individual data page and writes data, individual data page for the first time and writes data and individual data page for the first time more
New writen data.
Write in this programme that data page and updating the data is write and whole piece band is write data and writes the calculating of individual data page for the first time
The mode of verification data page is different.
As shown in fig. 6, for whole band writes data for the first time, its detailed process can be as follows:Such as Fig. 6 (A), band is
Sky, data D0, D1, D2, D3 place page are sky at the beginning, and the page that verification data P is located is also sky.While data D0 are write, D1,
After D2, D3, verification data P is done XOR and is obtained by data D0, D1, D2, D3, finally by data D0, D1, D2, D3 and verification
Data P all write flash memory and constitute a band.
As shown in fig. 6, for individual data page writes data for the first time, its detailed process can be as follows:Such as Fig. 6 (B), to certain
New data D0 is write in one band, original checksums data P is read into DRAM from flash memory first, then verification data P and number
New verification data P is obtained according to D0 XORs ', finally by data D0 and new verification data P ' write flash memory.
As shown in fig. 6, for individual data web update writes data, its detailed process can be as follows:Such as Fig. 6 (C), when will make
When updating legacy data D0 with new data D0 ', first legacy data D0 and verification data P0 are read out from flash memory, then will
Legacy data D0 and new data D0 ' step-by-step XORs, the result after XOR again with verification data P0 step-by-step XOR, result P0 for obtaining '
As new verification data, finally by new data D0 ' and new verification data P0 ' write back in flash memory.
As shown in fig. 6, for whole band more new writen data, its detailed process can be as follows:As shown in Fig. 6 (D), while
Using new data D0 ', D1 ', D2 ', D3 ' updates legacy data D0, D1, D2, D3, new verification data P ' by new data D0 ', D1 ',
D2 ', D3 ' XOR obtain, finally by new data D0 ', D1 ', D2 ', D3 ' and new verification data P ' flash memory is write back together.
As shown in fig. 7, the process of description data reconstruction:Now there is read request read data D2 on channel2, D2
Affiliated die is wiped free of operation and takes, and short duration failure occurs, by way of the data XOR of other pages in RAID5 bands
To recover data D2.
In the present invention, lifting SSD methods of reading performance during garbage reclamation mainly includes:
RAID is set up in each interchannel by band of data page;
When processing read request, if the wafer (die) that the data page that read request is accessed is located is not carrying out erasing behaviour
Make, now read request normal response;
If it is shared that the wafer (die) that the data page that read request is accessed is located just is wiped free of operation, there is short duration failure,
Now carry out RAID reconstructions using the interchannel RAID by reading the data of other data pages on same band, so as to and
When respond the read request.
Wherein RAID rebuild i.e. by other data pages in RAID stripe by way of XOR recovering the read request
The data page of required reading.
Specifically, as shown in figure 8, the method for lifting SSD reading performances during garbage reclamation of one embodiment of the invention,
Carry out by way of in the built-in RAID5 response systems read request of interchannel, which comprises the following steps that:
1. there is read request, put counter i=1, i is used for recording the completed port number of reconstruction read request;
2. judge whether the die that the read request is located is wiped free of operation occupancy, be to go to step 3., 8. no step turns;
3. judge that i, whether equal to 1, is to go to step 4., otherwise goes to step 5.;
4. generate (n-1) individual reconstruction read request to be mounted on (n-1) individual passage respectively;Read the number of remaining RAID stripe
According to n is total number of channels;
5. the individual channel of parallel processing (n-i);Existing read-write requests in treatment channel successively;
6. the respective data blocks of same band on channel are read, the reconstruction read request on a channel, i is often processed
++;
7. judge whether (n-i) is equal to 0, be to turn 9., no turn 2.;
8. data block is directly read, the vestige that RAID rebuilds is removed, is turned 10.;
9. according to existing n-1 data block, recover data block to be read using RAID5;
10. respond read request.
In this programme, the request on passage is sequential processes, rebuilds read request and is also sequentially mounted to correspondence channel
On.Because existing number of request is typically differed on each channel, the time for completing to rebuild reading is also different.By wheel
The mode of inquiry come ensure all channel all complete rebuild read.It is because there may be before RAID reconstruction operations do not complete, former
The upper erasing operations of channel are completed one step ahead, now need not carry out RAID reconstructions again, need what the raid that will do not completed rebuild
Vestige is removed (mainly remove the data for rebuilding that read request is generated), if that is, in RAID process of reconstruction is carried out, erasing operation is complete
Into wafer (die) can be accessed again, then eliminate the RAID and rebuild vestige, normal response read request.
As it will be easily appreciated by one skilled in the art that the foregoing is only presently preferred embodiments of the present invention, not to
The present invention, all any modification, equivalent and improvement made within the spirit and principles in the present invention etc. is limited, all should be included
Within protection scope of the present invention.
Claims (10)
1. a kind of method of lifting SSD reading performances during garbage reclamation, including:
RAID is set up in each interchannel by band of data page;
When processing read request, if the wafer (die) that the data page that read request is accessed is located is not carrying out erasing operation, this
When read request normal response;
If it is shared that the wafer (die) that the data page that read request is accessed is located just is wiped free of operation, there is short duration failure, now
RAID reconstructions are carried out using the interchannel RAID by the data for reading other data pages on same band, so as to ring in time
Should read request.
2. the method for a kind of lifting SSD reading performances during garbage reclamation according to claim 1, wherein, the RAID
Rebuild the data i.e. by other data pages in RAID stripe by way of XOR to recover to read required for the read request
Page.
3. a kind of method of lifting SSD reading performances during garbage reclamation according to claim 1 and 2, wherein, is being carried out
If the erasing operation is completed in RAID process of reconstruction, i.e., wafer (die) can be accessed again, then eliminate the RAID and rebuild trace
Mark, normal response read request.
4. the method for a kind of lifting SSD reading performances during garbage reclamation according to any one of claim 1 to 3, its
In, the RAID of each passage rebuild read operation can parallel processing, and do not affect the normal read-write requests on the passage.
5. the method for a kind of lifting SSD reading performances during garbage reclamation according to any one of claim 1 to 4, its
In, in write request per treatment calculate or update verification data page, the calculating or update verification data page include it is whole
Individual band is write data, individual data page for the first time and writes the renewal of data, individual data web update data and whole band for the first time
Data.
6. a kind of method of lifting SSD reading performances during garbage reclamation according to claim 5, wherein, described whole
Band first time data packets are included while writing data to each data page, verification page data does XOR by each data page data and obtains
Arrive, each data page data and verification page data are constituted into a band finally and writes flash memory.
7. the method for a kind of lifting SSD reading performances during garbage reclamation according to claim 5 or 6, wherein, the list
Individual data page first time data packets are included to the new data page write data in any bar band, first by original checksums page from
DRAM is read in flash memory, and the verification data for then verifying page obtains new verification data with the new data page data XOR, most
Afterwards by the new data page data and new verification data write flash memory.
8. the method for a kind of lifting SSD reading performances during garbage reclamation according to any one of claim 5-7, its
In, the individual data web update data packets include by the former data of the individual data page and verification page former verification data from
Read out in flash memory, then by the former data and new data step-by-step XOR, the result after XOR is pressed with former verification data again
Position XOR, the result for obtaining are new verification data, are finally write back in flash memory new data with new verification data.
9. the method for a kind of lifting SSD reading performances during garbage reclamation according to any one of claim 5-8, its
In, the whole band updates data packets and includes while updating each data page data using new data, verification page data is by each number
Obtain according to the new data XOR of web update, the new data of each data web update and new verification page data are write back into flash memory finally.
10. a kind of method by the built-in RAID response systems read request of interchannel, its step are as follows:
1) RAID is set up in each interchannel by band of data page;
2) there is read request, put counter i=1;
3) judge whether the wafer (die) that the read request is located is wiped free of operation occupancy, be to go to step 4), 9) no step turns;
4) judge that i, whether equal to 1, is to go to step 5), otherwise goes to step 6);
5) generate n-1 reconstruction read request to be mounted on n-1 passage respectively, read the data of remaining RAID stripe, wherein n is
Total number of channels;
6) n-i passage of parallel processing, successively existing read-write requests in treatment channel;
7) respective data blocks of same band on passage are read, the reconstruction read request on a passage, i++ is often processed;
8) judge that n-i, whether equal to 0, is to go to step 10), otherwise goes to step 3);
9) data block is directly read, the vestige that RAID rebuilds is removed, is gone to step 11);
10) according to existing n-1 data block, recover data block to be read using RAID5;
11) respond read request.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610877009.XA CN106527971B (en) | 2016-10-08 | 2016-10-08 | A method of promoting SSD reading performance during garbage reclamation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610877009.XA CN106527971B (en) | 2016-10-08 | 2016-10-08 | A method of promoting SSD reading performance during garbage reclamation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106527971A true CN106527971A (en) | 2017-03-22 |
CN106527971B CN106527971B (en) | 2019-11-26 |
Family
ID=58333045
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610877009.XA Active CN106527971B (en) | 2016-10-08 | 2016-10-08 | A method of promoting SSD reading performance during garbage reclamation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106527971B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107678690A (en) * | 2017-09-29 | 2018-02-09 | 山东超越数控电子有限公司 | A kind of implementation method of solid state hard disc and its RAID array |
CN109144407A (en) * | 2017-06-16 | 2019-01-04 | 希捷科技有限公司 | For monitoring memory so as to the system and method for resignation |
CN110119250A (en) * | 2019-05-13 | 2019-08-13 | 湖南国科微电子股份有限公司 | Non-volatile memory medium data processing method and non-volatile memory medium |
CN110895445A (en) * | 2018-09-12 | 2020-03-20 | 华为技术有限公司 | Data processing method and system |
CN112527694A (en) * | 2020-12-15 | 2021-03-19 | 深圳星火半导体科技有限公司 | Fragmentation degree alarm method and device |
WO2021068517A1 (en) * | 2019-10-10 | 2021-04-15 | 苏州浪潮智能科技有限公司 | Stored data sorting method and device |
CN113495830A (en) * | 2020-04-02 | 2021-10-12 | 深圳星火半导体科技有限公司 | Method and system for testing and analyzing efficiency of garbage recovery algorithm |
CN114327287A (en) * | 2021-12-30 | 2022-04-12 | 苏州浪潮智能科技有限公司 | Wafer setting method, device, equipment and medium for solid state disk |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103336744A (en) * | 2013-06-20 | 2013-10-02 | 华中科技大学 | Garbage recovery method for solid-state storage device and system for garbage recovery method |
CN103412826A (en) * | 2013-07-18 | 2013-11-27 | 记忆科技(深圳)有限公司 | Garbage collection method and system of solid state disk |
-
2016
- 2016-10-08 CN CN201610877009.XA patent/CN106527971B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103336744A (en) * | 2013-06-20 | 2013-10-02 | 华中科技大学 | Garbage recovery method for solid-state storage device and system for garbage recovery method |
CN103412826A (en) * | 2013-07-18 | 2013-11-27 | 记忆科技(深圳)有限公司 | Garbage collection method and system of solid state disk |
Non-Patent Citations (1)
Title |
---|
吴素贞、陈晓熹、毛波: "GC-RAIS:一种基于垃圾回收感知的固态盘阵列", 《计算机研究与发展》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109144407A (en) * | 2017-06-16 | 2019-01-04 | 希捷科技有限公司 | For monitoring memory so as to the system and method for resignation |
CN107678690A (en) * | 2017-09-29 | 2018-02-09 | 山东超越数控电子有限公司 | A kind of implementation method of solid state hard disc and its RAID array |
CN110895445A (en) * | 2018-09-12 | 2020-03-20 | 华为技术有限公司 | Data processing method and system |
CN110895445B (en) * | 2018-09-12 | 2021-09-14 | 华为技术有限公司 | Data processing method and system |
CN110119250A (en) * | 2019-05-13 | 2019-08-13 | 湖南国科微电子股份有限公司 | Non-volatile memory medium data processing method and non-volatile memory medium |
WO2021068517A1 (en) * | 2019-10-10 | 2021-04-15 | 苏州浪潮智能科技有限公司 | Stored data sorting method and device |
CN113495830A (en) * | 2020-04-02 | 2021-10-12 | 深圳星火半导体科技有限公司 | Method and system for testing and analyzing efficiency of garbage recovery algorithm |
CN112527694A (en) * | 2020-12-15 | 2021-03-19 | 深圳星火半导体科技有限公司 | Fragmentation degree alarm method and device |
CN114327287A (en) * | 2021-12-30 | 2022-04-12 | 苏州浪潮智能科技有限公司 | Wafer setting method, device, equipment and medium for solid state disk |
CN114327287B (en) * | 2021-12-30 | 2023-08-04 | 苏州浪潮智能科技有限公司 | Method, device, equipment and medium for setting wafer of solid state disk |
Also Published As
Publication number | Publication date |
---|---|
CN106527971B (en) | 2019-11-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106527971A (en) | Method for improving read performance of SSD during junk recovery period | |
CN104503706B (en) | A kind of data storage and read method based on disk array | |
CN103488583B (en) | The solid-state disk implementation method that a kind of high-performance is highly reliable | |
CN106681934B (en) | Method and device for recycling garbage of storage device | |
US8234519B2 (en) | Fast data recovery from HDD failure | |
CN102866956B (en) | System and method for data real-time tracking and storage based on solid-state storage medium | |
CN103218274B (en) | A kind of method that trouble saving is cumulative and solid state hard disc | |
CN106528438A (en) | Segmented junk recovery method for solid-state storage device | |
US20150378886A1 (en) | Software-defined ssd and system using the same | |
US10942849B2 (en) | Use of a logical-to-logical translation map and a logical-to-physical translation map to access a data storage device | |
CN103049222A (en) | RAID5 (redundant array of independent disk 5) write IO optimization processing method | |
US9251059B2 (en) | Storage system employing MRAM and redundant array of solid state disk | |
CN104750618A (en) | Data storage device and flash memory control method | |
CN109358809B (en) | RAID data storage system and method | |
CN110347613B (en) | Method for realizing RAID in multi-tenant solid-state disk, controller and multi-tenant solid-state disk | |
CN101567211A (en) | Method for improving usability of disk and disk array controller | |
CN110187830A (en) | A kind of method and system accelerating disk array reconstruction | |
CN1692335A (en) | Method and means for tolerating multiple dependent or arbitrary double disk failures in a disk array | |
CN103488432B (en) | Hybrid disk array, deferred write verification method for hybrid disk array, and data recovery method for hybrid disk array | |
WO2021139166A1 (en) | Error page identification method based on three-dimensional flash storage structure | |
US20140173223A1 (en) | Storage controller with host collaboration for initialization of a logical volume | |
CN103019893A (en) | Multi-disk fault-tolerant two-dimensional hybrid disk RAID4 system architecture and read-write method thereof | |
CN102023902B (en) | Disc array reconstruction method | |
CN102520890A (en) | RS (Reed-Solomon) - DRAID( D redundant array of independent disk) system based on GPUs (graphic processing units) and method for controlling data of memory devices | |
CN109032513B (en) | RAID (redundant array of independent disks) architecture based on SSD (solid State disk) and HDD (hard disk drive) and backup and reconstruction methods thereof |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |