CN107544747A - Abrasion inverse equalization methods and device - Google Patents
Abrasion inverse equalization methods and device Download PDFInfo
- Publication number
- CN107544747A CN107544747A CN201610498832.XA CN201610498832A CN107544747A CN 107544747 A CN107544747 A CN 107544747A CN 201610498832 A CN201610498832 A CN 201610498832A CN 107544747 A CN107544747 A CN 107544747A
- Authority
- CN
- China
- Prior art keywords
- bulk
- storage device
- storage
- life
- span
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000005299 abrasion Methods 0.000 title abstract description 14
- 238000003860 storage Methods 0.000 claims abstract description 366
- 230000008859 change Effects 0.000 claims description 3
- 230000014759 maintenance of location Effects 0.000 claims 1
- 230000004224 protection Effects 0.000 description 14
- 230000009758 senescence Effects 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000000151 deposition Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 7
- 238000009826 distribution Methods 0.000 description 6
- 230000000717 retained effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000007689 inspection Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 210000000352 storage cell Anatomy 0.000 description 2
- 206010011906 Death Diseases 0.000 description 1
- 206010064930 age-related macular degeneration Diseases 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 208000002780 macular degeneration Diseases 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Techniques For Improving Reliability Of Storages (AREA)
Abstract
Provide abrasion inverse equalization methods and device.It is balanced to realize that abrasion is intended in storage object is created.The method of the establishment storage object provided, including：Select the M bulk from N number of storage device；If the life-span for providing at least the first storage device and the second storage device in the storage device of M bulk exceedes first threshold, the first bulk from the first storage device is rejected from the M bulk, and M bulk for selecting the second bulk to replace first bulk and after update so that the difference for providing the life-span of the 3rd storage device of the second bulk and the life-span of the second storage device is no more than first threshold more than the life-span of the 3rd storage device of the second bulk of Second Threshold or offer；And create storage object using M bulk after renewal.
Description
Technical field
The present invention relates to storage system, more particularly, within the storage system to the driver implementation mill of restricted lifetime
The inverse balanced method and system of damage.
Background technology
Existing RAID (Redundant Arrays of Independent Disks, RAID) skill
Multiple phisical drives (for example, disk) are aggregated into disk group by art, and band is divided in disk group, and each bar is taken by redundancy skill
Art provides protection to data.In the RAID system of prior art, when one or more of multiple drivers break down, it can open
Dynamic spare drive, and data reconstruction is carried out, to keep the data protection ability of RAID system and performance.In failed drive
It is replaced rear, it is necessary to which the data duplication in spare drive is returned into new driver again.RAID system can provide a variety of data protections
Rank.Conventional RAID has RAID0, RAID1, RAID5, RAID6.Different RAID algorithms can tolerate the quantity of low-quality disk
It is different.Broken down for example, RAID5 systems can correct a block drive, and RAID6 systems can correct the generation of two block drives
Failure.
Traditional RAID technique solves the problems, such as data protection, but the problem of bring be reconstruct data speed it is slower.Cause
Be likely to read data from other all disks to reconstruct the process of data, calculate recover to write after data one piece it is new
On driver, therefore new driver will turn into the performance bottleneck of data recovery procedure.In order to solve this problem, and to depositing
Storage resource, which is realized, more flexibly and effectively to be managed, and proposes the thinking SRM and data protection division.I.e. first drive
Storage resource on dynamic device is cut into small storage cell, then by the storage cell on different driving device with certain algorithm organization
Cheng Geng little but more flexible RAID groups.Thus when a phisical drive is damaged, reconstructing the data come, just have can
It can write on other different drivers, performance bottleneck caused by so as to avoid single driver.
RAID group technologies realize the distribution according to need of resource, improve the utilization ratio of storage.The group of each RAID group
No longer it is whole driver into unit, and simply a part of storage resource on driver.RAID groups are nor static state creates
, but distribute according to need.
The storage resource of consideration reasonable employment driver is needed in RAID group technologies, is avoided to one or a few driver mistake
Degree abrasion, causes these drivers to reach first and writes the life-span.Global resource abrasion equilibrium technology is thereby produced, is substantially distribution
By considering that the idle capacity of storage dish considers the life-span of driver indirectly during storage resource, to reach the longevity of All Drives
Life is all more balanced, is not in the situation that a small number of drivers are first written into end-of-life.There are various ways to characterize driver
Life-span, for example, for solid-state drive, the erasable number undergone with its flash memory (Program Erase Cycles) or solid
Mark of the data writing of state driver as the life-span, its flash memory can also be responded to the time of reading and writing and/or erasing operation
Measurement as the life-span.
The content of the invention
Inventors realized that if the abrasion equilibrium for being always ensured that All Drives is not always beneficial in RAID system
's.As it means that all driver life-spans are synchronous ageing, the end can be reached simultaneously by also implying that.To RAID systems
It is possible that the situation that polylith or All Drives collective break down in same time or very short time for system.Any
The low-quality disk quantity of RAID system tolerance all has certain limitations, and such case is all disaster for any RAID system
Property.
Therefore when abrasion equilibrium all also not up to closes to an end suitable for the life-span of All Drives, if All Drives
All in the young or middle age (early stage of life cycle), the abrasion condition for now balancing All Drives is advantageous to drive
The service life of device and the reliability of RAID system.But when all or when having more block drives all in senectitude, then
It is continuing with abrasion equilibrium strategy and is introduced into high fault rate.
The purpose of the present invention is the situation generation for avoiding global abrasion equilibrium from causing All Drives to damage together.Reach and work as
There is driver because the life-span is exhausted when being damaged, for impacted RAID groups, failed drive number is all in its RAID
Within the scope of protection level can be tolerated, so as to avoid loss of data.
According to the first aspect of the invention, there is provided a kind of side of establishment storage object according to a first aspect of the present invention
Method, including：Select the M bulk from N number of storage device；If provide at least first storage in the storage device of M bulk
The life-span of equipment and the second storage device exceedes first threshold, by first from the first storage device from the M bulk
Bulk is rejected, and M bulk for selecting the second bulk to replace first bulk and after update so that offer is second largest
The difference in the life-span of the 3rd storage device of block and the life-span of the second storage device is more than Second Threshold or provides the second bulk
Life-span of the 3rd storage device be no more than first threshold；And create storage object using M bulk after renewal.
According to the second aspect of the invention, there is provided first according to a second aspect of the present invention creates the side of storage object
Method, including：The M bulk from multiple storage devices is selected, wherein M bulk is used to build depositing with the first RAID
Object is stored up, first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；If
The life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, the p storage device
Including the first storage device and the p-1 storage device in addition to described first stores, reselect described in p-1 bulk replacement
Bulk from the p-1 equipment in M bulk so that in multiple storage devices that the p-1 bulk reselected is provided
Life-span exceedes in q storage device and first storage device of first threshold, the life-span difference of any two storage device
More than Second Threshold.
The method that first according to a second aspect of the present invention creates storage object, there is provided according to a second aspect of the present invention
Second method for creating storage object, wherein p, q, M, m is positive integer.
The method that first or second according to a second aspect of the present invention creates storage object, there is provided according to the present invention second
The method that the 3rd of aspect creates storage object, in addition to：Storage object is created using the M bulk after replacement.
According to the third aspect of the invention we, there is provided first according to a third aspect of the present invention creates the side of storage object
Method, including：The M bulk from multiple storage devices is selected, wherein M bulk is used to build depositing with the first RAID
Object is stored up, first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；If
The life-span for having at least m+1 storage device in the multiple storage device exceedes first threshold, checks and provides the M bulk
The life-span for whether having p storage device in storage device exceedes first threshold, wherein p>M, the p storage device include first
Storage device and the p-1 storage device in addition to described first stores, reselect p-1 bulk and replace the M bulk
In the bulk from the p-1 equipment so that providing the life-span in multiple storage devices of the p-1 bulk reselected exceedes
In q storage device of first threshold and first storage device, the life-span difference of any two storage device is more than second
Threshold value.
The method that first according to a third aspect of the present invention creates storage object, there is provided according to a third aspect of the present invention
Second create storage object method, if wherein the life-span for having at least m+1 storage device in the multiple storage device exceed
First threshold, methods described also include：Storage object is built using the M bulk after replacement.
The method that first or second according to a third aspect of the present invention creates storage object, there is provided according to the present invention the 3rd
Aspect the 3rd create storage object method, if wherein in the multiple storage device the life-span exceed first threshold storage set
It is standby to be no more than m, build storage object using the M bulk.
First according to a third aspect of the present invention is to the 3rd one of the method for creating storage object, there is provided according to the present invention
The method that the 4th of the third aspect creates storage object, wherein the M bulk of the selection from multiple storage devices, including：
M storage device is randomly choosed from the multiple storage device.
First according to a third aspect of the present invention is to the 3rd one of the method for creating storage object, there is provided according to the present invention
The method that the 5th of the third aspect creates storage object, wherein the M bulk of the selection from multiple storage devices, including：
Storage resource service condition according to storage device is each setting weight of the multiple storage device, according to weight from described
M storage device is selected in multiple storage devices, the wherein high storage device of weight is had higher by the low storage device of weight
Probability is chosen.
According to the fourth aspect of the invention, there is provided first according to a fourth aspect of the present invention creates the side of storage object
Method, including：The M bulk from multiple storage devices is selected, wherein M bulk is used to build depositing with the first RAID
Object is stored up, first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；If
The life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, the p storage device
Including the first storage device and the p-1 storage device in addition to described first stores, reselect the first bulk and replace the M
One of bulk from the p-1 equipment in individual bulk so that the life-span for providing the storage device of first bulk does not surpass
Cross first threshold or the difference in the life-span of the storage device of first bulk and the life-span of first storage device is provided
More than Second Threshold.
According to the fifth aspect of the invention, there is provided first according to a fifth aspect of the present invention creates the side of storage object
Method, including：The M bulk from multiple storage devices is selected, wherein M bulk is used to build depositing with the first RAID
Object is stored up, first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；If
The life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, reselect at least p-m
Individual bulk replaces at least p-m bulk in the p bulk from the p storage device in the M bulk, wherein providing
The life-span of multiple storage devices of at least p-m bulk reselected is no more than first threshold.
According to the sixth aspect of the invention, there is provided first according to a sixth aspect of the present invention creates the dress of storage object
Put, including：Bulk selecting module, for selecting the M bulk from N number of storage device；Bulk replacement module, if for providing
The life-span of at least the first storage device and the second storage device in the storage device of M bulk exceedes first threshold, from the M
The first bulk from the first storage device is rejected in individual bulk, and selects the second bulk to replace first bulk and obtain
M bulk after to renewal so that the life-span of the 3rd storage device of the second bulk of offer and the life-span of the second storage device
The life-span that difference is more than the 3rd storage device of the second bulk of Second Threshold or offer is no more than first threshold；And storage pair
As creation module, for creating storage object using M bulk after updating.
According to the seventh aspect of the invention, there is provided first according to a seventh aspect of the present invention creates the dress of storage object
Put, including：Bulk selecting module, for selecting the M bulk from multiple storage devices, wherein M bulk is used to build and had
There is the storage object of the first RAID, first RAID can tolerate that the m that bulk is provided to the storage object is individual
Storage device breaks down；Bulk replacement module, if the life-span for p storage device in the storage device of M bulk of offer
More than first threshold, wherein p>M, the p storage device include the first storage device and the p-1 in addition to described first stores
Individual storage device, reselect p-1 bulk and replace the bulk from the p-1 equipment in the M bulk so that provide
Life-span deposits more than q storage device of first threshold with described first in multiple storage devices of the p-1 bulk reselected
Store up in equipment, the life-span difference of any two storage device is more than Second Threshold.
According to the eighth aspect of the invention, there is provided first according to a eighth aspect of the present invention creates the dress of storage object
Put, including：Bulk selecting module, for selecting the M bulk from multiple storage devices, wherein M bulk is used to build and had
There is the storage object of the first RAID, first RAID can tolerate that the m that bulk is provided to the storage object is individual
Storage device breaks down；Bulk replacement module, if for there is the longevity of at least m+1 storage device in the multiple storage device
Life exceedes first threshold, checks in the storage device for providing the M bulk whether there is the life-span of p storage device more than first
Threshold value, wherein p>M, the p storage device include the first storage device and set with the p-1 storage in addition to described first stores
It is standby, reselect p-1 bulk and replace the bulk from the p-1 equipment in the M bulk so that offer reselects
P-1 bulk multiple storage devices in the life-span exceed first threshold q storage device and first storage device in,
The life-span difference of any two storage device is more than Second Threshold.
According to the ninth aspect of the invention, there is provided first according to a ninth aspect of the present invention creates the dress of storage object
Put, including：Bulk selecting module, for selecting the M bulk from multiple storage devices, wherein M bulk is used to build and had
There is the storage object of the first RAID, first RAID can tolerate that the m that bulk is provided to the storage object is individual
Storage device breaks down；Bulk replacement module, if the life-span for p storage device in the storage device of M bulk of offer
More than first threshold, wherein p>M, the p storage device include the first storage device and the p-1 in addition to described first stores
Individual storage device, reselect the first bulk and replace one of bulk from the p-1 equipment in the M bulk so that
The life-span for providing the storage device of first bulk is no more than first threshold or provides the storage device of first bulk
The difference in life-span of life-span and first storage device be more than Second Threshold.
According to the tenth aspect of the invention, there is provided a kind of computer program for including computer program code, when being loaded into
Computer system and when performing on the computer systems, the computer program code makes the computer system perform according to this
The data reconstruction method based on storage object that invention first aspect provides to the 5th aspect.
According to the eleventh aspect of the invention, there is provided a kind of program including program code, when being loaded into storage system simultaneously
When being performed in storage system, the meter program code makes the storage system perform according to a first aspect of the present invention to the 5th side
The data reconstruction method based on storage object that face provides.
Brief description of the drawings
When being read together with accompanying drawing, by reference to the detailed description of illustrative embodiment, will be best understood below
The present invention and preferable use pattern and its further objects and advantages, wherein accompanying drawing include：
Fig. 1 shows the framework of storage system according to embodiments of the present invention；
Fig. 2 shows the structure of storage object according to embodiments of the present invention；
Fig. 3 A- Fig. 3 D describe bulk selection course when storage object according to embodiments of the present invention creates；And
Fig. 4 shows that storage object according to embodiments of the present invention creates the flow chart of process.
Embodiment
Referring to Fig. 1, Fig. 1 shows the framework of storage system according to embodiments of the present invention.According to the storage system of the present invention
System includes computer or server (being referred to as main frame) and is coupled to multiple storage devices (for example, driver) of main frame.It is excellent
Selection of land, driver are solid-state drive (SSD).Alternatively, according to an embodiment of the invention in can also include disc driver.
The storage resource that each driver provides is safeguarded by memory resource pool.Each drive is have recorded in memory resource pool
Data block or data chunk (Chunk, abbreviation bulk) in dynamic device.As an example, data chunk is that the logic of predefined size is empty
Between or the continuous multiple data blocks of physical space.The size of data chunk, can be hundreds of KB or MB as an example.Alternatively,
What is recorded in storage resource is the data block or data chunk that storage object is not yet assigned in each driver, these data blocks
Or data chunk is known as freed data blocks or idle data bulk again.Memory resource pool is also a kind of virtualization technology, in the future
Data block or data chunk are virtually turned to for upper access or use from the storage resource of phisical drive.In storage system
In, memory resource pool can have multiple, and in the example in fig 1, show only single memory resource pool.
In an embodiment according to the present invention, the life-span of each driver is safeguarded, so as to would know that belonging to each bulk
The life-span of driver.In the life-span for there are various ways to characterize driver, for example, for solid-state drive, undergone with its flash memory
Mark of the data writing of erasable number (Program Erase Cycles) or solid-state drive as the life-span.It can also use
The life-span of the index characterization drivers such as the working time of driver, the wrong bitrate of driver., will according to the life-span of driver
Stage residing for driver is divided into nonage, middle age and/or senectitude., will by setting certain lifetime threshold
The driver that life-span exceedes threshold value is defined as old driver, and not less than the driver of threshold value are defined as into young/middle age the life-span
Driver.New driver is represented for example, the life-span of driver is quantified as into 0-100,0, and 100 represent the drive that the life-span exhausted
Dynamic device, is defined as threshold value, the driver that the life-span meets or exceeds 90 is old driver by 90.
The storage object provided according to embodiments of the present invention, represent the part memory space of storage system.Storage object
The memory cell with RAID functions, behind can combine Fig. 2 storage object structure is described in detail.According to an embodiment of the invention
Middle offer storage object layer, storage object layer provide multiple storage objects.Storage object can be created and destroy.One storage
When object is created, the bulk of requirement is obtained from memory resource pool by distributor, these bulks composition storage object.
One bulk can only belong to a storage object in synchronization.Already being allocated to the bulk of storage object will not be divided again
Other storage objects of dispensing.When storage object is destroyed, the bulk for forming the storage object is discharged back into memory resource pool, and can
It is assigned again to other storage objects.
Fig. 2 shows the structure of storage object according to embodiments of the present invention.Storage object includes multiple data blocks or number
According to bulk.In the example in figure 2, storage object includes bulk 220, bulk 222, bulk 224 and bulk 226.Form storage
The bulk of object is from different drivers.Each driver provides an at most bulk to a storage object.Referring to Fig. 2,
Bulk 220 carrys out output from driver 210, and bulk 222 carrys out output from driver 212, and bulk 224 carrys out output from driver 214 and bulk 226 comes from
Driver 216.So as to which when single driver breaks down, one in storage object or a small number of bulks can not access, and lead to
Other bulks of storage object are crossed, the data of storage object can be rebuild, to meet the requirement of data reliability.
Data protection is provided for storage object by RAID technique, and the high-performance for providing storage object accesses.Referring to figure
2, storage object includes multiple RAID stripes (band 230, band 232...... bands 238), and each band is by from different big
The memory space composition of block.The memory space from different bulks of same band can have identical or different address realm.
Band is the minimum write-in unit of storage object, so as to by being written in parallel to data to more block drives come improving performance.Deposit
The read operation of storage object does not have size limitation.RAID technique is realized in band.Still come from referring to Fig. 2, constituting bar band 230
In the memory space of 4 bulks, 3 memory spaces are used to store user data, and another 1 memory space is used to store check number
According to so that the data protection of RAID5 ranks is provided on band 230.
Alternatively, metadata information is also stored in each bulk.In the example in figure 2, bulk 220- bulks 226 is each
On all preserve identical metadata information, so as to ensure that the reliability of metadata information, even if belonging to same storage object
Part bulk is broken down, and metadata information can be still obtained from other bulks.Metadata is used to record depositing belonging to bulk
Store up object, the data protection level (RAID) of storage object where bulk, the information such as erasable number of bulk.
In an embodiment according to the present invention, when creating storage object, the drive of bulk is provided by being selected based on the life-span
Dynamic device, to realize that abrasion is inverse balanced.
The storage resource (bulk) of each storage object must meet certain condition, be considered as just safe.Each
When distributing resource (bulk) to create storage object, the storage resource of distribution must meet certain condition, could form storage
Object.As an example, the data storage for offer safety is, it is necessary to which the condition met includes：(1) the arbitrarily large of storage object is formed
Block both be from/young driver；(2) if condition (1) is unsatisfactory for, the bulk from old driver of storage object is formed
In, the life-span difference of any two bulk is more than a safe difference.Wherein, the life-span of bulk uses the driver of offer bulk
Life-span indicates.The implication of safe difference is, for any two pieces of old drivers, if their life-span difference is pacified at this
More than full difference, probability that they break down simultaneously is also considered as within the scope of acceptable, such two pieces old to drive
Dynamic device can provide bulk simultaneously to form storage object.
When not having driver to be in the senescence phase in storage system, global abrasion equilibrium strategy can be used.
, it is necessary to enter when considering creating storage object to the life-span of the bulk of distribution after thering is driver to enter the senescence phase
Row checks.In one example, after thering is driver to enter the senescence phase in storage system, when creating storage object every time, to dividing
The bulk life-span matched somebody with somebody is checked.In another example, the RAID of storage object is considered.If into the drive in senescence phase
Dynamic device negligible amounts, can be without checking when creating storage object when not influenceing the data protection ability of storage object.Example
The storage object of the ranks of RAID 5 is such as created, and only has a block drive in storage system and is in the senescence phase, or creates RAID
The storage object of 6 ranks, and only have two block drives in storage system and be in the senescence phase, now in the worst cases, driver
Life-span do not interfere with the data safety of storage object yet, can be checked when creating storage object without the life-span.
, then will be to every when the RAID protection levels for having more than storage object allow the driver of low-quality disk number to enter the senescence phase
The secondary bulk distributed during storage object that creates carries out life-span inspection.For example, creating the storage object of the ranks of RAID 5, and store system
The senescence phase is in more than a block drive in system, or creates the storage object of the ranks of RAID 6, and is had more than in storage system
Two block drives are in the senescence phase, then to carry out life-span inspection to the bulk of distribution when creating storage object.
When creating storage object, multiple bulks are selected first.Can be according to a variety of policy selection bulks.For example, according to reality
Existing global balanced bulk selection strategy, the strategy of bulk is randomly selected, and/or according to the driver idle capacity for providing bulk
Bulk selection strategy, selection provide bulk driver.Then life-span inspection is carried out to driver where these bulks or bulk
Look into.If current bulk has chosen " old age " of the RAID maximum tolerance failed drive number more than storage object
Driver, then carry out reselecting for bulk.
When bulk reselects, one piece of longevity arbitrarily selected first exceed threshold value driver (i.e. old driver) as
Destination drive, then by driver rejecting in other selected drivers and similar in the destination drive life-span, carry out weight
New selection.Life-span, the close life-span difference for referring to two block drives was less than safe difference.During the driver specifically selected is contemplated to be
Young driver or the driver with destination drive life-span difference outside safe difference range.If can not
Find the driver for meeting above-mentioned requirements, then current storage object creates failure, or can notify user's storage system
Data are in unsafe state.
The step of bulk reselects can be repeatedly executed, until all selected by the storage object that is created
Any life-span difference between any two of old driver is both greater than safe difference.Alternatively, if having retried enough times
Number does not find satisfactory result also, then the establishment failure of current storage object.
With reference to Fig. 3 A- Fig. 3 D, bulk when storage object according to embodiments of the present invention creates is described in detail and selected
Journey.Bulk is selected when storage object creates by the method provided according to embodiments of the present invention, to realize the drive of storage system
The abrasion of dynamic device is inverse balanced.
In Fig. 3 A- Fig. 3 D example, driver of the life-span not less than 90 is defined as old driver, and the driver longevity
The safe difference of life is 3, and storage object includes 4 bulks from 4 drivers, and data protection level is RAID 5.
Referring to Fig. 3 A, to create storage object, in the selection of first time bulk, it have selected and come output from driver 310, driver
320th, the bulk of driver 330 and driver 340.Wherein there are 3 pieces of old drivers, the life-span of driver 310 is 95, is driven
The life-span of dynamic device 330 is 96 and the life-span of driver 340 is 94.In figure 3 a, by the numeral instruction driver below driver
Life-span.Origin output from driver 310, driver 320, the bulk of driver 330 and driver 340 form storage object and are clearly
It is unsafe, because old driver 310, driver 330 may break down with driver 340 simultaneously, cause storage object
Data can not read.Safe storage object is obtained, it is necessary to so that in storage object while two bulk failures occur
Probability is sufficiently low.In this case it is desirable to carry out second of bulk selection for the storage object.To carry out second of bulk choosing
Select, from old driver 310, driver 330 with randomly selecting one piece of old driver as target drives in driver 340
Device, in Fig. 3 A example, driver 340 is selected as destination drive (in figure 3 a, indicated with arrows).Second of bulk
The target of selection is is less than 91 or old driver of the life-span more than 97 with young and middle-aged driver, or life-span, to replace except mesh
Mark other two pieces old drivers (old driver 310 and driver 330) outside driver.
Fig. 3 B show the result of second of bulk selection.In second of bulk selects, driver 312 is have selected to replace
Driver 310 is changed, and have selected driver 332 to replace driver 330.
Checked after second sub-distribution, the life-span of driver 312 and the life-span of driver 332 are 98.Due to driving
Dynamic device 312 belongs to age-related macular degeneration with driver 332, and the life-span difference (0) of the two is not more than safe difference, then needs to this
Two pieces of disks carry out third time bulk selection.When third time bulk selects, select one of driver 312 and driver 332 as
Destination drive.In figure 3b, as indicated by arrow, driver 332 is chosen to be used as destination drive.The mesh of third time bulk
Mark is the driver that safe difference is differed with the life-span of driver 332 (98).It is pointed out that selection bulk drives with selection
Device has identical effect, because for a storage object, each driver, which provides, is no more than 1 bulk.And the third-largest
In block selection, it is desirable to select destination drive should be (1) non-aged driver, or (2) and the longevity of driver 340
Life difference is more than safe difference, and with the old driver of the life-span difference of driver 332 more than safe difference.However, in basis
In one embodiment of the present of invention, to be easy to implement, the third-largest piece selection in, can only with current goal driver 332
Life-span difference more than safe difference be target.
With continued reference to Fig. 3 C, in the selection of third time bulk, driver 314 (life-span 91) is have selected to replace driver
312, because driver 314 and driver 332, the life-span difference of driver 340 are all higher than safe difference, thus current bulk choosing
Work(is selected to, suitable bulk is have found for storage object.
And Fig. 3 D then show the example that the selection of third time bulk still fails.In fig. 3d, third time is chosen
Driver 314 ' (life-span 93), because driver 314 ' belongs to old driver, and with the destination drive selected for the first time
340 life-span difference is unsatisfactory for condition, then the selection of third time bulk still fails.As an example, because 3 bulk selections are lost
Lose, it is believed that be difficult to find that suitable multiple bulks within the storage system to create storage object, the establishment of storage object is lost
Lose.
In the example shown in Fig. 3 A- Fig. 3 D, when selecting bulk every time, bulk is selected in a random way, and check institute
Whether the life-span for the bulk selected is eligible.One of ordinary skill in the art will realize the hand of various other selection bulks
Section.For example, the bulk of candidate is grouped by the life-span, so as to actively choose bulk or the old driving from young driver
The bulk of device.Further, age class can further be divided to old driver, for example, driver is divided into the age
The multiple grades such as 90-92,93-95,96-98,99-100 are belonging respectively to, if destination drive belongs to 90-92 age classes, from
96-98 is with the group of 99-100 age classes, can easily select the driver for meeting age difference condition.At another
In example, in the situation of figure 3 a, when have selected 3 bulks from old driver, for depositing for the protection levels of RAID 5
Object is stored up, retains one of bulk, and other 2 bulks are replaced with into the bulk from young driver.
In still another embodiment, the data protection grade of storage object is RAID 6, it is allowed to forms storage object
Bulk in, have 2 bulks while break down.So when creating storage object, if having in the M bulk selected for the first time
More than 2 bulks (for example, p) come from old driver, select bulk of the two of which from old driver as target
Block or destination drive.And p-2 bulk is selected again, to replace the p-2 bulks from old driver.As act
Example, when the p-2 bulk reselected is all from non-aged driver, disclosure satisfy that the requirement of the data safety of storage object.
As another example, reselecting the target of bulk is, in the old bulk for forming storage object, is arbitrarily less than safe differential value
Age difference section in big number of blocks be not more than 2.
Fig. 4 shows that storage object according to embodiments of the present invention creates the flow chart of process.To create storage object, from
M bulk (410) is selected in the memory resource pool of storage system.An each driver of the bulk from storage system, and keep away
Exempt to select two or more bulks from same driver, so as to when driver breaks down, pass through its of storage object
His bulk can recover the data of the bulk of failed drive.Storage object has RAID, for example, RAID5, RAID6
Or other RAIDs.M bulk is randomly selected from memory resource pool, and meets each driver to multidirectional one
Storage object provides the requirement of a bulk.Can be also that driver setting is selected according to the capacity of driver when selecting bulk
In probability.Higher selected probability is set in driver with more free spaces.
Check whether the big number of blocks p from old driver exceedes threshold value m (420) in the M bulk selected.M, m with
And p is positive integer.Threshold value is related to the RAID of storage object.If the RAID of storage object is RAID5, Zhi Nengrong
Bear in M bulk for forming storage object while a failure bulk occur, then threshold value m is 1；And for RAID6, threshold value m
For 2.Alternatively, it is not that each storage object that creates is required for implementation steps 420.For example, the RAID of storage object is
RAID 5, and old driver is not present in multiple drivers of storage system or only exists an old driver, then choosing
In the M bulk then gone out, it is impossible to two bulks from old driver be present, in the case, without implementation steps
420.Thus alternatively, when the old number of drives in recognizing storage system exceedes threshold value, just M bulk is being selected
Afterwards, the inspection of step 420 is performed.
If in the M bulk selected, the big number of blocks p from old driver is no more than threshold value m, then to select
M bulk creates storage object (450).If in the M bulk selected, the big number of blocks p from old driver exceedes threshold
Value m, reselect p-m bulk (430).There are a variety of strategies for reselecting p-m bulk.For example, make the p-m reselected
Individual bulk is all from non-aged driver, and causes M bulk after reselecting to be all from different drivers, each driving
The big number of blocks that device provides is no more than 1.In another example, m=1,1 bulk from old driver is randomly choosed
(referring also to Fig. 3 A, carrying out the bulk of output from driver 340) is retained；And by its in the bulk from p old drivers
It is big that his p-m bulks (referring also to Fig. 3 A, carrying out the bulk of output from driver 310 and driver 330) replace with p-m reselected
Block so that p-m bulk either from non-aged disk or after replacing of p-m bulk after replacement and retained big
The life-span difference of any two bulk is more than safe difference between block.Alternatively, when reselecting p-m bulk, often
It is secondary to reselect 1 bulk.When reselecting bulk every time, a bulk is randomly choosed from memory resource pool, to replace
One not be replaced in p-m bulk, and cause a selected bulk either from non-aged disk or with p greatly
Any one age difference of retained bulk is more than safe difference in block, and will be selected qualified
Bulk is as retained bulk.If a selected bulk is unsatisfactory for above-mentioned requirements, reselected from memory resource pool
Bulk.Further, number of retries threshold value is set, if can not select qualified bulk still after retrying repeatedly, it is believed that storage
So that satisfactory bulk can not be provided in system, can be created by the use of storage object unsuccessfully as response, or prompt the user with nothing
Method obtains satisfactory bulk.
In another example, m>1, retain the m bulks from old driver, and make the drive of this m bulk of offer
The life-span difference of dynamic device is more than safe difference；And other p-m bulk in the p bulks from old driver are replaced with
The p-m bulks reselected.Further, when reselecting p-m bulk, 1 bulk is reselected every time.Weighing every time
During new selection bulk, a bulk is randomly choosed from memory resource pool, to replace be not replaced in p-m bulk one,
And cause a selected bulk either from non-aged disk or any with bulk retained in p bulk
The age difference of one is more than safe difference, and using selected qualified bulk as retained bulk.
The embodiment of the present invention also provides a kind of program for including program code, when being loaded into CPU and being performed in CPU,
Program makes CPU perform one of method according to embodiments of the present invention provided above.
The embodiment of the present invention also provides a kind of program including program code, when being loaded into main frame and performed on main frame
When, described program makes one of the computing device of main frame method according to embodiments of the present invention provided above.
It should be understood that the combination of the frame of each frame and block diagram and flow chart of block diagram and flow chart can respectively by including
The various devices of computer program instructions are implemented.These computer program instructions can be loaded into all-purpose computer, special meter
To produce machine on calculation machine or other programmable data control devices, so as in computer or other programmable data control devices
The instruction of upper execution creates the device for realizing the function of being specified in one or more flow chart box.
These computer program instructions, which can also be stored in, can guide computer or other programmable data control devices
Computer-readable memory in so as to working in a specific way, so as to using being stored in computer-readable memory
Instruction manufacture including the product for realizing the computer-readable instruction of specified function in one or more flow chart box.
Computer program instructions can also be loaded on computer or other programmable data control devices with cause computer or its
A series of operation operation is performed on his programmable data control device, so as to produce computer implemented process, and then is being counted
The instruction performed on calculation machine or other programmable data control devices is provided for realizing institute in one or more flow chart box
Specify the operation of function.
Thus, the frame of block diagram and flow chart supports the combination of the device for performing specified function, for performing specified work(
The combination of operation and the combination of the program instruction means for performing specified function of energy.It should also be understood that block diagram and flow chart
The combination of frame of each frame and block diagram and flow chart can specify function or operation, hardware based special meter by performing
Calculation machine system is realized, or is realized by the combination of specialized hardware and computer instruction.
Although the example of present invention reference is described, it is intended merely to the purpose explained rather than the limit to the present invention
System, the change to embodiment, increase and/or deletion can be made without departing from the scope of the present invention.
It is involved by these embodiments, from the description above and in the field that benefits of the teaching that is presented in associated accompanying drawing
Technical staff will be recognized that many modifications of the invention recorded here and other embodiment.It should therefore be understood that this hair
It is bright to be not limited to disclosed embodiment, it is intended to include modification and other embodiment in the scope of the appended claims
It is interior.Although employing specific term herein, them are used only on general significance and describing significance and not is
The purpose of limitation and use.
Claims (10)
1. a kind of method for creating storage object, including：
Select the M bulk from N number of storage device；
If the life-span of at least the first storage device and the second storage device in the storage device of M bulk of offer is more than the first threshold
Value, the first bulk from the first storage device is rejected from the M bulk, and the second bulk of selection replaces described the
M bulk a bulk of and after being updated so that providing the life-span of the 3rd storage device of the second bulk and second stores and set
The life-span that the difference in standby life-span is more than the 3rd storage device of the second bulk of Second Threshold or offer is no more than first threshold；
And
Storage object is created using M bulk after renewal.
2. a kind of method for creating storage object, including：
The M bulk from multiple storage devices is selected, wherein M bulk is used to build the storage pair with the first RAID
As first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；
If the life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, the p are deposited
Storing up equipment includes the first storage device and the p-1 storage device in addition to described first stores, and reselects p-1 bulk and replaces
Change the bulk from the p-1 equipment in the M bulk so that provide multiple storages of the p-1 bulk reselected
The life-span exceedes in q storage device and first storage device of first threshold in equipment, the longevity of any two storage device
Life difference is more than Second Threshold.
3. a kind of method for creating storage object, including：
The M bulk from multiple storage devices is selected, wherein M bulk is used to build the storage pair with the first RAID
As first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；
If the life-span for having at least m+1 storage device in the multiple storage device exceedes first threshold, check that offer is described M
The life-span for whether having p storage device in the storage device of bulk exceedes first threshold, wherein p>M, the p storage device bag
The first storage device and the p-1 storage device in addition to described first stores are included, p-1 bulk is reselected and replaces the M
Bulk from the p-1 equipment in individual bulk so that the longevity in multiple storage devices of the p-1 bulk reselected is provided
For life more than in q storage device of first threshold and first storage device, the life-span difference of any two storage device is big
In Second Threshold.
4. the method according to claim 11, wherein
If the life-span for having at least m+1 storage device in the multiple storage device exceedes first threshold, methods described also includes：
Storage object is built using the M bulk after replacement.
5. according to the method described in one of claim 3-4, wherein
If the storage device that the life-span exceedes first threshold in the multiple storage device utilizes the M bulk structure no more than m
Build storage object.
6. according to the method described in one of claim 3-5, wherein
The M bulk of the selection from multiple storage devices, including：M are randomly choosed from the multiple storage device to deposit
Store up equipment.
7. according to the method described in one of claim 3-5, wherein
The M bulk of the selection from multiple storage devices, including：Storage resource service condition according to storage device is institute
Each setting weight of multiple storage devices is stated, M storage device is selected from the multiple storage device according to weight, wherein
The high storage device of weight has higher probability to be chosen by the low storage device of weight.
8. a kind of method for creating storage object, including：
The M bulk from multiple storage devices is selected, wherein M bulk is used to build the storage pair with the first RAID
As first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；
If the life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, the p are deposited
Storing up equipment includes the first storage device and the p-1 storage device in addition to described first stores, and reselects the first bulk and replaces
Change one of bulk from the p-1 equipment in the M bulk so that provide the longevity of the storage device of first bulk
Life-span and the life-span of first storage device of the life no more than first threshold or the storage device for providing first bulk
Difference be more than Second Threshold.
9. a kind of method for creating storage object, including：
The M bulk from multiple storage devices is selected, wherein M bulk is used to build the storage pair with the first RAID
As first RAID can tolerate that the m storage device that bulk is provided to the storage object breaks down；
If the life-span for providing p storage device in the storage device of M bulk exceedes first threshold, wherein p>M, reselect to
Few p-m bulk replaces at least p-m bulk in the M bulk in the p bulk from the p storage device, its
The life-span of the middle multiple storage devices for providing at least p-m bulk reselected is no more than first threshold.
10. a kind of device for creating storage object, including：
Bulk selecting module, for selecting the M bulk from N number of storage device；
Bulk replacement module, if being set at least the first storage device in the storage device of M bulk of offer and the second storage
The standby life-span exceedes first threshold, rejects the first bulk from the first storage device from the M bulk, and selects the
Two bulks are come M bulk replacing first bulk and after being updated so that provide the 3rd storage device of the second bulk
The difference in life-span of life-span and the second storage device be more than Second Threshold or the 3rd storage device of the second bulk be provided
Life-span is no more than first threshold；And
Storage object creation module, for creating storage object using M bulk after updating.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610498832.XA CN107544747B (en) | 2016-06-29 | 2016-06-29 | Wear reverse balancing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610498832.XA CN107544747B (en) | 2016-06-29 | 2016-06-29 | Wear reverse balancing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107544747A true CN107544747A (en) | 2018-01-05 |
| CN107544747B CN107544747B (en) | 2022-06-03 |
Family
ID=60965980
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610498832.XA Active CN107544747B (en) | 2016-06-29 | 2016-06-29 | Wear reverse balancing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107544747B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111104055A (en) * | 2018-10-29 | 2020-05-05 | 伊姆西Ip控股有限责任公司 | Method, apparatus and computer program product for managing a storage system |
| CN113535082A (en) * | 2021-06-09 | 2021-10-22 | 杭州电子科技大学 | A method to achieve inverse balance of wear |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7865761B1 (en) * | 2007-06-28 | 2011-01-04 | Emc Corporation | Accessing multiple non-volatile semiconductor memory modules in an uneven manner |
| CN102209952A (en) * | 2009-02-20 | 2011-10-05 | 株式会社日立制作所 | Storage system and method for operating the storage system |
| US20130046945A1 (en) * | 2011-08-15 | 2013-02-21 | Fujitsu Limited | Storage apparatus and storage apparatus control method |
| US8453036B1 (en) * | 2010-02-01 | 2013-05-28 | Network Appliance, Inc. | System and method for dynamically resizing a parity declustered group |
| CN103688248A (en) * | 2013-06-29 | 2014-03-26 | 华为技术有限公司 | Management method, device and controller of a storage array |
| US8825938B1 (en) * | 2008-03-28 | 2014-09-02 | Netapp, Inc. | Use of write allocation decisions to achieve desired levels of wear across a set of redundant solid-state memory devices |
| CN104571947A (en) * | 2014-12-05 | 2015-04-29 | 华为技术有限公司 | Method for partitioning hard disk domains in storage array, as well as controller and storage array |
-
2016
- 2016-06-29 CN CN201610498832.XA patent/CN107544747B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7865761B1 (en) * | 2007-06-28 | 2011-01-04 | Emc Corporation | Accessing multiple non-volatile semiconductor memory modules in an uneven manner |
| US8825938B1 (en) * | 2008-03-28 | 2014-09-02 | Netapp, Inc. | Use of write allocation decisions to achieve desired levels of wear across a set of redundant solid-state memory devices |
| CN102209952A (en) * | 2009-02-20 | 2011-10-05 | 株式会社日立制作所 | Storage system and method for operating the storage system |
| US8453036B1 (en) * | 2010-02-01 | 2013-05-28 | Network Appliance, Inc. | System and method for dynamically resizing a parity declustered group |
| US20130046945A1 (en) * | 2011-08-15 | 2013-02-21 | Fujitsu Limited | Storage apparatus and storage apparatus control method |
| CN103688248A (en) * | 2013-06-29 | 2014-03-26 | 华为技术有限公司 | Management method, device and controller of a storage array |
| CN104571947A (en) * | 2014-12-05 | 2015-04-29 | 华为技术有限公司 | Method for partitioning hard disk domains in storage array, as well as controller and storage array |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111104055A (en) * | 2018-10-29 | 2020-05-05 | 伊姆西Ip控股有限责任公司 | Method, apparatus and computer program product for managing a storage system |
| CN111104055B (en) * | 2018-10-29 | 2023-10-27 | 伊姆西Ip控股有限责任公司 | Method, apparatus and computer program product for managing a storage system |
| CN113535082A (en) * | 2021-06-09 | 2021-10-22 | 杭州电子科技大学 | A method to achieve inverse balance of wear |
| CN113535082B (en) * | 2021-06-09 | 2024-05-31 | 杭州电子科技大学 | Method for realizing wear inverse equalization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107544747B (en) | 2022-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101923501B (en) | Disk array multi-level fault tolerance method | |
| CN104956309B (en) | Storage system and data management method | |
| US7886111B2 (en) | System and method for raid management, reallocation, and restriping | |
| US8843782B2 (en) | Method and apparatus for reconstructing redundant array of inexpensive disks, and system | |
| US8566673B2 (en) | Method for improving performance in RAID systems | |
| US9575844B2 (en) | Mass storage device and method of operating the same to back up data stored in volatile memory | |
| CN105892934B (en) | Method and apparatus for storage device management | |
| CN110096217A (en) | The data from end-of-life memory driver are relocated in the data-storage system using mapping RAID technique | |
| WO2009129054A2 (en) | Dynamically quantifying and improving the reliability of distributed data storage systems | |
| CN103718162A (en) | Method and apparatus for flexible raid in ssd | |
| US11860746B2 (en) | Resilient data storage system with efficient space management | |
| CN109582219A (en) | Storage system, computing system and its method | |
| CN110096218A (en) | In response to reducing the distribution variation of driver panel to using the data-storage system of mapping RAID technique to add memory driver | |
| CN103870352A (en) | Method and system for data storage and reconstruction | |
| CN118779146B (en) | Data storage method, device, medium and product | |
| CN107665096A (en) | Weighted data slitting | |
| CN107239233A (en) | A kind of disk array and data reconstruction method and storage device based on it | |
| US11327666B2 (en) | RAID member distribution for granular disk array growth | |
| CN107544747A (en) | Abrasion inverse equalization methods and device | |
| US9547554B2 (en) | Mass storage device and method of operating the same to store parity data | |
| CN104572374B (en) | Processing method, device and the storage device of storage | |
| CN107665090A (en) | The method for realizing global abrasion equilibrium | |
| US11531500B1 (en) | Coalescing storage space of heterogeneous nodes for flexible expansion | |
| CN111694509A (en) | Apparatus and method for automatic configuration of storage space | |
| CN106997277A (en) | Efficiently utilize the method and device of the memory space reclaimed |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 100192 room A302, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Applicant after: Beijing yihengchuangyuan Technology Co.,Ltd. Address before: 100192 room A302, building B-2, Dongsheng Science Park, Zhongguancun, 66 xixiaokou Road, Haidian District, Beijing Applicant before: BEIJING MEMBLAZE TECHNOLOGY Co.,Ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |