CN103377152A - Write operation control method and write operation device for solid state disk - Google Patents

Abstract

An embodiment of the invention provides a write operation control method and a write operation device for a solid state disk. The solid state hard disk comprises a first storage unit and a second storage unit. The first storage unit and the second storage unit are both nonvolatile storage devices. The method includes: writing data into the first storage unit; writing the data in the first storage unit into the second storage unit. According to the write operation control method and the write operation device for the solid state disk, the data is first written into the first storage unit which is the nonvolatile storage device to be cached, then the cache data stored in the first storage unit is written into the second storage unit which is the nonvolatile storage device to be stored permanently instantly or at intervals, and therefore even under the condition of unsafe shutdown, the cache data which is in the first storage unit and is not written into the second storage unit to be stored permanently can not be lost, and data loss rate of the solid state disk is reduced.

Description

The write operation control method of solid state hard disc and write operation device

Technical field

The present invention relates to field of computer technology, be specifically related to a kind of write operation control method and write operation device of solid state hard disc.

Background technology

Solid state hard disc (Solid-State Drive, SSD) SSD is a kind of than less, the lighter novel storer of hard disk drive (Hard Disk Drive, HDD), and SSD is widely used in computer realm.SSD also is called electronic magnetic disc (electronic disk) sometimes, is a kind of data storage device that needs permanent data of preserving and the IO access mode the same with traditional hard disk is provided simultaneously of preserving with solid-state memory.Tradition SSD is comprised of NAND Flash, DRAM and SSD controller, as shown in Figure 1, the data write operation of traditional SSD according to the following steps in the prior art: at first the SSD controller writes DRAM with data and carries out buffer memory, data cached among the DRAM is write persistence among the NAND Flash again.

Most enterprises level SSD relies on power-fail circuit, to monitor supply voltage, can produce " early warning " signal and sends to the SSD controller if voltage is lower than predefined threshold value.The secondary voltage holding circuit of reserving is used for guaranteeing that driver in emergency circumstances has time enough that data are dumped to NAND Flash in by DRAM any.Reserve voltage power-on time meaning figure but be illustrated in figure 2 as, generally, the secondary voltage of reservation can be Millisecond for the time of drive operation.During safety shutdown, most of host computer system all can start a provisional order to SSD so that SSD has sufficient time to shutdown system, thereby so that SSD the data of extra buffer among the current DRAM can be saved among the nonvolatile memory NAND FLASH.But, when non-security shutdown, SSD is suddenly power down before host computer system starts provisional order, this situation is so that be kept at the data of carrying out buffer memory among the DRAM and also do not have enough time to be saved among the nonvolatile memory NAND, and DRAM is not Nonvolatile memory devices, so during non-security shutdown, the data that are buffered among the DRAM can be lost.In a single day these data are just lost and can't be recovered.So the write operation flow process utmost point of this solid state hard disc is unfavorable for the preservation of data.

Summary of the invention

In view of this, the object of the present invention is to provide write operation control method and the write operation device of the extremely low solid state hard disc of a kind of non-security when shutdown data loss rate, cause the defective of loss of data to overcome power fail in the prior art.

For achieving the above object, one embodiment of the present of invention provide a kind of write operation control method of solid state hard disc, described solid state hard disc comprises the first storage unit and the second storage unit, wherein said the first storage unit and described the second storage unit are Nonvolatile memory devices, and described method comprises:

Data are write the first storage unit;

Data in described the first storage unit are write the second storage unit.

Preferably, described solid state hard disc also comprises the 3rd storage unit, and described the 3rd storage unit is volatile storage, and the writing speed of described the first storage unit is greater than the writing speed of described the second storage unit; Wherein,

Described data are write the first storage unit, comprising:

Data are write the 3rd storage unit;

During described solid state hard disc power down, the described data in described the 3rd storage unit are write the first storage unit;

Described data in described the first storage unit are write the second storage unit, comprising:

After described solid state hard disc powers on, the described data in described the first storage unit are write described the 3rd storage unit;

Described data in described the 3rd storage unit are write the second storage unit.

Preferably, described data are write the first storage unit before, also comprise:

Whether the data buffer area of judging described the first storage unit is full, if so, described data write described the second storage unit; If not, described data are write the first storage unit.

Preferably, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described data in described the first storage unit are write the second storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the second storage unit.

Preferably, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the 3rd storage unit; Described data are write the 3rd storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the 3rd storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the 3rd storage unit.

Preferably, described map information also comprises erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described described data in described the 3rd storage unit are write the second storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the second storage unit.

Preferably, described method also comprises:

Update stored in the physical address of physical block in described the first storage unit in described the first storage unit and/or described the second storage unit and the mapping relations of logical address.

Preferably, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the first storage unit; Described method also comprises:

In described map information, the historical physical block of storing described data in described the first storage unit is labeled as invalid block.

In addition, the embodiment of the invention also provides a kind of write operation device of solid state hard disc, described solid state hard disc comprises the first storage unit and the second storage unit, and wherein said the first storage unit and described the second storage unit are Nonvolatile memory devices, and described device comprises:

The first writing unit is used for data are write the first storage unit;

The second writing unit writes the second storage unit with the data in described the first storage unit.

Preferably, described solid state hard disc also comprises the 3rd storage unit, and described the 3rd storage unit is volatile storage, and the writing speed of described the first storage unit is greater than the writing speed of described the second storage unit; Wherein,

Described the first writing unit comprises:

First writes subelement, is used for data are write the 3rd storage unit;

Second writes subelement, is used for when described solid state hard disc power down the described data in described the 3rd storage unit being write the first storage unit;

Described the second writing unit comprises:

The 3rd writes subelement, is used for after described solid state hard disc powers on the described data in described the first storage unit being write described the 3rd storage unit;

The 4th writes subelement, is used for the described data of described the 3rd storage unit are write the second storage unit.

Preferably, described system also comprises:

The first judging unit is used for judging whether the data buffer area of described the first storage unit is full.

Preferably, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described the second writing unit comprises:

The first scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

The 5th writes subelement, is used for the described mapping relations that described the first scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the second storage unit.

Preferably, described the first storage unit map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the 3rd storage unit; Described first writes subelement comprises:

The second scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the 3rd storage unit;

The 6th writes subelement, is used for the described mapping relations that described the second scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the 3rd storage unit.

Preferably, described the first cell stores map information, described map information also comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; The described the 4th writes subelement comprises:

The 3rd scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

The 7th writes subelement, is used for the described mapping relations that described the 3rd scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the second storage unit.

Preferably, described system also comprises:

Updating block is for the physical address of physical block and the mapping relations of logical address in described the first storage unit that updates stored in described the first storage unit and/or described the second storage unit.

Preferably, described the first storage unit map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the first storage unit; Described system also comprises:

Indexing unit is used at described map information, and the historical physical block of storing described data in described the first storage unit is labeled as invalid block.

Write operation control method and the write operation device of the solid state hard disc that provides according to the embodiment of the invention, the first storage unit that data is written as first Nonvolatile memory devices is carried out buffer memory, and then data cached the second storage unit that is written as Nonvolatile memory devices instant or that will be stored in the first storage unit every the set time is carried out persistence, like this, even in the situation of non-security shutdown, the i.e. unexpected power down of solid state hard disc in data writing process, can not cause data cached the losing that does not write in first storage unit of carrying out persistence in the second storage unit yet, thereby reduced the probability of solid state hard disc loss of data, improved the reliability of solid state hard disc data storages.

Description of drawings

In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, the below will do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.

Fig. 1 is the structural representation of conventional solid-state hard disk in the prior art;

But Fig. 2 is the secondary voltage power-on time synoptic diagram of reserving in the prior art;

Fig. 3 is the structural representation of solid state hard disc in the write operation control method of the embodiment of the invention;

Fig. 4 is the process flow diagram of write operation control method of the solid state hard disc of the embodiment of the invention;

Fig. 5 is the structural representation of solid state hard disc in the write operation control method of solid state hard disc of the embodiment of the invention one;

Fig. 6 is the data structure synoptic diagram of conventional solid-state hard disk in the prior art;

Fig. 7 is the data structure synoptic diagram of solid state hard disc in the write operation control method of solid state hard disc of the embodiment of the invention one;

Fig. 8 is the write operation control method process flow diagram of the solid state hard disc of the embodiment of the invention one;

Fig. 9 is the structural representation of solid state hard disc in the write operation control method of solid state hard disc of the embodiment of the invention two;

Figure 10 is the structural representation of solid state hard disc in the write operation control method of solid state hard disc of an instantiation of the embodiment of the invention two;

Figure 11 is the data structure synoptic diagram of solid state hard disc in the write operation control method of solid state hard disc of an instantiation of the embodiment of the invention two;

Figure 12 is the write operation control method process flow diagram of solid state hard disc of an instantiation of the embodiment of the invention two;

Figure 13 is the structural representation of the write operation device of the solid state hard disc that provides of the embodiment of the invention.

Embodiment

For the purpose, technical scheme and the advantage that make the embodiment of the invention clearer, below in conjunction with the accompanying drawing in the embodiment of the invention, technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that obtains under the creative work prerequisite.

Non-security shutdown is caused by unexpected power down usually, cause the factor of unexpected power down (the following stated " suddenly power down " i.e. is representative " non-security shutdown ") that fault, pcb board or the connector fault of the fault of power supply grid or UPS, the power fail of system (comprise and merging and wiring), SSD voltage adjusting assembly may be arranged, and the impact of heating etc.Power fail may cause the delay (situation that needs reconstructed mapped information when driver) of system and data are nonvolatil loses.

Traditional SSD adopts volatile storage (such as DRAM) to carry out data buffer storage, and when non-security shutdown, data cached the having little time that is kept among the DRAM is saved in the nonvolatile memory, so may cause losing of data.

For the possibility that is reduced in non-security when shutdown solid state hard disc loss of data, improve the reliability of solid state hard disc data storage, the embodiment of the invention provides a kind of write operation control method of solid state hard disc, wherein, be illustrated in figure 3 as the structural representation of this solid state hard disc, this solid state hard disc 300 comprises the first storage unit 301 and the second storage unit 302, and wherein the first storage unit 301 and the second storage unit 302 are Nonvolatile memory devices; As shown in Figure 4, be the process flow diagram of the write operation control method of solid state hard disc 300, this write operation control method may further comprise the steps:

Step S401: data are write the first storage unit 301;

Because the first storage unit 301 be Nonvolatile memory devices, the data that namely write in the first storage unit 301 can not cut off the power supply suddenly or other factors cause non-security shutdown to cause loss of data owing to power supply; So the data that write the first storage unit 301 can be carried out buffer memory, also can carry out persistence.

In addition, the first storage unit 301 can also be stored the physical address of physical block in some hot spot data, working procedure and storage unit (comprising the first storage unit and/or the second storage unit) that need frequent access and the mapping relations of logical address etc.

Particularly, the first storage unit 301 can be Nonvolatile memory devices, such as EEPROM (Electrically Erasable Programmable Read Only Memo) (Electrically Erasable Programmable Read-Only Memory, EEPROM), flash memory (Flash Memory) and ferroelectric memory (Ferroelectric Memory, FRAM), magnetic RAM (Magnetoresistantive Random Access Memory, MRAM), phase transition storage (Phase-Change Memory, PCM) etc.

Step S402: the data in the first storage unit 301 are write the second storage unit 302;

Particularly, can be with the first storage unit 301 as buffer memory, with writing data in the first storage unit 301 after the first storage unit 301 is carried out buffer memory, immediately or at regular intervals these data cached second storage unit 302 that write are carried out persistence again.

Particularly, the second storage unit can be Flash Memory, NAND Flash etc.

The write operation control method of the solid state hard disc that provides according to the embodiment of the invention, data are written as first in the first storage unit of Nonvolatile memory devices and carry out buffer memory, and then data cached the second storage unit that is written as Nonvolatile memory devices instant or that will be stored in the first storage unit every the set time is carried out persistence, like this, even in the situation of non-security shutdown, the i.e. unexpected power down of solid state hard disc in data writing process, can not cause data cached the losing that does not write in first storage unit of carrying out persistence in the second storage unit yet, thereby the probability of solid state hard disc loss of data when having reduced non-security shutdown has improved the reliability of solid state hard disc data storages.

The write operation control method of the solid state hard disc in the embodiment of the invention specifically can have multiple implementation; the below describes these modes with several instantiations; need to prove; these examples only are the explanations to embodiment of the invention technical scheme; be not intended to limit the present invention technical scheme; those of ordinary skills can expect the replacement scheme that other are similar or be equal to easily on the basis of technical solution of the present invention, need not the similar techniques scheme that can realize the object of the invention that creative work obtains on the basis of thinking of the present invention all in protection scope of the present invention.

Embodiment one

As an instantiation, the first storage unit 301 in the embodiment of the invention can be PCM, the second storage unit 302 can be NAND Flash, below the write operation control method of the solid state hard disc that the embodiment of the invention provided as example be elaborated.

The embodiment of the invention one provides a kind of write operation control method of solid state hard disc, this solid state hard disc comprises the first storage unit 301 and the second storage unit 302, wherein, the first storage unit 301 is PCM, the second storage unit 302 is NAND Flash, as shown in Figure 5, is the structural representation of the solid state hard disc of the write operation control method of the embodiment of the invention one, except PCM and NAND Flash, this SSD can also comprise host interface and SSD controller.

As shown in Figure 7, data structure synoptic diagram for the solid state hard disc of the write operation control method of the embodiment of the invention one, in order to realize flash translation layer (FTL) (FTL), PCM is except can memory buffers data (Data Cache), all right storage running code (Routine Code) and map information (mapping information), this map information comprises erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in the storage unit (comprising the first storage unit 301 and/or the second storage unit 302), erase count table (Erase Count Table, ECT), free block list (Free Block Table, FBT), invalid block table (Invalid Block Table, IBT) etc.; Wherein, compare with the data structure synoptic diagram (namely shown in Figure 6) of conventional solid-state hard disk, in the solid state hard disc of the embodiment of the invention one, what map information can be lasting is kept among the PCM, and NAND Flash can not preserve any map information; In addition, we can preserve performance and the life-span that those hot spot datas (Hot Data) improve SSD with a part of space of PCM.

PCM is a kind of nonvolatile memory.PCM utilizes chalcogenide glass can switch this characteristic between crystal or noncrystal two states in the situation of heating to store data.PCM is a kind of emerging memory technology, has a lot of significant advantages: non-volatile, and the byte addressing, the high 2-4 of density ratio DRAM doubly read latency, write delay, writes durability degree than the excellent several quantity grades of NAND Flash.Table 1 is that each characteristic of the storer of PCM and other kinds compares.

Table 1

As shown in Figure 8, be the process flow diagram of the method, the method may further comprise the steps:

Step S801: judge whether the data buffer area among the PCM is full, if so, execution in step S802; If not, execution in step S803.

Particularly, data being write before PCM carries out buffer memory, can judge at first whether the data buffer area of PCM is full, if full, can directly data be write NAND Flash, and not carry out buffer memory.Can reduce to a certain extent writing speed like this, but can guarantee that data can not lose.

Step S802: data are write NAND Flash.

Particularly, this step can be carried out in the following manner:

The map information of scanning among the PCM finds the physical address of the minimum physical block of erasing times among the NAND Flash that is stored among the PCM and the mapping relations of logical address; In embodiments of the present invention, can find a physical block that erasing times is minimum by scanning ECT and FBT.

According to these mapping relations, data are write the minimum physical block of erasing times among the NAND Flash.

In this way, data can be write be not wiped free of or NAND Flash that erasing times is minimum in physical block, like this, can be by the allotment of SSD controller, write operation is distributed to physical blocks different among the NAND Flash fifty-fifty, cause the life-span of this physical block to reduce to avoid that a certain physical block is repeatedly carried out write operation, namely carry out abrasion equilibrium (Wear Leveling, WL) function is to improve the bulk life time of solid state hard disc.

Step S803: data are write PCM.

Particularly, this step can be carried out in the following manner:

The map information of scanning among the PCM finds the physical address of the minimum physical block of erasing times among the PCM that is stored among the PCM and the mapping relations of logical address;

According to these mapping relations, data are write the minimum physical block of erasing times among the PCM of institute.

In this way, data can be write be not wiped free of or PCM that erasing times is minimum in physical block, like this, can be by the allotment of SSD controller, write operation is distributed to physical blocks different among the PCM fifty-fifty, cause the life-span of this physical block to reduce to avoid that a certain physical block is repeatedly carried out write operation, namely carry out abrasion equilibrium (Wear Leveling, WL) function is to improve the bulk life time of solid state hard disc.

Step S804: with the data cached NAND Flash that writes among the PCM.

Data are write PCM carries out immediately or periodically the data cached NAND of the writing Flash in the PCM data buffer area being carried out persistence again after the buffer memory.

Particularly, this step can be carried out in the following manner:

The map information of scanning among the PCM finds the physical address of the minimum physical block of erasing times among the NAND Flash that is stored among the PCM and the mapping relations between the logical address;

According to these mapping relations, data are write the minimum physical block of erasing times among the NAND of institute.

In this way, can with data cached write be not wiped free of or NAND Flash that erasing times is minimum in physical block, like this, can be by the allotment of SSD controller, write operation is distributed to physical blocks different among the NAND Flash fifty-fifty, cause the life-span of this physical block to reduce to avoid that a certain physical block is repeatedly carried out write operation, namely carry out abrasion equilibrium (Wear Leveling, WL) function is to improve the bulk life time of solid state hard disc.

For example, the hot spot data that is kept among the PCM immediately can be write NAND Flash, perhaps the map information that is kept among the PCM periodically can be write NAND Flash, also the map information among the PCM for good and all can be kept among the PCM.

The write operation control method of the solid state hard disc that provides by the embodiment of the invention one, on the one hand, when non-security shutdown (such as the System Sudden power down), do not write among the data cached PCM that still is kept at as Nonvolatile memory devices that carries out persistence among the NAND Flash, after re-powering etc. system, can continue to be kept among the data cached NAND of the writing Flash among the PCM and carry out persistence, can not cause data cached losing because of suddenly power down.Namely, even the unexpected power down of solid state hard disc in data writing process, also can not cause and not write data cached the losing of carrying out persistence in the second storage unit, thereby the probability of solid state hard disc loss of data when having reduced non-security shutdown has improved the reliability of solid state hard disc data storages.

On the other hand, in the situation of safety shutdown, within the time that the secondary voltage of reserving can be powered, may continue to carry out will write the data of carrying out among the PCM behind the buffer memory and write the action that NAND Flash carries out persistence in the solid state hard disc, loss of data can not occur because of shutdown in the data that write behind the NAND Flash; After reservation voltage can't work, also be not written into NAND Flash after may still having partial data through the PCM buffer memory and carry out persistence, because PCM is Nonvolatile memory devices, loss of data can not occur because of shutdown in the data that are kept among the PCM, like this, the probability of loss of data when just having reduced the security of system shutdown has further improved the reliability of data storages.

In addition, after execution of step S802 and/or step S804, can also may further comprise the steps:

At first, upgrade the mapping relations among the PCM, corresponding one by one with physical address and the logical address of the physical block of save data to guarantee the map information among the PCM.This step can comprise following two aspects: upgrade the mapping relations of the physical address of physical block among the PCM be kept among the PCM and logical address and renewal and be kept at the physical address of physical block among the NAND Flash among the PCM and the mapping relations of logical address, can guarantee that like this mapping relations are corresponding one by one with the address of the physical block of current storage data, can find quickly and easily the physical block of these data of storage by these mapping relations user.

Owing to be to write data into logical address, must logical address be mapped to physical address by the logical relation of the logical address among the PCM and physical address; Simultaneously, new physical block should be deleted from the FBT table, and data can write same logical block and can not write same physical block, can further improve like this life-span of SSD.

Secondly, in the map information in PCM, historical physical block among the PCM is labeled as invalid block, so that this piece is wiped in refuse collection, use this piece in the convenient follow-up ablation process, here, historical physical block refers to that the data of storing have been written into NAND Flash and have carried out persistence or transfer to the physical block that preserve other positions, and namely historical physical block is not stored data.

Particularly, can historical physical block be labeled as invalid block by the invalid block table that upgrades in the map information, wherein, historical physical block refers to preserve the physical block of these data, in the embodiment of the invention one, data cached write NAND Flash by PCM after, the physical block of before storing among this data cached PCM is historical physical block.

Above-mentioned steps in the embodiment of the invention one can be finished by the control of the SSD controller in the solid state hard disc.

Because PCM is Nonvolatile memory devices, so that comparing with traditional SSD, the SSD that is made of PCM and NAND Flash that the embodiment of the invention one provides has better reliability.

The reliability of the SSD that the embodiment of the invention one is provided below by concrete formula and traditional SSD compares.

In traditional SSD, power fail may cause SSD system delay or nonvolatil loss of data, and reason is as follows: when power fail occurring, still exist some data cached and/or map informations not to be written into NAND Flash in DRAM.Traditional SSD write-back map information to time of NAND Flash no longer than reserving the continuable time of voltage, traditional SSD write-back map information can use formula (1) to represent to the needed time of NAND Flash:

T _{traditional_writeback}＝(N _data+N _mapinf?or)*(T _scan+T _{write_NAND}+T _map+T _mark)..................(1)

Wherein, T _{Traditional_writeback}The needed time of the data cached NAND of writing Flash among traditional SSD, N _DataThe piece number that is buffered in the data among the DRAM, N _{Mapinf or}The piece number that is buffered in the map information among the DRAM, T _ScanThat scanning ECT and FBT obtain the needed time of physical block that erasing times is minimum, T _{Write_NAND}The needed time of the data cached NAND of writing Flash, T _MapTo upgrade map information and revise needed time of FBT, T _MarkThat historical physical block is labeled as the needed time of invalid block.

Yet, in the SSD that is consisted of by PCM and NAND Flash that the embodiment of the invention one provides, owing to map information is permanently stored among the PCM, so N _{Mapinf or}Be zero, so in the situation of other parameter constants, the SSD that the embodiment of the invention one provides is higher than traditional SSD reliability.

In addition, in the process of the map information in the physical block that data is write minimum erasing times among the NAND Flash and renewal PCM, if data cachedly do not write NAND Flash or logical address is not mapped to NAND Flash, then data cached or logical address must be mapped to PCM, causes losing of data cached or logical address to avoid the unexpected power down of SSD.Data in PCM will exist always, so data will be lost never.If power down occurs in the map information among the renewal PCM or historical physical block is labeled as in the process of invalid block, when restoring electricity, should reclaim immediately these pieces.

In addition, can be some hot spot datas, exist among the PCM such as the partial data of file configuration table (File Location Table, FAT).There is dual mode to choose hot spot data.A kind of is static mode, take the SSD of FAT file system format as example, the partial logic of FAT can be mapped to the hot spot data space.Another is dynamical fashion, and in this mode, can which decide according to access times is hot spot data.Because FAT is the data that need frequent access, and the read-write of PCM time-delay will be good than NANDFlash many, the permanance of PCM also is better than NAND Flash, so if adopt the hot spot data scheme, the life-span of the SSD that the embodiment of the invention one provides will be longer than the life-span of traditional SSD.

Although tradition also can adopt the hot spot data scheme with the SSD of DRAM, but the SSD that is made of PCM and NAND Flash that traditional SSD does not have the embodiment of the invention one to provide is reliable, because the hot spot data of the SSD that is made of PCM and NAND Flash that the embodiment of the invention provides is to be kept among the Nonvolatile memory devices PCM, and DRAM is volatile memory, non-volatile these characteristics of PCM here, the SSD that the embodiment of the invention one is provided has higher reliability.Suppose that the abrasion equilibrium algorithm is enough outstanding, the life-span of traditional SSD can represent with following expression formula:

The expected life of tradition SSD can be represented by equation (2):

$\mathrm{Traditional}\_\mathrm{Lifetime}\left(\mathrm{years}\right)=\frac{(\mathrm{SSD}\_\mathrm{Capacity})(P/E_{\mathrm{NAND}})(\mathrm{Percent}\_\mathrm{Utilization})}{(\mathrm{Usage}/\mathrm{Day})(\mathrm{Capacity}\_\mathrm{Rate})(365\mathrm{days}/\mathrm{year})}---\left(2\right)$

Wherein, Traditional_Lifetime (years) is the life-span of traditional SSD, P/E _NAND:NANDBe the durability times of SSD, SSD_Capacity is the capacity of SSD, and Percent_Utilization is utilization factor, and Usage/Day represents the capacity used every day, and Capacity_Rate is the active volume ratio, 365days/year namely 365 days/year.

In the hybrid solid-state hard disk of this static layering memory module, PCM will share erasing times jointly with NAND Flash.If PCM Program code space is invalid, PCM still can be used as the storer of run time version so.The life-span of the SSD of the embodiment of the invention one can be used equation (3) expression:

$\begin{array}{c}\mathrm{Hybrid}\_\mathrm{Lifetime}\left(\mathrm{years}\right)=\frac{(\mathrm{NAND}\_\mathrm{Capa\; city})(P/E_{\mathrm{NAND}})(\mathrm{Percent}\_\mathrm{Utilization}\_\mathrm{NAND})}{(\mathrm{Usage}/\mathrm{Day})(\mathrm{Capacity}\_\mathrm{Rate})(365\mathrm{days}/\mathrm{year})}\\ +\frac{(\mathrm{PCM}\_\mathrm{Capac\; ity})(P/E_{\mathrm{PCM}})(\mathrm{Percent}\_\mathrm{Utilization}\_\mathrm{PDM})}{(\mathrm{Usage}/\mathrm{Day})(\mathrm{Capacity}\_\mathrm{Rate})(365\mathrm{dats}/\mathrm{year})}\end{array}---\left(3\right)$

Wherein, Hybrid_Lifetime (years) is the life-span of the SSD of the embodiment of the invention one, P/E _NAND:NANDBe the durability times of the SSD of the embodiment of the invention one, P/E _PCMBe the PCM durability times among the SSD of the embodiment of the invention one, Percent_Utilization_NAND is the NAND utilization factor among the SSD of the embodiment of the invention one, SSD_Capacity is the capacity of the SSD of inventive embodiments one, Percent_Utilization_PCM is the PCM utilization factor, Usage/Day is the capacity of using every day, Capacity_Rate is the active volume ratio, 365days/year namely 365 days/year.

According to equation (2) and (3), if the level storage algorithm of traditional SSD is enough outstanding.Multiple relation by life-span of life-span of PCM mixing SSD and traditional SSD can use equation (4) to represent:

$\mathrm{Multiple}=1+\frac{(\mathrm{PCM}\_\mathrm{Capacity})(P/E_{\mathrm{PCM}})(\mathrm{Percent}\_\mathrm{Utilization}\_\mathrm{PCM})}{(\mathrm{NAND}\_\mathrm{Capacity})(P/E_{\mathrm{NAND}})(\mathrm{Percent}\_\mathrm{Utilization}\_\mathrm{NAND})}---\left(4\right)$

As shown in Table 1, the permanance of PCM and NAND Flash is respectively 10 ⁶, 10 ⁸With 10 ⁴, 10 ⁵The capacity of common SSD is 128GB, and the capacity of common PCM is 128MB.If 50% capacity with PCM is preserved hot spot data, the permanance that can be released SSD by equation (4) just increases by 1.005 to 6 times.

As seen, reliability and the life-span of the SSD that is made of PCM and NAND Flash that provides of the embodiment of the invention all are better than traditional SSD.

In addition, the write operation control method of the solid state hard disc that provides of the embodiment of the invention one can also may further comprise the steps:

The free block that is arranged in the IBT table is collected in refuse collection (Garbage Collection), then wipes the data in the executing garbage take piece as unit, so that follow-up data writing uses;

Refuse collection increases the value of the erase counters among the ECT, with the piece number of physical block available in the statistics storage unit;

Refuse collection sends back to the free block that reclaims in the FBT table that is arranged in PCM.Final these pieces accumulate pool of free blocks.

In this process, when the physical block that detects damage, the data that refuse collection will damage in the physical block are written in the minimum physical block of physical block erase count again, and the logical address that then remaps is to new physical address.When physical block is written to pool of free blocks, logical address is counted and be remapped to the minimum physical block of erasing times again to new physical address.The erroneous block that detects from IBT is used never with deleted.

Whether buffer memory program checkout data buffer area is full, if data buffer area is full, the buffer memory program will repeat above step and clear up data buffer area.

Must wipe first traditional SSD that just can carry out write operation with main storage medium NAND Flash compares, the write operation control method of the solid state hard disc that the embodiment of the invention one provides can be wiped free block before write operation, garbage collector can not need to wipe whole data block simultaneously, thereby can improve write efficiency.

Embodiment two

The embodiment of the invention two provides a kind of write operation control method of solid state hard disc, as shown in Figure 9, structural representation for this SSD900, this SSD comprises: the first storage unit 901, the second storage unit 902 and the 3rd storage unit 903, wherein the 3rd storage unit 903 can be volatile storage, and the writing speed of the first storage unit 901 is greater than the writing speed of the second storage unit 902.In the embodiment of the invention, specifically take the first storage unit 901 as PCM, the second storage unit 902 is an instantiation as NAND Flash, the 3rd storage unit 903 as DRAM, the data write operation control method of the SSD that the embodiment of the invention two is provided is elaborated.As shown in figure 10, be the structural representation of solid state hard disc in the write operation control method of the solid state hard disc of an instantiation of the embodiment of the invention two; As shown in figure 11, the data structure synoptic diagram of solid state hard disc in the write operation control method that provides for an instantiation of the embodiment of the invention two, wherein, PCM can be used for some hot spot datas of storage, and DRAM can be used for some working procedures of storage etc.

Need to prove; the first storage unit 901 is not limited to PCM in the instantiation of the embodiment of the invention two; the second storage unit 902 is not limited to NAND Flash; the 3rd storage unit 903 also is not limited to DRAM, all satisfy an instantiation of the embodiment of the invention two qualifications, and can realize the similar of the object of the invention or the technical scheme that is equal to all in the scope of protection of present invention.

As shown in figure 12, the process flow diagram of the write operation control method of the SSD that is consisted of by PCM, NANDFlash and DRAM that provides for an instantiation of the embodiment of the invention two, the method may further comprise the steps:

Step S121: data are write DRAM;

Particularly, this step can be carried out in the following manner:

The map information of scanning among the PCM finds the physical address of the minimum physical block of erasing times among the DRAM that is stored among the PCM and the mapping relations between the logical address;

According to these mapping relations, data are write the minimum physical block of erasing times among the DRAM.

In this way, can with data cached write be not wiped free of or DRAM that erasing times is minimum in physical block, like this, can be by the allotment of SSD controller, write operation is distributed to physical blocks different among the DRAM fifty-fifty, cause the life-span of this physical block to reduce to avoid that a certain physical block is repeatedly carried out write operation, namely carry out abrasion equilibrium (Wear Leveling, WL) function is to improve the bulk life time of solid state hard disc.

Particularly, data can be write DRAM and carry out buffer memory, and then immediately or data cached the writing that will deposit among the DRAM every the set time carry out persistence in the Nonvolatile memory devices.

During the unexpected power down of step S122:SSD, with the data cached PCM that writes among the DRAM;

In the traditional SSD of DRAM as buffer memory, after the unexpected power down of SSD, the data that are not written among the DRAM of Nonvolatile memory devices will be lost, and this also is to cause one of low reason of the data storing reliability of traditional SSD.In the instantiation of the embodiment of the invention two, when the SSD power down, with data cached the writing rapidly among the PCM among the DRAM, can reduce the probability of loss of data.

Particularly, before execution in step S122, can also carry out following steps:

Judge whether the data buffer area among the PCM is full, if so, data write NAND Flash; If not, data are write PCM.Data being write before PCM carries out buffer memory, can judge at first whether the data buffer area of PCM is full, if full, can directly data be write NAND Flash.Can reduce to a certain extent writing speed like this, but can guarantee that data can not lose.

After step S123:SSD powers on, the data among the PCM are write DRAM.

Particularly, can adopt the faster speed carried out to improve this step as the first storage unit 901 of Nonvolatile memory devices of a kind of data reading speed, reduce the needed time of whole write operation.

Step S124: the data among the DRAM are write NAND Flash, to realize the persistence of data.

In the write operation control method of the SSD that is consisted of by DRAM, PCM and NAND Flash that an instantiation of the embodiment of the invention two provides, during the System Sudden power down, with the data cached PCM that writes among the DRAM, particularly can be immediately or each set time with the data cached PCM that writes among the DRAM.Because the writing speed of PCM is greater than the writing speed of NAND Flash, so compare with the write operation control method of traditional SSD, within the identical time interval (as reserving voltage in the continuable time), data cachedly PCM can be write more, therefore, the loss of data probability of SSD in the time of can reducing the System Sudden power down improves the reliability of SSD data storage.

After execution of step S124, can also may further comprise the steps:

At first, upgrade the mapping relations among the PCM, corresponding one by one with physical address and the logical address of the physical block of save data to guarantee the map information among the PCM.This step can comprise following three aspects: upgrade the mapping relations of the physical address of physical block among the PCM be kept among the PCM and the mapping relations of logical address, the physical address that upgrades physical block among the DRAM that is kept among the PCM and logical address and renewal and be kept at the physical address of physical block among the NAND Flash among the PCM and the mapping relations of logical address.

In addition, in described map information, the historical physical block of storing data among PCM and/or the DRAM is labeled as invalid block, so that this piece is wiped in refuse collection, uses this piece in the follow-up ablation process in aspect.Here, historical physical block refers to that the data of storing have been written into NAND Flash and have carried out persistence or transfer to the physical block that preserve other positions, i.e. the current data of not storing of historical physical block.

Because PCM is Nonvolatile memory devices, so that comparing with traditional SSD, the solid state hard disc that is made of PCM, DRAM and NAND Flash that an instantiation of the embodiment of the invention two provides has better reliability.

The SSD that the embodiment of the invention is provided below by concrete formula and the reliability of traditional SSD compare.

In traditional SSD, power fail may cause SSD system delay or nonvolatil loss of data, and reason is as follows: when power fail occurring, still exist some data cached and/or map informations not to be written into NAND Flash in DRAM.The write-back map information of the SSD that an instantiation of the embodiment of the invention two provides can be used formula (5) expression to the needed time of NAND Flash:

T _{hybrid_wrtieback}＝(N _data+N _map?inf?or)*T _{write_PCM}..........................................(5)

Wherein, T _{Hybrid_writeback}For writing data cached time to PCM, N _DataBe piece number data cached among the DRAM, N _MapinforBe the piece number of map information among the DRAM, T _{Write_PCM}For writing a data block to the PCM required time.

$\frac{T_{\mathrm{traditional}\_\mathrm{wrtieback}}}{T_{\mathrm{hybrid}\_\mathrm{wrtieback}}}=\frac{(N_{\mathrm{data}}+N_{\mathrm{map\; inf\; o}})*(T_{\mathrm{scan}}+T_{\mathrm{write}\_\mathrm{NAND}}+T_{\mathrm{map}}+T_{\mathrm{mark}})}{(N_{\mathrm{data}}+N_{\mathrm{map\; inf\; o}})*T_{\mathrm{write}\_\mathrm{PCM}}}---\left(6\right)$

$\≈\frac{T_{\mathrm{write}\_\mathrm{NAND}}}{T_{\mathrm{write}\_\mathrm{PCM}}}=\frac{T_{\mathrm{wtite}\_\mathrm{latency}\_\mathrm{NAND}}}{T_{\mathrm{write}\_\mathrm{latency}\_\mathrm{PCM}}}$

Wherein, T _{Traditional_writeback}For writing data cached time to NAND Flash, T _{Hybrid_writeback}For writing data cached time to PCM, N _DataBe piece number data cached among the DRAM, N _MapinforBe the piece number of map information among the DRAM, T _{Write_PCM}For writing a data block to PCM required time, T _{Write_latency_NAND}Be NAND write delay time, T _{Write_latency_PCM}Be the PCM write delay time.

According to expression formula (6), can calculate, when power fail occurs, traditional SSD with data from DRAM dump to the required time of NAND Flash approximately be the SSD that provides of an instantiation of the embodiment of the invention two 500*64B/4KB=7.81 doubly.

As shown in figure 10, compare with traditional SSD, the SSD of this structure has increased a PCM and has come Storage Mapping information, and map information comprises map information, ECT, IBT and the FBT of logical address and physical address.In case power on, these data will be cached to DRAM.Simultaneously, the data among DRAM when spatial cache comes the buffer memory power fail in PCM, have been reserved.Compare with the SSD in the embodiment of the invention one, the SSD of an instantiation of the embodiment of the invention two can be placed on operation code among the DRAM and carry out.

Correspondingly, the embodiment of the invention also provides a kind of write operation device of solid state hard disc, wherein, this solid state hard disc comprises the first storage unit and the second storage unit, the first storage unit and the second storage unit are Nonvolatile memory devices, as shown in figure 13, the structural representation of the write operation device of the solid state hard disc that provides for the embodiment of the invention, this device 130 comprises:

The first writing unit 131 is used for data are write the first storage unit;

The second writing unit 132 writes the second storage unit with the data in described the first storage unit.

In addition, solid state hard disc can also comprise the 3rd storage unit, and the 3rd storage unit can be volatile storage, and the writing speed of the first storage unit is greater than the writing speed of the second storage unit; Wherein,

The first writing unit 131 can comprise:

First writes subelement, is used for data are write the 3rd storage unit;

Second writes subelement, is used for when the solid state hard disc power down data in the 3rd storage unit being write the first storage unit;

The second writing unit 132 can comprise:

The 3rd writes subelement, is used for after solid state hard disc powers on the data in the first storage unit being write the 3rd storage unit;

The 4th writes subelement, is used for the data of the 3rd storage unit are write the second storage unit.

This system can also comprise:

The first judging unit is used for judging whether the data buffer area of the first storage unit is full.

In embodiments of the present invention, the first storage unit can Storage Mapping information, and this map information can comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in the second storage unit; The second writing unit can comprise:

The first scanning element, the map information for scanning the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in the second storage unit;

The 5th writes subelement, is used for the mapping relations that foundation the first scanning element finds, and data are write the minimum physical block of erasing times in the second storage unit.

In embodiments of the present invention, this map information can also comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in the 3rd storage unit; First writes subelement can comprise:

The second scanning element, the map information for scanning the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in the 3rd storage unit;

The 6th writes subelement, is used for the mapping relations that foundation the second scanning element finds, and data are write the minimum physical block of erasing times in the 3rd storage unit.

In addition, this map information can also comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in the second storage unit; The 4th writes subelement can comprise:

The 3rd scanning element, the map information for scanning the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in the second storage unit;

The 7th writes subelement, is used for the mapping relations that foundation the second scanning element finds, and data are write the minimum physical block of erasing times in the second storage unit.

In addition, this map information also can comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in the first storage unit; This system can also comprise:

Updating block is for the physical address of physical block and the mapping relations of logical address in the first storage unit that updates stored in the first storage unit and/or the second storage unit; Can also be for the physical address of the physical block that updates stored in the 3rd storage unit and the mapping relations of logical address.

Indexing unit is used at map information, and the historical physical block of storing data in the first storage unit in the storage unit is labeled as invalid block; The historical physical block of storage data in the 3rd storage unit can also be labeled as invalid block.Here, historical physical block refers to that the data of storing have been written into the second storage unit and have carried out persistence or transfer to the physical block that preserve other positions, i.e. the current data of not storing of historical physical block.

Write operation control method and the write operation device of the solid state hard disc that provides according to the embodiment of the invention, the first storage unit that data is written as first Nonvolatile memory devices is carried out buffer memory, and then data cached the second storage unit that is written as Nonvolatile memory devices instant or that will be stored in the first storage unit every the set time is carried out persistence, like this, even in the situation of non-security shutdown, the i.e. unexpected power down of solid state hard disc in data writing process, can not cause data cached the losing that does not write in first storage unit of carrying out persistence in the second storage unit yet, thereby reduced the probability of solid state hard disc loss of data, improved the reliability of solid state hard disc data storages.

The above only is preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (16)

1. the write operation control method of a solid state hard disc is characterized in that, described solid state hard disc comprises the first storage unit and the second storage unit, and wherein said the first storage unit and described the second storage unit are Nonvolatile memory devices, and described method comprises:

Data are write the first storage unit;

Data in described the first storage unit are write the second storage unit.

2. control method according to claim 1, it is characterized in that, described solid state hard disc also comprises the 3rd storage unit, and described the 3rd storage unit is volatile storage, and the writing speed of described the first storage unit is greater than the writing speed of described the second storage unit; Wherein,

Described data are write the first storage unit, comprising:

Data are write the 3rd storage unit;

During described solid state hard disc power down, the described data in described the 3rd storage unit are write the first storage unit;

Described data in described the first storage unit are write the second storage unit, comprising:

After described solid state hard disc powers on, the described data in described the first storage unit are write described the 3rd storage unit;

Described data in described the 3rd storage unit are write the second storage unit.

3. control method according to claim 1 is characterized in that, described data are write the first storage unit before, also comprise:

Whether the data buffer area of judging described the first storage unit is full, if so, described data write described the second storage unit; If not, described data are write the first storage unit.

4. control method according to claim 1, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described data in described the first storage unit are write the second storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the second storage unit.

5. control method according to claim 2, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the 3rd storage unit; Described data are write the 3rd storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the 3rd storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the 3rd storage unit.

6. control method according to claim 5 is characterized in that, described map information also comprises erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described described data in described the 3rd storage unit are write the second storage unit, comprising:

Scan the map information in described the first storage unit, find minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

According to the described mapping relations that find, described data are write the minimum physical block of erasing times in described the second storage unit.

7. control method according to claim 1 and 2 is characterized in that, described method also comprises:

Update stored in the physical address of physical block in described the first storage unit in described the first storage unit and/or described the second storage unit and the mapping relations of logical address.

8. control method according to claim 1 and 2, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the first storage unit; Described method also comprises:

In described map information, the historical physical block of storing described data in described the first storage unit is labeled as invalid block.

9. the write operation device of a solid state hard disc is characterized in that, described solid state hard disc comprises the first storage unit and the second storage unit, and wherein said the first storage unit and described the second storage unit are Nonvolatile memory devices, and described device comprises:

The first writing unit is used for data are write the first storage unit;

The second writing unit writes the second storage unit with the data in described the first storage unit.

10. device according to claim 9, it is characterized in that, described solid state hard disc also comprises the 3rd storage unit, and described the 3rd storage unit is volatile storage, and the writing speed of described the first storage unit is greater than the writing speed of described the second storage unit; Wherein,

Described the first writing unit comprises:

First writes subelement, is used for data are write the 3rd storage unit;

Second writes subelement, is used for when described solid state hard disc power down the described data in described the 3rd storage unit being write the first storage unit;

Described the second writing unit comprises:

The 3rd writes subelement, is used for after described solid state hard disc powers on the described data in described the first storage unit being write described the 3rd storage unit;

The 4th writes subelement, is used for the described data of described the 3rd storage unit are write the second storage unit.

11. device according to claim 9 is characterized in that, described system also comprises:

The first judging unit is used for judging whether the data buffer area of described the first storage unit is full.

12. device according to claim 9, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; Described the second writing unit comprises:

The first scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

The 5th writes subelement, is used for the described mapping relations that described the first scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the second storage unit.

13. device according to claim 10, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the 3rd storage unit; Described first writes subelement comprises:

The second scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the 3rd storage unit;

The 6th writes subelement, is used for the described mapping relations that described the second scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the 3rd storage unit.

14. device according to claim 13 is characterized in that, described map information also comprises erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the second storage unit; The described the 4th writes subelement comprises:

The 3rd scanning element, the map information for scanning described the first storage unit finds minimum mapping relations corresponding to physical block of erasing times in described the second storage unit;

The 7th writes subelement, is used for the described mapping relations that described the 3rd scanning element of foundation finds, and described data are write the minimum physical block of erasing times in described the second storage unit.

15. according to claim 9 or 10 described devices, it is characterized in that described system also comprises:

Updating block is for the physical address of physical block and the mapping relations of logical address in described the first storage unit that updates stored in described the first storage unit and/or described the second storage unit.

16. according to claim 9 or 10 described devices, it is characterized in that, described the first cell stores map information, described map information comprise erasing times and the physical address of physical block and the mapping relations between the logical address of physical block in described the first storage unit; Described system also comprises:

Indexing unit is used at described map information, and the historical physical block of storing described data in described the first storage unit is labeled as invalid block.

Images