CN105808150B - Solid state disk fast taking system for hybrid storage facilities - Google Patents
Solid state disk fast taking system for hybrid storage facilities Download PDFInfo
- Publication number
- CN105808150B CN105808150B CN201410834245.4A CN201410834245A CN105808150B CN 105808150 B CN105808150 B CN 105808150B CN 201410834245 A CN201410834245 A CN 201410834245A CN 105808150 B CN105808150 B CN 105808150B
- Authority
- CN
- China
- Prior art keywords
- data
- solid state
- block
- state disk
- cache
- 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.)
- Active
Links
Landscapes
- Memory System Of A Hierarchy Structure (AREA)
Abstract
The invention discloses a kind of solid state disk fast taking systems for hybrid storage facilities.The system includes：One solid state disk, to store caching data, which is divided into a repeat pattern cache area and a dynamic replacement cache area；And a cache management module, including：One input/output analytic unit to the block of a hard disk into the input/output request of line access, and sequentially stores the first data of corresponding first block to the repeat pattern cache area for detecting in multiple continuous detection time intervals；And one dsc data search unit, for detecting in an independent detection time interval, to a hard disk into the input/output request of line access, and sequentially store the second data of corresponding second block to the dynamic replacement cache area.
Description
Technical field
The present invention relates to a kind of solid state disk fast taking systems, more particularly to a kind of solid-state for hybrid storage facilities
Hard disk fast taking system can have higher performance by storing configuration mechanism.
Background technology
Traditionally, excessively high defeated caused by arbitrary access workload in order to solve in data center or enterprise
Enter/delay issue is exported, large-scale and excessive amounts of magnetic hard-disk may be disposed.Thus, more hard disk read heads
It can reduce by two probability continuously read on same disk, and then improve the performance integrally accessed on hard disk.However, because
Excessive deployment and there are a variety of disadvantages, for example, more and more protect equipment, exceptional space to use, system operation it is more
Power supply and cooling system and high maintenance cost etc..In addition, excessive capacity increase causes the utilization rate of system that may subtract
It is few.
Recently, a kind of solution widely adopted is to take the quick reading of solid state disk (random or sequence) cooperation hard
The operation of disk.It is obvious that being a hybrid storage facilities using the stocking system of such solution.Solid state disk is used as
A kind of function of cache.That is, only dsc data (data being most often used) is temporarily stored into solid state disk, access is provided.
Once the data no longer so " heat " of these storages, they will be removed and original space is left other dsc datas for and used.
Many non-dsc datas can be stored in hard disk.Such stocking system has many benefits, first, for spike demand, Gu
State hard disk cache can provide more preferably spiking behaviour numerical value.Secondly, solid state disk cache can be under virtual stocking system environment, altogether
It is shifted between the workload used.Furthermore it allocates dsc data in advance to do flexibly performance requirement to solid state disk cache
Response.Under the background of solid state disk cache transfer, solid state disk cache can discharge the data being soon not used in advance, to move
Turn solid state disk and get other workloads soon, other workloads should be able to be predicted or periodically occur.Really, if
It can predict that the peak period time of individual work load, solid state disk cache also can only operate on single workload.
Solid state disk can become fast taking system with random access memory.Although having when cache and hard disk operate together exhausted
Good storge quality, solid state disk cache do not defer to the principle of traditional random access memory cache but.Its reading is far faster than writing
Enter, sequence read/write is faster than random read take/write-in.Most of all, solid state disk can lose after the write-in of certain number
Effect, thus when the such hybrid storage facilities of operation, the write-in opportunity for controlling solid state disk cache is most important.
Many existing patents provide new technology to meet the above demand.For example, U.S. Patent Publication No.
No. 20140244959 disclose a kind of stocking system having for different kenel storage facilities control means.The stocking system
Including：One hard disc storage equipment；One solid state disk storage facilities；And a storage facilities controller, the wherein storage facilities control
Device collects the load information about multiple each loads in area in the hard disc storage equipment.It can be based on the load letter that those are collected
Breath selects a migration candidate regions in hard disc storage equipment, and the Data Migration in the candidate regions of the selection is hard to solid-state
In disk storage facilities.Storage facilities controller collects the counting of input/output request in per unit time as the load information,
It selects candidate regions also based on the average life span being delayed from the load.The average life span can be further by subtracting in hard disc storage
The load in each area of equipment is calculated by the time., it is clear that the selection for moving to solid-state hard disc equipment data is to be based on
The lasting delay of load information.The system does not include being repeated periodically load, those loads may not continue to grow very much one section
Time, but containing regularly with frequent request mode.This repeated boad access from hard disk can absolutely injure the storage
The performance of deposit system.
U.S. Patent Publication No. the 20140258668th provides in hybrid data storage system for managing storage
The System and method in space.The System and method for of the application is based on data source, data types, data function or other similar ginsengs
Number intelligently distributes data to solid state disk (or other relatively high performance disks) and other Data Holding Equipments, such as firmly
In disk.Smart allocation between an at least solid state disk and at least one other disk forms, allows through the effective of storage space
Using promoting system performance.It is concentrated in a hybrid data storage, which can adaptively choose from the option of link
Most fast storage facilities, while maintaining maximum performance and most effectively low volume data write-in.In addition, data are assigned to hybrid number
May be automatically to control or specifically by user setting according to the storage facilities in stocking system.Although many factors are all considered
Data are distributed, but the hit rate of solid state disk can not improve, this further restricts the uses of solid state disk.
Therefore, in order to efficiently operate hybrid storage facilities, it is a kind of be used for controlling be used together with other hard disks
Solid state disk fast taking system is in demand.In particular, the system be operable to improve the hit rate that uses of solid state disk and
Reach higher performance.
Invention content
To solve the above-mentioned problems, it according to the aspect of the present invention, discloses a kind of for hybrid storage facilities
Solid state disk fast taking system.The system includes a solid state disk and a cache management module, wherein the solid state disk is storing
Caching data, the solid state disk are divided into a repeat pattern cache area and a dynamic replacement cache area；And the cache management module packet
Unit is searched containing an input/output analytic unit and a dsc data, wherein the input/output analytic unit is for detecting more
In a continuous detection time interval, to the block of a hard disk into the input/output request of line access, and sequentially storage corresponding first
First data of block are to the repeat pattern cache area, and in each continuous detection time interval, which is at least weighed
Access multiple is secondary；And the dsc data searches unit for detecting in an independent detection time interval, to a hard disk into line access
Input/output request, and sequentially store the second data of corresponding second block to the dynamic replacement cache area, independently examine at this
It surveys in time interval, which is at least accessed secondary.
According to the present invention, which further includes a write-in hard disc module, if first data
It is asked by write instruction and when the input/output analytic unit finds the repeat pattern cache area space quilt of one first percentage
When occupied by data, directly to store in first data a to hard disk.First percentage is higher than for 90% or any
90% percentage.If second data are asked by write instruction and the input/output analytic unit finds one the 200th
When the space in the dynamic replacement cache area of point ratio is occupied by data, the write-in hard disc module be further used for directly storing this
In two data a to hard disk.Second percentage is the 90% or any percentage for being higher than 90%.No matter the first and second percentages
Than be all by the Configuration Values set by system manager and consider system performance, SLA (Service Level Agreement) or
QoS (Quality of Service), 90% or preferable higher than 90% numerical value.
Solid state disk fast taking system further includes a RAM module, which includes
One area's Virtual table, wherein area's Virtual table are a mapping table, the block shape for keeping track first block and the second block
State；If correspondence image block of the logical blocks in area's Virtual table in the solid state disk belongs to dynamic replacement cache area,
Logical blocks in the areas Ze Virtual table are denoted as dynamic replacement cache block；If the logical blocks in area's Virtual table exist
Correspondence image block in the solid state disk belongs to repeat pattern cache area, and the logical blocks in the areas Ze Virtual table are denoted as
Repeat pattern cache block；When the input/output analytic unit finds specific second number in the dynamic replacement cache area
When according to becoming first data, then image to the dynamic replacement cache area of specific second data in the dynamic replacement cache area
Block is denoted as repeat pattern cache block, and the dynamic replacement cache area Central Plains data content does not make any change.
In the case of preference, when the input/output analytic unit finds that the first block of a group is repeated access, under
Before secondary first block is accessed, corresponding first data will be stored into the repeat pattern cache area.When the input/output
When analytic unit finds that the first block of a group is not repeated access, after in next time, first block is accessed, corresponding first number
According to will be removed from the repeat pattern cache area.When the first new data of a group and/or the second data predicting are special following one
A meeting of fixing time is accessed, the input/output analytic unit remove a part of space in the repeat pattern cache area and/or
A part of space in the dynamic replacement cache area.One cache replacement algorithm is searched unit by the dsc data and is used, with clear
Except a part of the second data in dynamic replacement cache area.The cache replacement algorithm is minimum use (Least recently
Recently Used, LRU) algorithm, it is at least relatively often used (Least Frequently Used, LFU) algorithm, or adaptively replace
Change cache (Adaptive Replacement Caching, ARC) algorithm.By the dsc data search in the dynamic replacement cache area
Algorithm used in unit selected by the different occupation mode of storage facilities, or is stopped from those cache replacement algorithms
Only it is applied to reading or write instruction.When hybrid storage facilities encounter routinely, input and output operation number height per second with
When cache hit rate is low, for avoid it is inefficient can with meaningless cache, the dsc data search unit stop storage by read instruct
Second data of request.The cache management module and the form of write-in hard disc module are an adapter in server or one connect
Device, or the software that is installed in server.
The present invention utilizes the advantages of repeat pattern cache area detached in solid state disk is with dynamic replacement cache area, respectively stores up
Deposit the data that (or cache) is repeated and be commonly used.Other solid state disk cache mechanism may lose the data of reuse and
It does not get soon, the efficiency of integer composite storage facilities can be reduced in this way.To which the present invention smooth can efficiently operate a set of
Hybrid storage facilities.Most of all, the hit rate of solid state disk cache can improve.
Description of the drawings
Fig. 1 is the schematic diagram for the fast taking system of hybrid storage facilities according to the present invention；
Fig. 2 is the record that block is accessed in continuous detection time interval；
Fig. 3 is the record that block is accessed in independent detection time interval；
Fig. 4 is a kind of workload distribution in the hybrid storage facilities of ordinary day；
Fig. 5 is another workload distribution in the hybrid storage facilities of ordinary day.
Reference sign：10- fast taking systems；The hybrid storage facilities of 20-；100- solid state disks；102- repeat patterns
Cache area；104- dynamic replacement caches area；200- servers；202- cache management modules；2022- input/output analytic units；
2024- dsc datas search unit；Hard disc module is written in 204-；206- RAM modules；The areas 2062- Virtual table；302-
Hard disk；304- hard disks；401- applied host machines；402- applied host machines；The area A1, A2, A3, B1, B2, B3, B4, C1, C2, C3, C4-
Block.
Specific implementation mode
The present invention will be more specifically described with reference to following embodiment.
Referring to Fig. 1, one embodiment of the invention is illustrated by the figure.One fast taking system 10 is used for improving one hybrid
The performance of storage facilities 20.In the present embodiment, hybrid storage facilities 20 by a solid state disk 100 and two hard disks 302 with
304 compositions.According to the spirit of the present invention, the quantity of solid state disk and hard disk is not limited to one or two, according to a system
The demand operated is needed, they can be any quantity.Hybrid storage facilities 20 may be present in the form of disk array
In data center, it can also be only a solid state disk in single host mix dish (Solid State Hard Drive,
SSHD).In the present embodiment, which is a disk array.
Fast taking system 10 includes solid state disk 100 and a server 200, and fast taking system 10 is by change solid state disk 100
Originally operating type and reach.Therefore, any solid state disk with following data structure in hybrid storage facilities, all
It is considered the practice of the present invention.Solid state disk 100 is divided into the areas Liang Ge：One repeat pattern cache area 102 and a dynamic replacement cache
Area 104.Solid state disk 100 is used for storing caching data, and the not all data from applied host machine 401 and 402 write instructions.
Repeat pattern cache area 102 and dynamic replacement cache area 104 can also be used to storage caching data, only different classes of fast
Access evidence.
Server 200 includes a cache management module 202, one write-in hard disc module 204 and a random access memory mould
Block 206, each single item in these elements all will be described under.Cache management module 202 includes input/output analysis
Unit 2022 and a dsc data search unit 2024.Input/output analytic unit 2022 can be detected to hard disk 302 or 304 inner regions
Input/output request of the block into line access.That is input/output analytic unit 2022 detects hard disk 302 or 304 for this
Write instruction and reading instruction and its relevant block.The detection process continues multiple continuous detections before the time instantly
Time interval.If in each continuous detection time interval, there are the first data of corresponding first block, wherein the first block is extremely
Be repeated less access it is secondary, then those first data are sequentially stored into repeat pattern cache area 102.In order to have preferably to this point
Understanding, refer to Fig. 2.
Fig. 2 is the record for being accessed block.The block array in hard disk 302 is shown on the left of Fig. 2, and 2 areas are only painted in figure
Block is illustrating.In fact, a hard disk may include thousands of blocks, Fig. 2 be example for illustration purposes, be not intended to limit
The present invention.Certain blocks are denoted as A1, A2, A3, B1, B2, B3, B4, C1, C2, C3 and C4, they, which are detected, was once deposited
It took (herein, it will be apparent that " access " refers to that the reading received instructs, and is not write instruction.In general, reading refers to
It enables and is suitable for the present invention with write instruction).A1 to A3 represents a complete data.B1 and B2 indicates a data and B3 and B4
Indicate another data, the two data are accessed simultaneously.C1 to C4 represents another complete data.Right side description block access
Sequential.Time t0 indicates present.Ta1, ta2, ta3 are past time points with ta4.Between any two adjacent time points when
Between interval be exactly continuous detection time interval, such as between ta2 and ta3.It can see that A1, A2, be accessed between ta1 to t0 with A3
3 times, B1, B2, B3 and B4 occur 2 times, respectively between ta3 to ta2 and ta2 to ta1.C1, C2, C3 and C4 are in each continuous detection
All occur in time interval.If it is the first block to be defined on each all blocks being accessed in continuous detection time interval,
Corresponding first data can arrive repeat pattern cache area 102 by sequentially cache (storage), then C1, C2, C3 and C4 are first
Block.
If it is the first block to be defined on the block that is all accessed in 3 continuous detection time intervals, corresponding first
Data can arrive repeat pattern cache area 102 by sequentially cache (storage), that B1, B2, B3, B4, C1, C2, C3 and C4 are first
Block.These first blocks are sequentially stored in repeat pattern cache area 102 to be used so that the following sequence is read.Because of the invention
It is required that those blocks in each continuous detection time interval, should at least be repeated access twice, although A1, A2, with A3 in ta1
It is accessed to being concentrated between t0, they are not the first blocks.Meanwhile minimum access times can be set according to any hybrid storage
Change for 20 performance requirement, continuous detection time interval depends on workload and service level agreement (Service
Level Agreement, SLA) or service quality (Quality of Service, QoS) demand in definition.In general, it can
To be the several seconds (hard work loads and superior performance demand) to arrive dozens of minutes (light work loads and low performance demand).
Dsc data searches unit 2024 and is used for detecting in an independent detection time interval, to a hard disk into the defeated of line access
Enter/export request.Similarly, dsc data searches write instruction and the reading instruction that unit 2024 detects hard disk 302 or 304,
And its relevant block.Different from input/output analytic unit 2022, dsc data searches unit 2024 and sequentially stores corresponding second
In second data of block to dynamic replacement cache area 104, wherein the second block in the independent detection time interval at least by
It accesses secondary.Please refer to Fig. 3.By taking identical access block in hard disk in Fig. 2 302 as an example, if independent detection time interval is
(time of ta4 to ta2 is equal to the time of tb2 to tb1, and the time of ta2 to t0 is equal to tb1 for twice of continuous detection time interval
To the time of t0), according to definition, in independent detection time interval tb2 to tb1, C1, C2, C3 are the second block with C4, and
In independent detection time interval tb1 to t0, A1, A2, A3, C1, C2, C3 are the second block with C4.Certainly, independent detection
Time interval needs not be twice of continuous detection time interval, independent detection time interval and continuous detection time compartment
There can be any multiple proportion.
If the first data are asked by write instruction and input/output analytic unit 2022 finds repeat pattern cache area
102 is soon full, and write-in hard disc module 204 can directly store in the first data a to hard disk 302 or 304.As discussed above, it stores up
Write instruction is may be from there are first data in repeat pattern cache area 102 or reads instruction.To write caching, because of the first number
According to will finally be written in hard disk 302 or 304, certain first data can not be stored into repeat pattern cache area 102 and may not damage
The performance of harmful hybrid storage facilities 20.This is because the first data for being directly stored in hard disk 302 or 304 may be in the future not
It can often be accessed.Thus, if repeat pattern cache area 102 is rapidly written into or the repeat pattern of one first percentage
When 102 space of cache area is occupied by data, the first data occurred later will be directly stored into hard disk 302 or 304.Implementation
On, the first percentage can be the 90% or any percentage for being higher than 90%.Similarly, if the second data are write instruction institute
Request and input/output analytic unit 2022 find that dynamic replacement cache area 104 is rapidly written into or one second percentage
When 104 space of dynamic replacement cache area is occupied by data, write-in hard disc module 204 also can directly store the second data to hard disk
In 302 or 304.Second percentage can also be the 90% or any percentage for being higher than 90%.No matter the first and second percentages
All it is to consider system performance, service level agreement or QoS requirement together by the Configuration Values set by system manager.
90% or preferable higher than 90% numerical value.
In order to improve the efficiency in repeat pattern cache area 102 and dynamic replacement cache area 104, RAM module
206 apply area's Virtual table 2062.Area's Virtual table 2062 is a mapping table, (is repeated for keeping track solid state disk 100
Pattern cache area 102 and dynamic replacement cache area 104) Block status.If the logical blocks of Virtual table 2062 are solid in area
Correspondence image block in state hard disk 100 belongs to dynamic replacement cache area 104, and in area, the logical blocks of Virtual table 2062 are labeled
For dynamic replacement cache block.On the other hand, if area's Virtual table 2062 correspondence of the logical blocks in solid state disk 100
Image block belongs to repeat pattern cache area 102, and in area, the logical blocks of Virtual table 2062 are denoted as repeat pattern cache area
Block.When input/output analytic unit 2022 finds that specific 1 second data 104 become the first number in the dynamic replacement cache area
According to, it is not necessary to it removes the data from dynamic replacement cache area 104 and is rewritten in repeat pattern cache area 102.Image is arrived
The dynamic replacement cache block of specific second data of this in dynamic replacement cache area 104, they can only be attached most importance to by directly mark
Complex pattern cache block.Specific second data 104 do not make any change in dynamic replacement cache area.Because of random access memory
Running speed is faster than solid state disk 100, hybrid storage facilities 20 must with slower speed to repeat pattern cache area 102 with
Dynamic replacement cache area 104 modifies, and executing small change to RAM module 206 is absolutely beneficial to hybrid storage
Equipment 20.Therefore the performance of hybrid storage facilities 20 can be promoted.
It should be noted that a kind of searching cache replacement algorithm used by unit 2024 for removing a part by dsc data
The second data being stored in dynamic replacement cache area 104.This cache replacement algorithm is used for removing in dynamic replacement cache area 104
Certain less the second data used can be cached so as to the second data of new demand in dynamic replacement cache area 104.Have several
Kind cache replacement algorithm can use.For example, it is minimum use recently (Least Recently Used, LRU) algorithm, at least compared with
It is often used (Least Frequently Used, LFU) algorithm, or adaptively (Adaptive Replacement are fast for replacement cache
Take, ARC) algorithm, different operational versions need different algorithms.A kind of dsc data search by dynamic replacement cache area 104
Algorithm used in unit 2024 be from those cache replacement algorithms, selected by the different occupation mode of storage facilities,
Or stop applied to reading or write instruction.
According to different occupation modes, each algorithm can show property more better than other algorithms in certain behaviours in service
Energy.Have the problem that the situation that hybrid storage facilities 20 does not know that request occurs, there is also too many variables in operating environment.
But particular phenomenon can show that most recently used cache replacement algorithm is useless.For example, if hybrid storage facilities 20 (or
Dynamic replacement cache area 104) encounter routinely, input and output operation number per second is high and cache hit rate is low, this is just implied
The cache replacement algorithm for the application is an invalid selection.Thus, dsc data search unit 2024 will stop storage and come from
Read the second data of instruction request, instead another cache replacement algorithm.That is cache is read temporarily to stop
Only and write caching is unaffected.
According to the spirit of the present invention, before certain predictable (or repetition) situations occur, input/output analytic unit 2022
The caching data in dynamic replacement cache area 104 can be adjusted.See Fig. 4, which is shown in the hybrid storage facilities of ordinary day 20
Workload distribution situation.A workload from applied host machine 401 and 402 shows a kind of workload sharply increased
It is present approximately in 9：00, this may be to be reached the standard grade work by employee, be asked for caused by data to server 200.A group first
Block D was detected in this stage, can all be repeated access daily on ordinary days.Certainly, if first block of group D is stored in firmly
It in disk 302, when people's requirement accesses it and does not make any changes, input and output operation number per second will be got higher, and be postponed
It can occur.Therefore, when input/output analytic unit 2022 finds that first block of group D repeats to be accessed, next time this first
Before block is accessed, corresponding first data will be stored into repeat pattern cache area 102.For example, the first data will be flat
Day 8：30 are stored into repeat pattern cache area 102.That is it is based on input/output analytic unit 2022, data pre-fetching goes out
Into repeat pattern cache area 102, rather than dsc data is waited for search the cache replacement algorithm dynamic update of unit 2024 dynamically
Replace cache area 104.
On the contrary, if discovery the first block of a group of input/output analytic unit 2022 is not repeated access " or not, (they must
Repeatedly it is accessed at previous certain moment), after in next time, first block is accessed, corresponding first data will be by repeating mould
It is removed in formula cache area 102.Referring to Fig. 5, the figure shows another workload of hybrid storage facilities 20.Will be apparent that should
Workload is 9：00 to 21：00 is quite heavy.During this period, the first block of a group E, which is observed, is not accessed.Cause
This, first block of group E will be removed.For example, he will in the next day 8：It is removed after 50 points.It can be provided by removing storage data in advance
Its workload more cache spaces in dynamic replacement cache area 104.
The above-mentioned example of the present embodiment is the data detected based on history.If workload can be predicted or be given, this hair
It is bright to be also suitable.In other words, when the first new data of a group and/or the second data are predicted to click through in following specific time
Line access, input/output analytic unit 2022 can remove the space in a part of repeat pattern cache area 102 and/or a part is moved
State replaces the space in cache area 104.Thus, new the first data and/or the second data can be stored in the time point of the prediction.It is any
Being capable of providing suitable method, algorithm or the module serviced in this way can use.It is preferred special in the U.S. by identical inventor
A kind of storage facilities discharge model provided in sharp application case the 14/290th, 533 can obtain same technique by this application
Collective reference.The size in space is removed in repeat pattern cache area 102 and/or dynamic replacement cache area 104 to be depended on
Workload.
It should be noted that in the present embodiment, the form of cache management module 202 and write-in hard disc module 204 is server
An adapter in 200 or a connector.In fact, they also can be the software to be installed in server 200 and realize.
Although the present invention is disclosed above with embodiment, however, it is not to limit the invention, any affiliated technology neck
Have usually intellectual in domain, without departing from the spirit and scope of the present invention, when can make some changes and embellishment, therefore this
Subject to the protection domain of invention ought be defined depending on appended claims range.
Claims (13)
1. a kind of solid state disk fast taking system for hybrid storage facilities, which is characterized in that it includes a solid state disk and
One cache management module, wherein
For the solid state disk to store caching data, which is divided into a repeat pattern cache area and a dynamic replacement cache
Area；And
The cache management module includes that an input/output analytic unit and a dsc data search unit, wherein
The input/output analytic unit deposits the block of a hard disk for detecting in multiple continuous detection time intervals
The input/output request taken, and the first data for corresponding to the first block are sequentially stored to the repeat pattern cache area, in each company
In continuous detection time interval, it is secondary which is at least repeated access；And
The dsc data search unit for detect in an independent detection time interval, to a hard disk into line access input/it is defeated
Go out request, and sequentially stores the second data for corresponding to the second block to the dynamic replacement cache area, between the independent detection time
In, which is at least accessed secondary.
2. solid state disk fast taking system as described in claim 1 further includes a write-in hard disc module, if first number
The sky that the input/output analytic unit finds the repeat pattern cache area of one first percentage is asked and works as according to by write instruction
Between occupied by data when, directly to store in first data a to hard disk.
3. solid state disk fast taking system as claimed in claim 2, wherein first percentage are higher than 90% for 90% or any
Percentage.
4. solid state disk fast taking system as claimed in claim 2, wherein if second data asked by write instruction and
When the input/output analytic unit finds that the space in the dynamic replacement cache area of one second percentage is occupied by data, this is write
Enter hard disc module further directly to store in second data a to hard disk.
5. solid state disk fast taking system as claimed in claim 4, wherein second percentage are higher than 90% for 90% or any
Percentage.
6. solid state disk fast taking system as claimed in claim 2, wherein the cache management module are clothes with write-in hard disc module
The adapter or a connector being engaged in device, or the software that is installed in server.
7. solid state disk fast taking system as described in claim 1, further includes a RAM module, this is deposited at random
It includes area's Virtual table to take memory modules, wherein area's Virtual table be a mapping table, for keep track first block with
The Block status of second block；If correspondence image block of the logical blocks in area's Virtual table in the solid state disk belongs to
Dynamic replacement cache area, the logical blocks in the areas Ze Virtual table are denoted as dynamic replacement cache block；If virtual in area
Correspondence image block of the logical blocks in the solid state disk in table belongs to repeat pattern cache area, in the areas Ze Virtual table
Logical blocks are denoted as repeat pattern cache block；When the input/output analytic unit is found in the dynamic replacement cache area
In specific second data when becoming first data, then image to specific second data in the dynamic replacement cache area
Dynamic replacement cache block be denoted as repeat pattern cache block, and the dynamic replacement cache area Central Plains data content is not made
Any change.
8. solid state disk fast taking system as described in claim 1, wherein when the input/output analytic unit finds a group first
When block is repeated access, before in next time, first block is accessed, corresponding first data will be stored into the repeat pattern
In cache area.
9. solid state disk fast taking system as described in claim 1, wherein when the input/output analytic unit finds a group first
When block is not repeated access, after in next time, first block is accessed, corresponding first data will be from the repeat pattern cache
It is removed in area.
10. solid state disk fast taking system as described in claim 1, wherein when a group new the first data and/or the second data
It is contemplated that when following particular point in time will take when will be deposited, then the input/output analytic unit is removed in the repeat pattern
A part of space in cache area and/or a part of space in the dynamic replacement cache area.
11. solid state disk fast taking system as described in claim 1 a, wherein cache replacement algorithm searches unit by the dsc data
It is used, to remove the second data of a part in dynamic replacement cache area.
12. solid state disk fast taking system as claimed in claim 11, the wherein cache replacement algorithm are minimum recently using calculation
Method is at least relatively often used algorithm or adaptive replacement cache algorithm.
13. solid state disk fast taking system as claimed in claim 11, wherein when hybrid storage facilities encounters routinely, often
When second input and output operation number height is low with cache hit rate, which searches unit and stops storage by reading instruction request
Second data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410834245.4A CN105808150B (en) | 2014-12-29 | 2014-12-29 | Solid state disk fast taking system for hybrid storage facilities |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410834245.4A CN105808150B (en) | 2014-12-29 | 2014-12-29 | Solid state disk fast taking system for hybrid storage facilities |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105808150A CN105808150A (en) | 2016-07-27 |
| CN105808150B true CN105808150B (en) | 2018-09-07 |
Family
ID=56980498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410834245.4A Active CN105808150B (en) | 2014-12-29 | 2014-12-29 | Solid state disk fast taking system for hybrid storage facilities |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105808150B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10795583B2 (en) | 2017-07-19 | 2020-10-06 | Samsung Electronics Co., Ltd. | Automatic data placement manager in multi-tier all-flash datacenter |
| US10496541B2 (en) | 2017-11-29 | 2019-12-03 | Samsung Electronics Co., Ltd. | Dynamic cache partition manager in heterogeneous virtualization cloud cache environment |
| US11334274B2 (en) | 2018-02-09 | 2022-05-17 | Seagate Technology Llc | Offloaded data migration between storage devices |
| CN117041256B (en) * | 2023-10-08 | 2024-02-02 | 深圳市连用科技有限公司 | Network data transmission and storage method and system |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003216460A (en) * | 2002-01-21 | 2003-07-31 | Hitachi Ltd | Hierarchical storage device and its controller |
| CN101458613B (en) * | 2008-12-31 | 2011-04-20 | 成都市华为赛门铁克科技有限公司 | Method for implementing mixed hierarchical array, the hierarchical array and storage system |
| US8161241B2 (en) * | 2010-01-12 | 2012-04-17 | International Business Machines Corporation | Temperature-aware buffered caching for solid state storage |
| CN102129472B (en) * | 2011-04-14 | 2012-12-19 | 上海红神信息技术有限公司 | Construction method for high-efficiency hybrid storage structure of semantic-orient search engine |
| CN103186350B (en) * | 2011-12-31 | 2016-03-30 | 北京快网科技有限公司 | The moving method of mixing storage system and hot spot data block |
| CN103399713B (en) * | 2013-08-02 | 2016-01-20 | 浙江大学 | The data buffering method of balance dynamic data attemper performance and solid-state disk service life |
-
2014
- 2014-12-29 CN CN201410834245.4A patent/CN105808150B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN105808150A (en) | 2016-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8782369B2 (en) | Apparatus to manage efficient data migration between tiers | |
| EP2519883B1 (en) | Efficient use of hybrid media in cache architectures | |
| US8909887B1 (en) | Selective defragmentation based on IO hot spots | |
| JP3962368B2 (en) | System and method for dynamically allocating shared resources | |
| CN105808150B (en) | Solid state disk fast taking system for hybrid storage facilities | |
| US8285936B2 (en) | Cache memory with power saving state | |
| US20160179683A1 (en) | Ssd caching system for hybrid storage | |
| CN103019962B (en) | Data buffer storage disposal route, device and system | |
| CN104317742B (en) | Thin provisioning method for optimizing space management | |
| CN107018172A (en) | System and method for the adaptive partition in distributed caching memory | |
| ATE422688T1 (en) | STRATEGIES FOR READING INFORMATION FROM A MASS STORAGE MEDIUM USING A CACHE MEMORY | |
| CN105027093A (en) | Methods and apparatus for compressed and compacted virtual memory | |
| JP6402647B2 (en) | Data arrangement program, data arrangement apparatus, and data arrangement method | |
| CN103702057A (en) | Block storage algorithm applicable to multiple paths of concurrent-written stream media data | |
| CN104111896B (en) | Virtual memory management method and its device in big data processing | |
| CN1928804A (en) | Method and system for power management in a distributed file system | |
| CN109154933B (en) | Distributed database system and method for distributing and accessing data | |
| US9699254B2 (en) | Computer system, cache management method, and computer | |
| CN103699681B (en) | The treating method and apparatus of data rewind | |
| CA2415018C (en) | Adaptive parallel data clustering when loading a data structure containing data clustered along one or more dimensions | |
| WO2019103647A1 (en) | Method for performing read-ahead in data storage systems | |
| CN107256235A (en) | A kind of caching system hot spot data access method | |
| CN102779017A (en) | Control method of data caching area in solid state disc | |
| US20200401529A1 (en) | Gpu cache management based on locality type detection | |
| JP2019040343A (en) | Information processing apparatus, process management method, and process management program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1226833 Country of ref document: HK |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |