CN110413537A - A kind of flash translation layer (FTL) and conversion method towards hybrid solid-state hard disk - Google Patents
A kind of flash translation layer (FTL) and conversion method towards hybrid solid-state hard disk Download PDFInfo
- Publication number
- CN110413537A CN110413537A CN201910675390.5A CN201910675390A CN110413537A CN 110413537 A CN110413537 A CN 110413537A CN 201910675390 A CN201910675390 A CN 201910675390A CN 110413537 A CN110413537 A CN 110413537A
- Authority
- CN
- China
- Prior art keywords
- cmt
- cold
- hot
- medium
- page
- 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
- G06F12/00—Accessing, addressing or allocating within memory systems or architectures
- G06F12/02—Addressing or allocation; Relocation
- G06F12/0223—User address space allocation, e.g. contiguous or non contiguous base addressing
- G06F12/023—Free address space management
- G06F12/0238—Memory management in non-volatile memory, e.g. resistive RAM or ferroelectric memory
- G06F12/0246—Memory management in non-volatile memory, e.g. resistive RAM or ferroelectric memory in block erasable memory, e.g. flash memory
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F12/00—Accessing, addressing or allocating within memory systems or architectures
- G06F12/02—Addressing or allocation; Relocation
- G06F12/0223—User address space allocation, e.g. contiguous or non contiguous base addressing
- G06F12/0292—User address space allocation, e.g. contiguous or non contiguous base addressing using tables or multilevel address translation means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
- G06F2212/10—Providing a specific technical effect
- G06F2212/1016—Performance improvement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
- G06F2212/10—Providing a specific technical effect
- G06F2212/1032—Reliability improvement, data loss prevention, degraded operation etc
- G06F2212/1036—Life time enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
- G06F2212/21—Employing a record carrier using a specific recording technology
- G06F2212/214—Solid state disk
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of flash translation layer (FTL)s towards hybrid solid-state hard disk, the flash translation layer (FTL) includes an address of cache buffer area, and the address of cache buffer area penetrates cache table Hot-CMT by global translation directory GTD, mapping page cache slot TPCS, hot showing and cold mapped cache table Cold-CMT is formed；The GTD is used to record the actual physics page number of each translation page, when the TPCS is not hit by the buffer for currently mapping item, entire mapping page belonging to the current load mapping item of caching, the Hot-CMT be used to cache frequently access write mapping item, for cache, reading maps item to the Cold-CMT and what is infrequently accessed write mapping item；Logical page number (LPN) LPN, the physical page number PPN of the Hot-CMT and Cold-CMT record mapping item.The present invention solves the flash translation layer (FTL) design problem using a variety of media building SSD.The unified management of mapping item is realized, the performance of mixing SSD is promoted, reduces its unit capacity cost, the abrasion equilibrium between matter A and medium B is preferably realized, extends the service life of mixing SSD.
Description
Technical field
The present invention relates to computer memory system fields, and in particular to a kind of flash translation layer (FTL) towards hybrid solid-state hard disk
And its conversion method.
Background technique
As solid state hard disk (Solid State Disk, SSD) designing technique is constantly progressive, SSD and traditional machinery
Hard disk is compared, and has many advantages, such as read or write speed fast, low in energy consumption, light-weight, shockproof drop resistant, this deposits SSD in computer
Storage field gradually buds out into popularity.
Currently, the storage medium of the SSD of mainstream is nand flash memory.Nand flash memory has following physical characteristic: (1) only providing
Three kinds of reading and writing, erasing operations, the performance of three kinds of operations is asymmetric, and wherein reading rate is most fast, writes and takes second place, wipes most slow.(2) it dodges
It deposits and carries out tissue by the structure of page (page), block (block), plane (plane), page is the minimum unit of read/write, and block is erasing
Minimum unit.(3) flash memory does not support original place to update, and can only write once after flash memory erasing, that is, write preceding erasing (erase before
Write), this causes huge expense.(4) program/erase (P/E) number of each storage unit of flash memory is limited, is more than time
Flash memory cell storing data after number will be no longer reliable.
Since nand flash memory has These characteristics, in SSD design, a flash translation layer (FTL) (Flash is generally required
Translation Layer, FTL), it is responsible for the above-mentioned characteristic of hiding nand flash memory, SSD is modeled to only have read-write operation
The form of conventional hard, to adapt to current file system.
The key technology of FTL has three address of cache, garbage reclamation and abrasion equilibrium modules.Address of cache is responsible for come from
The logical address of file system is mapped as the physical address in flash memory, generally there is the mapping of page grade, the mapping of block grade and mixed-use developments three
Kind mapping mode.Garbage reclamation be responsible for will wipe block operation after invalid page recycle, invalid page can be reused.Mill
Damage equilibrium is responsible for keeping the rate of depreciation between each piece consistent as far as possible, prevents partial block from leading to premature deterioration due to rapid wear.
Flash chip includes 3 classes, SLC (Single-Level Cell, i.e. 1bit/Cell), MLC (Multi-Level
Cell, i.e. 2bit/Cell) and TLC (Trinary-Level Cell, i.e. 3bit/Cell).The performance SLC of above-mentioned three kinds of media
Preferably, MLC takes second place, and TLC is poor；But SLC highest in terms of the cost of unit memory capacity, MLC take second place, TLC is minimum.In the past,
SSD mainly uses homostyructure, i.e., constructs SSD using same flash media.It now, is the compromise for realizing performance and cost,
Start the SSD for heterojunction structure occur, i.e., the physical store of bottom is constructed using different storage mediums, such as SLC and MLC, SLC
SSD is mixed with TLC building.
The present invention is directed to the mixing SSD, i.e. the storage medium A and another kind of better performances constructed using two kinds of storage mediums
(such as A is SLC to the storage medium B that performance is taken second place, and B is MLC or TLC；Or A is MLC, B TLC), it discloses a kind of towards mixed
The flash translation layer (FTL) and its conversion method of solid state hard disk are closed, compromise of the mixing SSD between performance and cost is better achieved.
Summary of the invention
The purpose of the present invention is in view of the drawbacks of the prior art, provide a kind of flash memory conversion towards hybrid solid-state hard disk
Layer and its conversion method.Not only guarantee that the high-performance for mixing SSD and low cost are traded off, but also realize the abrasion between medium A and medium B
Equilibrium extends the service life of mixing SSD.
In order to achieve the goal above, the invention adopts the following technical scheme:
A kind of flash translation layer (FTL) towards hybrid solid-state hard disk, the hybrid solid-state hard disk are made of two kinds of flash medias,
The performance of wherein medium A better performances but unit capacity higher cost, medium B wants poor with respect to A, but cost is relatively low for unit capacity；
Medium A is divided into data block area and translation block area, all data block areas medium B；Translation block area is used to store translation page, data block
Area is used to storing data page；Translate the mapping relations that page is used between the logical address and physical address of storing data, data page
For storing real data；It translates page to be aligned according to the page-size of medium A, data page is aligned according to the page-size of medium B；
The flash translation layer (FTL) includes an address of cache buffer area, and the address of cache buffer area is by global translation directory GTD, mapping
Caching of page slot TPCS, hot showing penetrate cache table Hot-CMT and cold mapped cache table Cold-CMT composition；The GTD is every for recording
The actual physics page number of a translation page when the TPCS is not hit by the buffer for currently mapping item, caches current load and reflects
Penetrate entire mapping page belonging to item, the Hot-CMT be used to cache frequently access write mapping item, the Cold-CMT is for delaying
It deposits reading mapping item and what is infrequently accessed writes mapping item；The logical page number (LPN) of the Hot-CMT and Cold-CMT record mapping item
LPN, physical page number PPN.
Further, the flash translation layer (FTL) uses page grade address of cache mode.
Further, the present invention proposes a kind of conversion method of flash translation layer (FTL) towards hybrid solid-state hard disk, for
The flash translation layer (FTL) stated, includes the following steps:
S1 judges request type when access request arrives；If read request, S2 is executed, if write request then executes
S6；
Whether S2 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for read request；If reflecting
It penetrates item and is located at Hot-CMT, execute S5；If mapping item is located in Cold-CMT or TPCS, S3 is executed；Otherwise, mapping item does not exist
It caches in mapping table, then executes S4；
The mapping item is loaded into position MRU (Most Recently Used) of Cold-CMT, if Cold- at this time by S3
CMT size is greater than the targets threshold of setting, then starts the rejecting operation of Cold-CMT, then execute S5；
S4 accesses GTD according to the logical page number (LPN) of request, obtains the mapping item when mapping item not when caching in mapping table
The physical page number of corresponding translation page, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number,
And the mapping item is loaded into the MRU position of Cold-CMT simultaneously, if Cold-CMT size is greater than the targets threshold of setting at this time,
The rejecting operation for then starting Cold-CMT, finally executes S5；
S5 returns to physical page number corresponding to the mapping item, address of cache conversion end；
Whether S6 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for write request；If reflecting
Item is penetrated in Hot-CMT, Cold-CMT or TPCS hit then execute S7；Otherwise, mapping item executes not in caching mapping table
S8；
The mapping item is moved to the MRU position of Hot-CMT if hitting in Hot-CMT by S7；If being ordered in Cold-CMT
In, mapping item is also moved into the MRU position in Hot-CMT, if Hot-CMT size is greater than the targets threshold of setting at this time,
Start the rejecting operation of Hot-CMT；If hitting in TPCS, which is moved to the MRU position of Cold-CMT, if at this time
Cold-CMT size is greater than the targets threshold of setting, then starts the rejecting operation of Cold-CMT；Finally execute S9；
S8 accesses GTD according to the logical page number (LPN) of request, obtains the mapping item when mapping item not when caching in mapping table
The physical page number of corresponding translation page, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number,
And the mapping item is loaded into the MRU position of Cold-CMT simultaneously；If Cold-CMT size is greater than the targets threshold of setting at this time,
Then start the rejecting operation of Cold-CMT；Finally execute S9；
S9 adjusts the size of Hot-CMT and Cold-CMT cache table according to the rate of wear of medium A and B；It then is to write
Request distributes new free physical data page number, and new physical data page number is write in corresponding caching mapping table；Finally return
Return the pervious physical page number of mapping item and newly assigned free physical page number.
Further, the least recently used LRU that step S3, the Cold-CMT in S4, S7 and S8 uses clean item preferential
Rejecting mechanism: the fixed clean item of one length is set in the tail portion of Cold-CMT queue and preferentially rejects window；Work as Cold-CMT
When carrying out rejecting operation, since finding not more new mappings item in preferentially rejecting window LRU position, if can find,
Directly reject this clean mapping item；Otherwise, then the more new mappings item on LRU position is selected, by its information back to translation
It is rejected after page.
Further, the Hot-CMT in step S7 rejects mechanism using LRU, that is, directly selects Hot-CMT queue LRU bit
The mapping item set will reject after its information back to translation page.
Further, the adjustment Hot-CMT and Cold-CMT caches table size specifically:
It is consistent as far as possible come the rate of wear for guaranteeing them by the ratio α for controlling Hot-CMT and Cold-CMT；α is by medium A
It is controlled with the relative wear rate φ of medium B, the method for adjustment of α is as follows:
S21, calculation medium A and B current relative wear rate φ；
S22 compares φ and φtSize, φtTo preset relative wear balancing speed threshold value, if φ < φt, then it is assumed that when
Preceding medium A and medium B abrasion equilibrium, without adjusting α；Otherwise, following S23 is executed；
S23, if RWA>RWB, medium A reduces ratio value α=α-△ α of Hot-CMT, than medium B quick abrasion to subtract at this time
The data volume of medium A is written less；Conversely, then medium B increases ratio value α=α+△ α of Hot-CMT than medium A quick abrasion, come
Reduce the data volume of write-in medium B.
Further, φ is defined as:
Wherein, max () represents maximizing, and min () representative is minimized；RWAIndicate that medium A normalizes to medium
The average wear rate of B, RWBIndicate the average wear rate of medium B.
Further, new free physical data page number is distributed for write request in step S9 specifically:
Data block zoning is divided into hot data block area and cold data block area, i.e. the data in hot data block area are hot number
According to being stored in medium A；The data in hot data block area are not cold data, are stored in medium B；In data write-in, reflect
Penetrating page of the item in Hot-CMT is considered as dsc data, and distribution one belongs to the idle page number of medium A；Conversely, being then considered cold
Data, distribution one belong to the idle page number of medium B.
Further, when free block deficiency in medium A, the garbage reclamation mechanism of medium A is triggered, and operating process is as follows:
S31 selects the block with most invalid pages as recycling block, and then judgement recycling block whether there is valid data
Page；Valid data page then executes S32 if it exists, otherwise executes S35；
S32, successively judges whether the mapping item of each valid data page hits in Hot-CMT, executes if hit
Otherwise S33 executes S34；
S33, the data are still dsc data, which is moved to the free block of medium A, and modify its corresponding mapping
Item information；
S34, the data are cold data, which is moved to the free block of medium B, and modify its corresponding mapping letter
Breath；
Recycling block erasing is terminated garbage collection operation by S35；
Further, when free block deficiency in medium B, the garbage reclamation mechanism of medium B is triggered, and selection has most nothings
Then the block for imitating page will recycle free block of the valid data page migration in block to medium B as recycling block, and update the mapping
The mapping relations of item；Finally erasing recycling block, terminates garbage collection operation.
The present invention proposes a kind of flash translation layer (FTL) design method towards mixing SSD, solves and is constructed using a variety of media
The flash translation layer (FTL) design problem of SSD.On the one hand, the translation page in medium A and medium B is aligned according to the page size of medium A, real
The unified management of mapping item is showed.On the other hand, hot write request and other requests are distinguished using cold and hot mapped cache, so that hot write
Request is assigned to the good medium A of performance, and other write requests are assigned to medium B, promotes the performance of mixing SSD, reduces its unit appearance
Measure cost；Meanwhile according to the rate of wear of medium A and medium B, adjust the ratio of cold and hot mapped cache table, preferably realize matter A and
Abrasion equilibrium between medium B extends the service life of mixing SSD.The method that present aspect is announced has in mixing SSD design
Practicability and preferable application prospect.
Detailed description of the invention
It in order to illustrate the embodiments of the present invention more clearly or the technical solution of the prior art, below will be to embodiment or existing
The required attached drawing of technical description is briefly described, it should be apparent that, the accompanying drawings in the following description is only the present invention
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1: general frame figure of the invention.
Fig. 2: address mapping structures schematic diagram of the invention.
Fig. 3: address conversion flow chart of the invention.
Fig. 4: address conversion implementation example figure of the invention.
Fig. 5: friction equilibrium and the adjustment signal of cold and hot mapping table size adaptation between medium A and medium B of the invention
Figure.
Fig. 6: write request data distribution of the invention and garbage reclamation flow chart.
Specific embodiment
Illustrate embodiments of the present invention below by way of specific specific example, those skilled in the art can be by this specification
Other advantages and efficacy of the present invention can be easily understood for disclosed content.The present invention can also pass through in addition different specific realities
The mode of applying is embodied or practiced, the various details in this specification can also based on different viewpoints and application, without departing from
Various modifications or alterations are carried out under spirit of the invention.It should be noted that in the absence of conflict, following embodiment and implementation
Feature in example can be combined with each other.
It should be noted that illustrating the basic structure that only the invention is illustrated in a schematic way provided in following embodiment
Think, only shown in schema then with related component in the present invention rather than component count, shape and size when according to actual implementation
Draw, when actual implementation kenel, quantity and the ratio of each component can arbitrarily change for one kind, and its assembly layout kenel
It is likely more complexity.
The present invention will be further explained below with reference to the attached drawings and specific examples, but not as the limitation of the invention.
Embodiment one
The present embodiment proposes a kind of flash translation layer (FTL) towards hybrid solid-state hard disk, and the hybrid solid-state hard disk is by two kinds
Flash media is constituted, wherein medium A better performances but unit capacity higher cost, and the performance of medium B wants poor with respect to A, but unit
Capacity Cost is lower.Further, medium A is divided into data block area and translation block area, all data block areas medium B.Translate block area
For storing translation page, data block area is used to storing data page.Translate logical address and physical address that page is used to storing data
Between mapping relations, data page be used to store real data.It translates page to be aligned according to the page-size of medium A, data page is pressed
It is aligned according to the page-size of medium B (in general, different storage mediums has different page-sizes, the page-size of medium A
Less than the page-size of medium B, and meet integral multiple relation).
For example, i.e. SLC is upper as shown in Figure 1, mixing building SSD using two different storage medium SLC and MLC
The medium A in patent is stated, MLC is the medium B in above-mentioned patent.SLC is divided for data block area and translates block area, MLC is all
Data block area, dsc data page and translation page are stored in SLC, and cold data page is stored in MLC.The translation page size of request is pressed
It is aligned according to 2KB (the page size of SLC), data page size is aligned according to 4KB (the page size of MLC).By buffer area be divided into GTD,
Tetra- parts TPCS, Hot-CMT and Cold-CMT, wherein the proportionality coefficient of Hot-CMT and Cold-CMT can according to SLC and
The relative wear rate adaptation of MLC adjusts.
Flash translation layer (FTL) uses page grade address of cache mode, and includes an address of cache buffering in flash translation layer (FTL)
Area.Further, address of cache buffering divides into GTD (global translation directory), TPCS (mapping page cache slot), Hot-CMT (hot showing
Penetrate cache table) and Cold-CMT (cold mapped cache table).GTD is used to record the actual physics page number of each translation page, and TPCS is used
Entire mapping page (just exciting when only currently mapping item is not hit by the buffer) belonging to item is mapped to cache current load,
Hot-CMT be used to cache frequently access write mapping item, Cold-CMT, which is used to cache, reads mapping item and writing of infrequently accessing is reflected
Penetrate item.The logical page number (LPN) (LPN) of the Hot-CMT and Cold-CMT record mapping item, physical page number (PPN).
As shown in Fig. 2, including 4 tables in flash translation layer (FTL), it is respectively as follows: GTD, Hot-CMT, Cold-CMT and TPCS.GTD
For global map table, for recording the actual physics page number of each translation page.TVPN in GTD is virtual translation page number, is not had to
Storage.Any corresponding virtual translation page number of logical page (LPAGE) can use formulaIt obtains, LPN is here
Logical page number (LPN), NUM are the mapping item number that each translation page can store,Represent lower bracket function.Any logical page (LPAGE) is right at its
Answer the position in virtual translation page that can be obtained with formula No=Mod (LPN, NUM), LPN and NUM are same as above here, and Mod representative takes
Cofunction.Hot-CMT is that cache table is penetrated in hot showing, the mapping item for the higher write request of cache access frequency.It writes and asks when one
It asks in Cold-CMT hit, is identified as hot write, is moved to the MRU position of Hot-CMT；When Hot-CMT is full, adopt
It rejects strategy with LRU the mapping item infrequently accessed is written back in the translation page of bottom flash memory.Cold-CMT is that cold mapping is slow
Table is deposited, the mapping item and read request of the write request low for cache access frequency map item；When Cold-CMT is full, using clean
The preferential LRU of item rejects strategy.TPCS is mapping page cache slot, for caching entire mapping page belonging to current load mapping item
(just being excited when only currently mapping item is not hit by the buffer).In bottom flash memory storage layer, since the read-write of translation page updates
Compare frequently, so place it in the preferable SLC of readwrite performance, and by dsc data page (the corresponding data page of Hot-CMT)
It is stored in SLC, cold data page is stored in MLC.In addition, the size of Hot-CMT and Cold-CMT can according to SLC and
The relative wear rate of MLC dynamically adjusts ratio.
Embodiment two
As shown in figure 3, the present embodiment proposes a kind of hard towards hybrid solid-state based on the flash translation layer (FTL) in embodiment one
The conversion method of the flash translation layer (FTL) of disk, specifically includes:
S1 judges request type when access request arrives.If read request, S2 is executed, if write request then executes
S6。
Whether S2 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for read request.If reflecting
It penetrates item and is located at Hot-CMT, execute S5；If mapping item is located in Cold-CMT or TPCS, S3 is executed；Otherwise, mapping item does not exist
It caches in mapping table, then executes S4.
The mapping item is loaded into position MRU (Most Recently Used) of Cold-CMT, if Cold- at this time by S3
CMT size is greater than the targets threshold of setting, then starts the rejecting operation of Cold-CMT, then execute S5.
S4 accesses GTD according to the logical page number (LPN) of request, obtains the mapping item when mapping item not when caching in mapping table
The physical page number of corresponding translation page, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number,
And the mapping item is loaded into the MRU position of Cold-CMT simultaneously, if Cold-CMT size is greater than the targets threshold of setting at this time,
Then start the rejecting operation of Cold-CMT.Finally execute S5.
S5 returns to physical page number corresponding to the mapping item, address of cache conversion end.
Whether S6 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for write request.If reflecting
Item is penetrated in Hot-CMT, Cold-CMT or TPCS hit then execute S7；Otherwise, mapping item executes not in caching mapping table
S8。
The mapping item is moved to the MRU position of Hot-CMT if hitting in Hot-CMT by S7；If being ordered in Cold-CMT
In, mapping item is also moved into the MRU position in Hot-CMT, if Hot-CMT size is greater than the targets threshold of setting at this time,
Start the rejecting operation of Hot-CMT；If hitting in TPCS, which is moved to the MRU position of Cold-CMT, if at this time
Cold-CMT size is greater than the targets threshold of setting, then starts the rejecting operation of Cold-CMT.Finally execute S9.
S8 accesses GTD according to the logical page number (LPN) of request, obtains the mapping item when mapping item not when caching in mapping table
The physical page number of corresponding translation page, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number,
And the mapping item is loaded into the MRU position of Cold-CMT simultaneously；If Cold-CMT size is greater than the targets threshold of setting at this time,
Then start the rejecting operation of Cold-CMT.Finally execute S9.
S9 adjusts the size of Hot-CMT and Cold-CMT cache table according to the rate of wear of medium A and B；It then is to write
Request distributes new free physical data page number, and new physical data page number is write in corresponding caching mapping table；Finally return
Return the pervious physical page number of mapping item and newly assigned free physical page number.
Step S3, least recently used (LRU) the rejecting machine Cold-CMT in S4, S7 and S8 preferential using clean item
System.Concrete methods of realizing is as follows: the fixed clean item of a length is arranged in the tail portion of Cold-CMT queue and preferentially rejects window；
When Cold-CMT carries out rejecting operation, reflected since finding clean (not updating) in preferentially rejecting window LRU position
Penetrate item.If can find, this clean mapping item is directly rejected；Otherwise, then dirty (update) on LRU position is selected to map
, it will be rejected after its information back to translation page.
Hot-CMT in step S7 rejects mechanism using LRU, that is, directly selects the mapping on Hot-CMT queue LRU position
, it will be rejected after its information back to translation page.
In Fig. 4, it has been assumed that the logical address space for the mixing SSD being made of SLC and MLC is 32 data pages, and SLC
One Physical Page is only capable of 4 mapping items of storage.In addition, not involving Hot-CMT and Cold- during having also assumed that address conversion
The size of CMT adjusts.It should be pointed out that the VPN and VTPN in Fig. 4 do not account for memory space actually, effect is only retouched for convenience
It states.
The thought of page grade address of cache according to the present invention, 32 data pages 32 mapping items of needs, and each Physical Page of SLC
4 mapping items of storage are only capable of, then whole mapping table storages need the Physical Page of 8 SLC.Further, it is contemplated that flash memory is not supported
The particularity that original place updates, the practical storage for distributing 12 pages in SLC and being used for whole mapping tables, number T0-T11.Assuming that
The physical space that SLC, MLC there remains 48 data pages is stored for user data, number D0-D47, wherein first 8 are located at
SLC, latter 40 are located at MLC.The space User Data of each data page is used to store true user data, and the space OOB is used to
The metadata for storing each page, such as the state (invalidating/free time) of current page, the logical page number (LPN) of corresponding data, error checking and correction
Data etc..According to above-mentioned it is assumed that Fig. 4 (a) provides each mapping table of flash translation layer (FTL) and bottom flash parts translate the initial of page
Status diagram.In Fig. 4, it is 2 that the clean item of Cold-CMT, which preferentially rejects window, i.e. the mapping item of tail portion two is that clean item is preferential
Reject window；
Shade item in Hot-CMT and Cold-CMT is the mapping item for occurring to update.
It is now assumed that needing to handle following request in order: LPN=7 (read request), LPN=1 (read request), LPN=3
(write request), LPN=5 (write request), LPN=6 (write request).The method announced according to the present invention, shown in Fig. 4 (a) just
Under beginning state, address translation process and result are as follows:
C1, LPN=7 read request, the mapping item are hit in Cold-CMT, which is moved to Cold-
The MRU position of CMT returns to the PPN=D22 of the request.
C2, LPN=1 read request, the mapping item are hit in TPCS, which is loaded into Cold-CMT's
MRU position.At this point, Cold-CMT has expired, item is completely mapped since finding in preferentially rejecting window LRU position, due to (8,
D23 it is) clean mapping item, selects it as item is rejected and directly reject, then return to the PPN=D18 of the request.
C3, LPN=3 write request hit in Hot-CMT, which are moved to the MRU position of Hot-CMT.
Since the write request is hit in Hot-CMT, identifies that the request is hot write, therefore distribute the idle object for belonging to SLC for the request
Data page number PPN=D3 is managed, and data page D7 before is set in vain, mapping relations (3, D7) are updated to (3, D3), most
After return to the pervious physical page number PPN=D7 of the mapping item and newly assigned free physical page number PPN=D3.
C4, LPN=5 write request are not hit in mapped cache table, by inquiring GTD, obtain the corresponding translation of the mapping item
The physical page number TPPN=T1 of page, then loads the translation page into TPCS, and request mapping item (5, D35) is loaded into Cold-
The MRU position of CMT.At this point, Cold-CMT has expired, item is completely mapped since finding in preferentially rejecting window LRU position, this
When due to (9, D24) be clean mapping item, select it directly to reject as rejecting item.Due to this write request be not in Hot-CMT,
Therefore one is distributed for the request and belongs to the free physical data page number PPN=D36 of MLC, and data page D35 before is set to nothing
Mapping relations (5, D35) are updated to (5, D36) by effect, finally return to the pervious physical page number PPN=D35 of the mapping item and new
The free physical page number PPN=D36 of distribution.
C5, LPN=6 write request hit in Cold-CMT, which are moved to the position MRU of Hot-CMT
It sets.At this point, Hot-CMT has expired, the mapping item (0, D0) on LRU position is selected, by the information back of this mapping item to new translation
Rejected after page T8, then distribute the free physical data page number PPN=D4 in new SLC Flash area for the request, and by it
Preceding data page D21 is set in vain, and mapping relations (6, D21) are updated to (6, D4), finally returns to the pervious object of mapping item
Manage page number PPN=D21 and newly assigned free physical page number PPN=D4.
The rate of wear according to medium A and B in step S9 adjusts the machine of Hot-CMT and Cold-CMT caching table size
It makes as follows:
For the abrasion equilibrium for realizing medium A and B, the present invention is guaranteed by controlling the ratio α of Hot-CMT and Cold-CMT
Their rate of wear is consistent as far as possible.Specifically, α is controlled by medium A and the relative wear rate φ of medium B.φ is defined as:
In above formula, max () represents maximizing, and min () representative is minimized；RWAIndicate that medium A normalizes to Jie
The average wear rate of matter B, RWBIndicate the average wear rate of medium B.
The method of adjustment of α is as follows:
S21, calculation medium A and B current relative wear rate φ.
S22 compares φ and φtSize (φtTo preset relative wear balancing speed threshold value).If φ < φt, then it is assumed that when
Preceding medium A and medium B abrasion equilibrium, without adjusting α；Otherwise, following S23 is executed.
S23, if RWA>RWB, medium A reduces ratio value α=α-△ α of Hot-CMT, than medium B quick abrasion to subtract at this time
The data volume of medium A is written less；Conversely, then medium B increases ratio value α=α+△ α of Hot-CMT than medium A quick abrasion, come
Reduce the data volume of write-in medium B.
As shown in figure 5, the method for the present invention controlled by dynamically adjusting the size of cold and hot mapped cache table SLC and MLC it
Between abrasion equilibrium.Hot-CMT ratio value α is controlled by the relative wear rate Φ of SLC and MLC, in the present embodiment setting mill
Damage trimming threshold Φtα=0.05=1.3, Hot-CMT proportionality coefficient initial value α=0.4, △.
Calculate current relative wear rate value Φ0If Φ0< 1.3, then it is assumed that current SLC and MLC abrasion equilibrium is not necessarily to
Adjust α.
If Φ >=1.3, determines that current SLC and MLC abrasion is unbalanced, starts cold and hot mapped cache table dynamic adjustment mechanism,
When the SLC rate of wear faster when, i.e. RWSLC>RWMLC, need to reduce ratio value α=α-△ α of Hot-CMT, to reduce write-in SLC
Data volume, while increasing the data volume that SLC is migrated into MLC.When the MLC rate of wear faster when, i.e. RWSLC<RWMLC, need
Increase ratio value α=α+△ α of Hot-CMT, to reduce the data volume of write-in MLC, while increasing SLC and retaining dsc data
Ability.Relative wear rate is controlled again by adjusting the ratio value α of Hot-CMT, makes abrasion equilibrium between SLC and MLC.
It is that the mechanism that write request distributes new free physical data page number is as follows in step S9:
Data block zoning is divided into hot data block area and cold data block area by method proposed by the present invention, i.e., in hot data block area
In data be dsc data, be stored in medium A；The data in hot data block area are not cold data, are stored in medium B.
In data write-in, mapping page of the item in Hot-CMT is considered as dsc data, and distribution one belongs to the idle page number of medium A；
Conversely, being then considered cold data, distribution one belongs to the idle page number of medium B.
For example, as shown in Fig. 6 (a).The process of write request data distribution can be by the mapping item information pair of Hot-CMT
Data carry out cold and hot identification.In data write-in, belong to SLC's to mapping item corresponding data page distribution one in Hot-CMT
Idle data page number will map page of the item in Hot-CMT and be identified as dsc data page, is stored in high performance dielectric SLC.Instead
It, distributes the idle data page number for belonging to MLC to the corresponding data page of mapping item not in Hot-CMT, i.e., will not exist
Page in Hot-CMT is identified as cold data page, is stored in low performance medium MLC.
Shown in garbage reclamation mechanism of the invention such as Fig. 6 (b).The method of the present invention passes through valid data page in inquiry recycling block
Corresponding map information carries out garbage reclamation, if the mapping item of the data page is located in Hot-CMT, shows the data page still
For dsc data, which is copied in the freed data blocks of SLC, data block area on the contrary then that data are migrated into MLC.
When the freed data blocks deficiency in SLC, the garbage reclamation mechanism of SLC can be triggered, and carry out garbage collection operation,
Operating process is as follows:
C1 selects the block with most invalid pages as recycling block, and then judgement recycling block whether there is valid data page,
Valid data page then executes C2 if it exists, otherwise executes C5.
C2, successively judges whether the mapping item for recycling each valid data page of block hits in Hot-CMT, if hit
C3 is executed, C4 is otherwise executed.
C3, the data are still dsc data, which is moved to the free block of SLC, and are modified corresponding in Hot-CMT
Map item information.
C4, the data are cold data, which are moved to the free block of MLC, and modify its corresponding map information.
Recycling block erasing is terminated garbage collection operation by C5.
When the free block deficiency in MLC, the garbage reclamation mechanism of MLC is triggered, and selects the block with most invalid pages
As recycling block, free block of the valid data page migration in block to MLC then will be recycled, and the mapping for updating the mapping item is closed
System, finally erasing recycling block, terminates garbage collection operation.
The present invention proposes a kind of flash translation layer (FTL) design method towards mixing SSD, solves and is constructed using a variety of media
The flash translation layer (FTL) design problem of SSD.On the one hand, the translation page in medium A and medium B is aligned according to the page size of medium A, real
The unified management of mapping item is showed.On the other hand, hot write request and other requests are distinguished using cold and hot mapped cache, so that hot write
Request is assigned to the good medium A of performance, and other write requests are assigned to medium B, promotes the performance of mixing SSD, reduces its unit appearance
Measure cost；Meanwhile according to the rate of wear of medium A and medium B, adjust the ratio of cold and hot mapped cache table, preferably realize matter A and
Abrasion equilibrium between medium B extends the service life of mixing SSD.The method that present aspect is announced has in mixing SSD design
Practicability and preferable application prospect.
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (10)
1. a kind of flash translation layer (FTL) towards hybrid solid-state hard disk, the hybrid solid-state hard disk is made of two kinds of flash medias,
The performance of middle medium A better performances but unit capacity higher cost, medium B wants poor with respect to A, but cost is relatively low for unit capacity；It is situated between
Matter A points are data block area and translation block area, all data block areas medium B；Translation block area is used to store translation page, data block area
For storing data page；The mapping relations that page is used between the logical address and physical address of storing data are translated, data page is used
To store real data；It translates page to be aligned according to the page-size of medium A, data page is aligned according to the page-size of medium B；Its
It is characterized in that,
The flash translation layer (FTL) include an address of cache buffer area, the address of cache buffer area by global translation directory GTD,
Mapping page cache slot TPCS, hot showing penetrate cache table Hot-CMT and cold mapped cache table Cold-CMT composition；The GTD is for remembering
The actual physics page number for recording each translation page, when the TPCS is not hit by the buffer for currently mapping item, caching it is current plus
Carry entire mapping page belonging to mapping item, the Hot-CMT be used to cache frequently access write mapping item, the Cold-CMT is used
Mapping item is read in caching and what is infrequently accessed writes mapping item；The logical page (LPAGE) of the Hot-CMT and Cold-CMT record mapping item
Number LPN, physical page number PPN.
2. flash translation layer (FTL) according to claim 1, which is characterized in that the flash translation layer (FTL) uses page grade address of cache
Mode.
3. a kind of conversion method of the flash translation layer (FTL) towards hybrid solid-state hard disk, described in any item for claim 1-2
Flash translation layer (FTL), which comprises the steps of:
S1 judges request type when access request arrives；If read request, S2 is executed, if write request then executes S6；
Whether S2 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for read request；If mapping item
Positioned at Hot-CMT, S5 is executed；If mapping item is located in Cold-CMT or TPCS, S3 is executed；Otherwise, mapping item is not caching
In mapping table, then S4 is executed；
The mapping item is loaded into position MRU (Most Recently Used) of Cold-CMT, if Cold-CMT is big at this time by S3
The small targets threshold for being greater than setting then starts the rejecting operation of Cold-CMT, then executes S5；
S4 accesses GTD according to the logical page number (LPN) of request when mapping item not when caching in mapping table, and it is right to obtain the mapping item institute
The physical page number for the translation page answered, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number, and will
The mapping item is loaded into the MRU position of Cold-CMT simultaneously, if Cold-CMT size is greater than the targets threshold of setting at this time, opens
The rejecting operation of dynamic Cold-CMT, finally executes S5；
S5 returns to physical page number corresponding to the mapping item, address of cache conversion end；
Whether S6 checks request mapping item in Hot-CMT, Cold-CMT and TPCS in order for write request；If mapping item
In Hot-CMT, Cold-CMT or TPCS hit then execute S7；Otherwise, mapping item executes S8 not in caching mapping table；
The mapping item is moved to the MRU position of Hot-CMT if hitting in Hot-CMT by S7；If being hit in Cold-CMT,
Mapping item is moved into the MRU position in Hot-CMT, if Hot-CMT size is greater than the targets threshold of setting at this time, is started
The rejecting of Hot-CMT operates；If hitting in TPCS, which is moved to the MRU position of Cold-CMT, if at this time
Cold-CMT size is greater than the targets threshold of setting, then starts the rejecting operation of Cold-CMT；Finally execute S9；
S8 accesses GTD according to the logical page number (LPN) of request when mapping item not when caching in mapping table, and it is right to obtain the mapping item institute
The physical page number for the translation page answered, and the translation page comprising target mapping item is loaded into TPCS according to this physical page number, and will
The mapping item is loaded into the MRU position of Cold-CMT simultaneously；If Cold-CMT size is greater than the targets threshold of setting at this time, open
The rejecting operation of dynamic Cold-CMT；Finally execute S9；
S9 adjusts the size of Hot-CMT and Cold-CMT cache table according to the rate of wear of medium A and B；It then is write request
New free physical data page number is distributed, and new physical data page number is write in corresponding caching mapping table；Finally return to this
Map the pervious physical page number of item and newly assigned free physical page number.
4. conversion method according to claim 3, which is characterized in that step S3, the Cold-CMT in S4, S7 and S8 is used
The preferential least recently used LRU of clean item rejects mechanism: in the tail portion of Cold-CMT queue, one length of setting is fixed dry
Net item preferentially rejects window；When Cold-CMT carry out reject operation when, since LRU position preferentially reject window in seek
Not more new mappings item is looked for, if can find, directly rejects this clean mapping item；Otherwise, then the update on LRU position is selected
Item is mapped, will be rejected after its information back to translation page.
5. conversion method according to claim 3, which is characterized in that the Hot-CMT in step S7 uses LRU rejecting machine
System, that is, directly select the mapping item on Hot-CMT queue LRU position, will reject after its information back to translation page.
6. conversion method according to claim 3, which is characterized in that the adjustment Hot-CMT and Cold-CMT cache table
Size specifically:
It is consistent as far as possible come the rate of wear for guaranteeing them by the ratio α for controlling Hot-CMT and Cold-CMT；α is by medium A and is situated between
The relative wear rate φ of matter B is controlled, and the method for adjustment of α is as follows:
S21, calculation medium A and B current relative wear rate φ；
S22 compares φ and φtSize, φtTo preset relative wear balancing speed threshold value, if φ < φt, then it is assumed that it is current to be situated between
Matter A and medium B abrasion equilibrium, without adjusting α；Otherwise, following S23 is executed；
S23, if RWA>RWB, medium A reduces ratio value α=α-△ α of Hot-CMT, writes to reduce than medium B quick abrasion at this time
Enter the data volume of medium A；Conversely, then medium B increases ratio value α=α+△ α of Hot-CMT, than medium A quick abrasion to reduce
The data volume of medium B is written.
7. conversion method according to claim 6, which is characterized in that φ is defined as:
Wherein, max () represents maximizing, and min () representative is minimized；RWAIndicate that medium A normalizes to the flat of medium B
Equal rate of depreciation, RWBIndicate the average wear rate of medium B.
8. conversion method according to claim 3, which is characterized in that distribute new free physical in step S9 for write request
Data page number specifically:
Data block zoning is divided into hot data block area and cold data block area, i.e. the data in hot data block area are dsc data, are deposited
Storage is in medium A；The data in hot data block area are not cold data, are stored in medium B；In data write-in, item is mapped
Page in Hot-CMT is considered as dsc data, and distribution one belongs to the idle page number of medium A；Conversely, being then considered cold number
According to distribution one belongs to the idle page number of medium B.
9. conversion method according to claim 3, which is characterized in that when free block deficiency in medium A, the rubbish of medium A
The triggering of rubbish reclaim mechanism, operating process are as follows:
S31 selects the block with most invalid pages as recycling block, and then judgement recycling block whether there is valid data page；If
There are effective data pages then to execute S32, otherwise executes S35；
S32, successively judges whether the mapping item of each valid data page hits in Hot-CMT, executes S33 if hit, no
Then execute S34；
S33, the data are still dsc data, which is moved to the free block of medium A, and modify its corresponding mapping item letter
Breath；
S34, the data are cold data, which are moved to the free block of medium B, and modify its corresponding map information；
Recycling block erasing is terminated garbage collection operation by S35.
10. conversion method according to claim 3, which is characterized in that when free block deficiency in medium B, the rubbish of medium B
The triggering of rubbish reclaim mechanism selects the block with most invalid pages as recycling block, then moves the valid data page recycled in block
The free block of medium B is moved on to, and updates the mapping relations of the mapping item；Finally erasing recycling block, terminates garbage collection operation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910675390.5A CN110413537B (en) | 2019-07-25 | 2019-07-25 | Flash translation layer facing hybrid solid state disk and conversion method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910675390.5A CN110413537B (en) | 2019-07-25 | 2019-07-25 | Flash translation layer facing hybrid solid state disk and conversion method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110413537A true CN110413537A (en) | 2019-11-05 |
| CN110413537B CN110413537B (en) | 2021-08-24 |
Family
ID=68363092
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910675390.5A Active CN110413537B (en) | 2019-07-25 | 2019-07-25 | Flash translation layer facing hybrid solid state disk and conversion method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110413537B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111258924A (en) * | 2020-01-17 | 2020-06-09 | 中国科学院国家空间科学中心 | Mapping method based on satellite-borne solid-state storage system self-adaptive flash translation layer |
| CN111506517A (en) * | 2020-03-05 | 2020-08-07 | 杭州电子科技大学 | Flash memory page level address mapping method and system based on access locality |
| CN112000296A (en) * | 2020-08-28 | 2020-11-27 | 北京计算机技术及应用研究所 | Performance optimization system in full flash memory array |
| CN112506445A (en) * | 2020-12-29 | 2021-03-16 | 杭州电子科技大学 | Partition proportion self-adaptive adjusting method of homogeneous hybrid solid state disk |
| CN113220241A (en) * | 2021-05-27 | 2021-08-06 | 衢州学院 | Cross-layer design-based hybrid SSD performance and service life optimization method |
| CN113435109A (en) * | 2021-06-04 | 2021-09-24 | 衢州学院 | Performance and service life optimization method of hybrid SSD |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102122531A (en) * | 2011-01-27 | 2011-07-13 | 浪潮电子信息产业股份有限公司 | Method for improving stability in use of large-capacity solid state disk |
| US20120079232A1 (en) * | 2010-09-24 | 2012-03-29 | Glenn Hinton | Apparatus, method, and system for implementing micro page tables |
| CN103440206A (en) * | 2013-07-25 | 2013-12-11 | 记忆科技(深圳)有限公司 | Solid state hard disk and mixed mapping method of solid state hard disk |
| US20150095546A1 (en) * | 2013-10-02 | 2015-04-02 | Alan Bennett | System and method for bank logical data remapping |
| CN106776375A (en) * | 2016-12-27 | 2017-05-31 | 东方网力科技股份有限公司 | Data cache method and device inside a kind of disk |
| US20180113620A1 (en) * | 2016-10-24 | 2018-04-26 | SK Hynix Inc. | Memory system and operation method thereof |
| CN109446117A (en) * | 2018-09-06 | 2019-03-08 | 杭州电子科技大学 | A kind of solid state hard disk page grade flash translation layer (FTL) design method |
| CN109739780A (en) * | 2018-11-20 | 2019-05-10 | 北京航空航天大学 | Dynamic secondary based on the mapping of page grade caches flash translation layer (FTL) address mapping method |
-
2019
- 2019-07-25 CN CN201910675390.5A patent/CN110413537B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120079232A1 (en) * | 2010-09-24 | 2012-03-29 | Glenn Hinton | Apparatus, method, and system for implementing micro page tables |
| CN102122531A (en) * | 2011-01-27 | 2011-07-13 | 浪潮电子信息产业股份有限公司 | Method for improving stability in use of large-capacity solid state disk |
| CN103440206A (en) * | 2013-07-25 | 2013-12-11 | 记忆科技(深圳)有限公司 | Solid state hard disk and mixed mapping method of solid state hard disk |
| US20150095546A1 (en) * | 2013-10-02 | 2015-04-02 | Alan Bennett | System and method for bank logical data remapping |
| US20180113620A1 (en) * | 2016-10-24 | 2018-04-26 | SK Hynix Inc. | Memory system and operation method thereof |
| CN106776375A (en) * | 2016-12-27 | 2017-05-31 | 东方网力科技股份有限公司 | Data cache method and device inside a kind of disk |
| CN109446117A (en) * | 2018-09-06 | 2019-03-08 | 杭州电子科技大学 | A kind of solid state hard disk page grade flash translation layer (FTL) design method |
| CN109739780A (en) * | 2018-11-20 | 2019-05-10 | 北京航空航天大学 | Dynamic secondary based on the mapping of page grade caches flash translation layer (FTL) address mapping method |
Non-Patent Citations (2)
| Title |
|---|
| ILHOON SHIN: "Hot/cold clustering for page mapping in NAND flash memory", 《IEEE TRANSACTIONS ON CONSUMER ELECTRONICS》 * |
| 姚英彪等: "一种基于分类策略的聚簇页级闪存转换层算法", 《计算机研究与发展》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111258924A (en) * | 2020-01-17 | 2020-06-09 | 中国科学院国家空间科学中心 | Mapping method based on satellite-borne solid-state storage system self-adaptive flash translation layer |
| CN111258924B (en) * | 2020-01-17 | 2021-06-08 | 中国科学院国家空间科学中心 | Mapping method based on satellite-borne solid-state storage system self-adaptive flash translation layer |
| CN111506517A (en) * | 2020-03-05 | 2020-08-07 | 杭州电子科技大学 | Flash memory page level address mapping method and system based on access locality |
| CN112000296A (en) * | 2020-08-28 | 2020-11-27 | 北京计算机技术及应用研究所 | Performance optimization system in full flash memory array |
| CN112000296B (en) * | 2020-08-28 | 2024-04-09 | 北京计算机技术及应用研究所 | Performance optimization system in full flash memory array |
| CN112506445A (en) * | 2020-12-29 | 2021-03-16 | 杭州电子科技大学 | Partition proportion self-adaptive adjusting method of homogeneous hybrid solid state disk |
| CN112506445B (en) * | 2020-12-29 | 2022-05-20 | 杭州电子科技大学 | Partition proportion self-adaptive adjustment method for homogeneous hybrid solid state disk |
| CN113220241A (en) * | 2021-05-27 | 2021-08-06 | 衢州学院 | Cross-layer design-based hybrid SSD performance and service life optimization method |
| CN113435109A (en) * | 2021-06-04 | 2021-09-24 | 衢州学院 | Performance and service life optimization method of hybrid SSD |
| CN113435109B (en) * | 2021-06-04 | 2024-01-30 | 衢州学院 | Optimization method for performance and service life of mixed SSD |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110413537B (en) | 2021-08-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110413537A (en) | A kind of flash translation layer (FTL) and conversion method towards hybrid solid-state hard disk | |
| Kang et al. | Performance trade-offs in using NVRAM write buffer for flash memory-based storage devices | |
| US8261010B2 (en) | Methods for distributing log block associativity for real-time system and flash memory devices performing the same | |
| US10241919B2 (en) | Data caching method and computer system | |
| US10740251B2 (en) | Hybrid drive translation layer | |
| Jiang et al. | S-FTL: An efficient address translation for flash memory by exploiting spatial locality | |
| US9652386B2 (en) | Management of memory array with magnetic random access memory (MRAM) | |
| CN102981963B (en) | A kind of implementation method of flash translation layer (FTL) of solid-state disk | |
| CN103019958B (en) | Usage data attribute manages the method for the data in solid-state memory | |
| Wu et al. | GCaR: Garbage collection aware cache management with improved performance for flash-based SSDs | |
| CN105930282B (en) | A kind of data cache method for NAND FLASH | |
| US20130198439A1 (en) | Non-volatile storage | |
| CN105339910B (en) | Virtual NAND capacity extensions in hybrid drive | |
| CN104166634A (en) | Management method of mapping table caches in solid-state disk system | |
| CN109062505A (en) | A kind of write performance optimization method under cache policy write-in layering hardware structure | |
| CN105389135B (en) | A kind of solid-state disk inner buffer management method | |
| Wei et al. | CBM: A cooperative buffer management for SSD | |
| US20200310984A1 (en) | Storage device, controller and method for operating controller | |
| US20190303019A1 (en) | Memory device and computer system for improving read performance and reliability | |
| CN109783398A (en) | One kind is based on related perception page-level FTL solid state hard disk performance optimization method | |
| CN109446117A (en) | A kind of solid state hard disk page grade flash translation layer (FTL) design method | |
| CN108845957B (en) | Replacement and write-back self-adaptive buffer area management method | |
| US20130007341A1 (en) | Apparatus and method for segmented cache utilization | |
| CN110262982A (en) | A kind of method of solid state hard disk address of cache | |
| CN105607862A (en) | Solid state disk capable of combining DRAM (Dynamic Random Access Memory) with MRAM (Magnetic Random Access Memory) and being provided with backup power |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | 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 |