CN101609420A - Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array - Google Patents
Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array Download PDFInfo
- Publication number
- CN101609420A CN101609420A CNA2009100894183A CN200910089418A CN101609420A CN 101609420 A CN101609420 A CN 101609420A CN A2009100894183 A CNA2009100894183 A CN A2009100894183A CN 200910089418 A CN200910089418 A CN 200910089418A CN 101609420 A CN101609420 A CN 101609420A
- Authority
- CN
- China
- Prior art keywords
- disk
- raid
- zone
- inefficacy
- address
- 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.)
- Pending
Links
Images
Landscapes
- Debugging And Monitoring (AREA)
Abstract
The invention discloses a kind of method of redundant arrays of inexpensive disks (RAID) reconstruction, RAID and a kind of RAID controller that realization is rebuild realized.The present invention is when having member's disk failure to cause RAID to be in degrading state, even if member's disk failure is arranged again, as long as the zone, fail address of all inefficacy member disks is not overlapping, then can continue to use the not data in the zone, fail address of member's disk that respectively lost efficacy, all data in each inefficacy member disk are returned to respectively in the corresponding HotSpare disk.Like this, just make the fault tolerance degree in the RAID process of reconstruction be not limited to the RAID redundance.And, in each member's disk that lost efficacy not the data in the zone, fail address still can adopt the mode of direct copy, if in the inefficacy member disk not fail address zone the I/O write operation takes place, then can directly the I/O write data be copied to corresponding HotSpare disk, thereby reduced RAID algorithm computation amount, accelerated the RAID reconstruction speed.
Description
Technical field
The present invention relates to the reconstruction technique of redundant arrays of inexpensive disks (RAID), the method that the particularly a kind of RAID of realization rebuilds, a kind of RAID and a kind of RAID controller of rebuilding realized.
Background technology
RAID can provide the jumbo while by a plurality of disk aggregations together, and higher access speed, better availability and data redundancy protection are provided.Wherein, the disk among the RAID can be described as member's disk of this RAID, has not only stored the outside data that are written to this disk in each member's disk, also stores the redundant data in other member's disk.
RAID can change between following three kinds of states: normal condition, and promptly all the member's disks among the RAID all are in normal condition, and then the RAID array is in normal condition; Degrading state, i.e. part member disk failure, then RAID becomes degrading state; Failure state, i.e. loss of data among the RAID, corresponding RAID5, common judgement is to be in member's disk failure is arranged among the RAID of degrading state again, then this RAID promptly becomes failure state; Corresponding RAID6, common judgement is always to have three member's disk failure, then this RAID promptly becomes failure state.
Wherein, the RAID that is in degrading state no longer has redundancy, but can carry out the RAID algorithm computation by all data (comprising redundant data) that read in normal member's disk, obtain losing efficacy data in member's disk, thereby the RAID of this moment still can externally provide normal disk access function; The RAID that is in failure state can't externally provide normal disk access function, and the data that promptly leave among this RAID will thoroughly be lost.
Therefore, when RAID is in degrading state, need to make this RAID recover redundancy as early as possible by the reconstruction of RAID.
As shown in Figure 1, be example to be provided with 1 HotSpare disk among 4 member's disks formation RAID 5 and the RAID 5, realize the reconstruction of RAID in the prior art usually in the following way:
In the several strips, each stick of each band is distributed in member's disk 0～3.When 1 inefficacy (shown in " * " among Fig. 1) of member's disk makes RAID 5 become degrading state, immediately this member's disk 1 is kicked out of from RAID 5, and utilize the data in each stick in remaining member's disk 0, member's disk 2, the member's disk 3 to carry out XOR (Xor) computing, obtain belonging to band 0 stick D1, belong to band 1 stick D4, belong to each stick in the former member's of being distributed in disks 2 such as stick P2 of band 2, deposit in to HotSpare disk then.
Yet above-mentioned reconstruction model need carry out the Xor computing at all sticks in the inefficacy member disk, can cause reconstruction time longer.
In order to address the above problem, referring to Fig. 2, be example still to be provided with 1 HotSpare disk among 4 member's disks formation RAID 5 and the RAID5, can also realize the reconstruction of RAID in the prior art in the following way:
In the several strips, each stick of each band is distributed in member's disk 0～3.When 2 inefficacies (shown in " * " among Fig. 2) of member's disk make RAID 5 become degrading state, wouldn't will should from RAID 5, kick out of by inefficacy member disk 1, but earlier all unspoiled each stick in sector, place in the inefficacy member disk 1 directly is copied to HotSpare disk, and then should from RAID 5, kick out of by inefficacy member disk 1, and utilize the data in each stick in remaining member's disk 0, member's disk 2, the member's disk 3 to carry out the Xor computing, obtain the not copy in the former member's of being distributed in disk 1 each stick, deposit in to HotSpare disk.
Though above-mentioned another kind of reconstruction mode as shown in Figure 2 can carry out as few as possible Xor calculate, to shorten the RAID reconstruction time, but the RAID reconstruction time remains long, and in so long RAID reconstruction time, if there is member's disk failure will cause RAID to become failure state again, then the data in the whole RAID also just no longer complete, can't realize rebuilding.
As seen, the existing RAID reconstruction model in conjunction with copy and Xor computing as shown in Figure 2 only can shorten the RAID reconstruction time, but because the fault tolerance degree in the RAID process of reconstruction is subject to the RAID redundance, thereby be prone to the problem of data failure among the RAID.
Summary of the invention
In view of this, the invention provides method that a kind of RAID of realization rebuilds, a kind of RAID and a kind of RAID controller of rebuilding realized, can improve the fault tolerance degree of RAID in process of reconstruction.
A kind of method that realizes that RAID rebuilds provided by the invention, being in has other member's disk failure again among the RAID of degrading state and the zone, fail address between member's disk of respectively losing efficacy when not overlapping, and this method for reconstructing comprises the steps:
For each inefficacy member disk data in the zone, fail address not, directly be copied in the HotSpare disk of this member's disk correspondence;
For the data in each inefficacy member disk zone, fail address, read the corresponding data in other normal member's disks among the RAID and other inefficacies member disk, and be written to after calculating in the HotSpare disk of this member's disk correspondence.
This method for reconstructing further will copy in each inefficacy member disk or be disarmed state by calculating the not fail address zone marker that upgrades;
And, during the corresponding data of this method for reconstructing in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
In copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, this method for reconstructing directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
This method for reconstructing further is respectively the mapping MAP table that each member's disk is provided with a correspondence; In the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and I/O read-write error flag can take place in this address area for losing efficacy, and the data in this address area have copied or by calculate be marked as after upgrading invalid.
Each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by the calculating renewal.
An address area is one or more sticks or is one or more sector.
This method for reconstructing is further kicked out of inefficacy member disk the MAP table of RAID, deletion inefficacy member disk correspondence after reconstruction is finished, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
A kind of RAID of rebuilding of realizing provided by the invention, comprise: RAID controller and a plurality of member's disk that condenses together, when this RAID is in failure state, other member's disk failure are arranged again and the zone, fail address of respectively losing efficacy between member's disk not overlapping;
For each inefficacy member disk data in the zone, fail address not, the RAID controller directly is copied in the HotSpare disk of this member's disk correspondence;
For the data in each inefficacy member disk zone, fail address, the RAID controller reads the corresponding data in other normal member's disks among the RAID and other inefficacies member disk, and is written to after calculating in the HotSpare disk of this member's disk correspondence.
The RAID controller further will copy in each inefficacy member disk or be disarmed state by calculating the not fail address zone marker that upgrades;
And, during the corresponding data of RAID controller in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
In copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, the RAID controller directly directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
Further be respectively arranged with the mapping MAP table of a correspondence in the RAID controller for each member's disk; In the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and I/O read-write error flag can take place in this address area for losing efficacy, and the data in this address area have copied or by calculate be marked as after upgrading invalid.
Each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by the calculating renewal.
An address area is one or more sticks or is one or more sector.
The RAID controller is further kicked out of inefficacy member disk the MAP table of RAID, deletion inefficacy member disk correspondence after reconstruction is finished, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
A kind of RAID controller provided by the invention is connected in each member's disk and HotSpare disk among the RAID, and this RAID controller comprises:
Main control unit has other member's disk failure again and the zone, fail address between member's disk of respectively losing efficacy when not overlapping in being in the RAID of degrading state, keep enabling copy cell and computing unit;
Copy cell with inefficacy member disk data in the zone, fail address not, directly is copied in the HotSpare disk of this member's disk correspondence;
Computing unit, at the data in inefficacy member disk zone, fail address, read the data in the corresponding address zone in other normal member's disks among the RAID and other inefficacies member disk, and be written to after calculating in the HotSpare disk of this member's disk correspondence.
Copy cell is a disarmed state with copied not fail address zone marker in each inefficacy member disk further; Computing unit will be a disarmed state by calculating the not fail address zone marker that upgrades in each inefficacy member disk further;
And, during the corresponding data of computing unit in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
In copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, copy cell directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
Further comprise map unit in the RAID controller, be stored as the corresponding mapping MAP table that each member's disk is provided with respectively, in the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and can take place to be labeled as inefficacy after the I/O read-write mistake, and have finished copy or invalid by being marked as after calculating renewal in this address area in this address area.
Each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by calculating renewal back.
An address area is one or more sticks or is one or more sector.
Main control unit is further kicked out of inefficacy member disk the MAP table of RAID, deletion inefficacy member disk correspondence after reconstruction is finished, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
As seen from the above technical solution, the present invention is when having member's disk failure to cause RAID to be in degrading state, even if member's disk failure is arranged again, as long as the zone, fail address of all inefficacy member disks is not overlapping, then can continue to use the not data in the zone, fail address of member's disk that respectively lost efficacy, all data in each inefficacy member disk are returned to respectively in the corresponding HotSpare disk.Like this, just make the fault tolerance degree in the RAID process of reconstruction be not limited to the RAID redundance.
And, in each member's disk that lost efficacy not the data in the zone, fail address still can adopt the mode of direct copy, if in the inefficacy member disk not fail address zone the I/O write operation takes place, then can directly the I/O write data be copied to corresponding HotSpare disk, thereby reduced RAID algorithm computation amount, accelerated the RAID reconstruction speed.
Further, I/O read-write error flag can will take place for losing efficacy, in order to identification zone, fail address by for example mode of mapping (MAP) table in the present invention in each inefficacy member disk; Also will copy in each inefficacy member disk or be disarmed state by calculating the scheduler zone marker, data on the corresponding HotSpare disk after promptly upgrading are exactly effective, like this, can select according to this from inefficacy member disk reading of data or from HotSpare disk reading of data.
Alternatively, the address area in member's disk can be divided by stick, can make calculating simple relatively like this; Perhaps, address area in member's disk can be divided with littler granularity even sector, the overlapping nonoverlapping probability in zone, fail address is bigger in the member's disk that respectively lost efficacy though calculate relative complex this moment, can further improve the fault tolerance degree of RAID array.
Description of drawings
Fig. 1 is the synoptic diagram of existing a kind of RAID process of reconstruction;
Fig. 2 is the synoptic diagram of existing another kind of RAID process of reconstruction;
Fig. 3 realizes the synoptic diagram that RAID rebuilds in the embodiment of the invention;
Fig. 4 is the structural representation of RAID controller in the embodiment of the invention.
Embodiment
For making purpose of the present invention, technical scheme and advantage clearer, below with reference to the accompanying drawing embodiment that develops simultaneously, the present invention is described in more detail.
In the present embodiment, be example with RAID5, when having member's disk failure to cause RAID to be in degrading state in the ban, for example RAID 5 has 1 member's disk failure to cause degrading state, can realize in the following manner rebuilding:
For inefficacy member disk data in the zone, fail address not, directly be copied in the identical address zone in the HotSpare disk of this member's disk correspondence;
For the data in inefficacy member disk zone, fail address (the wrong address area of I/O read-write promptly takes place), read among the RAID in other normal member's disks the data in the corresponding address zone and carry out after Xor calculates, be written in the identical address zone in the HotSpare disk of this member's disk correspondence.
Need to prove that above-mentioned " corresponding address zone " is meant other address areas in the group of same address area, and the data in the arbitrary address zone in the group of same address area, can carry out Xor by the data in other address areas in this group and calculate.If each address area comprises one or more stick, then an address area group is the band of one or more equal number in fact just; If each address area comprises one or more sector, then address area group in fact just can regard with the sector to be the sub-band of one or more equal number of dividing of unit as.
After this, other member's disk failure are arranged again, for example the RAID 5 of degrading state has at least 1 member's disk failure again, in this case, present embodiment is not converted to failure state with RAID, do not stop to rebuild yet, but at first judge the zone, fail address respectively lost efficacy between member's disk whether not overlapping (" not overlapping " as herein described is meant that the zone, fail address of respectively losing efficacy between member's disk belongs to different address areas group), if the zone, fail address of respectively losing efficacy between member's disk is not overlapping, for example not overlapping between the zone, fail address of the individual inefficacy member disk of all n (n is smaller or equal to member's disk sum) among the RAID5, though then the inefficacy member number of disks in the whole RAID has reached 2, but for calculating each address area group of recovering data in the arbitrary address zone by Xor, when the zone, fail address between each inefficacy member disk is not overlapping, realize in the present embodiment that the method for reconstructing of RAID can be proceeded to rebuild in the following manner:
For each inefficacy member disk data in the zone, fail address not, directly be copied in the identical address zone in the HotSpare disk of this member's disk correspondence; Alternatively,, then directly the data of I/O write operation are write in the identical address zone in the corresponding HotSpare disk, represent that this has not finished copy in the zone, fail address this moment if in the zone, not fail address of current copy the I/O write operation is arranged; Need to prove, the part that the I/O write operation takes place may be less than an address area, also may just equal an address area, so for the situation of the part that the I/O write operation takes place less than an address area, except being copied to the I/O write data in the corresponding HotSpare disk, other parts that also need to the I/O write operation not to take place in this address area also are copied to HotSpare disk, to guarantee the data integrity in this address area;
For the data in each inefficacy member disk zone, fail address, read in other normal member's disks among the RAID and other inefficacies member disk and to belong to data in the corresponding address zone of same address area group with this address area and carry out after Xor calculates, be written in the identical address zone in the HotSpare disk of this member's disk correspondence.
Reconstruction mode when above-mentioned RAID is in failure state can be regarded that being distributed in each address areas of all member's disks each address area component virtual is a sub-RAID as, and realize the reconstruction of each virtual subnet RAID respectively.
In the practical application, because the data computation of each inefficacy member disk also can need the data in other inefficacies member disk, and the corresponding data in other inefficacies member disk may be resumed to HotSpare disk, therefore, realize in the process of reconstruction of method for reconstructing when RAID is in failure state of RAID in the present embodiment, except will cause owing to I/O read-write is wrong truly the fail address zone marker be the failure state, also can further will finish copy (comprise and copy two kinds of situations of I/O write data) from the address area copy, or be disarmed state by calculating the not fail address zone marker that upgrades.
Like this, during data in reading zone, fail address not,, then further read the data in the corresponding address zone in the corresponding HotSpare disk if this has not been marked as disarmed state in the zone, fail address.
In order to realize that a kind of optional mode in the present embodiment is to zone, true fail address and the mark of finishing the zone, not fail address of copy: be respectively the mapping MAP table that each member's disk is provided with a correspondence; In the MAP table of each member's disk correspondence, each element is respectively to an address area in should member's disk, and can be labeled as failure state after I/O read-write mistake takes place in this address area and the data in this address area copy (comprising from address area copy and two kinds of situations of copy I/O write data) or have been marked as disarmed state by calculating after upgrading.
Specifically, each element can only be made of dibit, wherein a bit represent the corresponding address zone whether be marked as owing to the I/O read-write is wrong inefficacys, promptly represent zone, real fail address, another bit represent the corresponding address zone whether by since finished copy (comprising) from address area copy and two kinds of situations of copy I/O write data or since by the calculating renewal be marked as invalid, be the zone, not fail address that data have been finished renewal.
In this case, can be further after inefficacy member disk be kicked out of RAID after reconstruction is finished, delete the MAP table of this inefficacy member disk correspondence, and further the HotSpare disk of inefficacy member disk correspondence is provided with the new MAP table of correspondence as newcomer's disk, interpolation.
As previously mentioned, the address area in the present embodiment in mentioned member's disk can be divided with stick, and correspondingly, each zone, fail address is one or more stick that includes the inefficacy sector; Perhaps, the address area in member's disk also can be divided with the sector, and correspondingly, each zone, fail address is one or more inefficacy sector.
Above-mentioned two kinds of different address area dividing mode are compared, and divide the complexity lower slightly (those skilled in the art can directly select existing any mode for use) of the Xor account form of address area with stick; Divide the complexity slightly high (those skilled in the art need and can improve slightly existing any mode) of the Xor account form of address area with the sector, but the nonoverlapping probability in zone, fail address overlapping in the member's disk that respectively lost efficacy is bigger, can further improve the fault tolerance degree of RAID array.
Therefore, in the practical application, can need to select arbitrarily a kind of address area dividing mode according to situation, but no matter select any address area dividing mode, the principle of reconstruction is all identical.
Referring to Fig. 3, be example still to be provided with 2 HotSpare disk and to comprise a stick among 4 member's disks formations RAID 5, the RAID 5 with an address area, realize the reconstruction of RAID in the present embodiment in the following way:
In the several strips, each stick of each band is distributed in member's disk 0～3.Lost efficacy when stick D1, P2 place sector corruption in member's disk 1 cause member's disk 1 (shown in " * " in the square frame of expression member disk 1 among Fig. 3), when making RAID 5 become degrading state, wouldn't will should from RAID 5, kick out of by inefficacy member disk 1;
For the data in do not lose efficacy in the inefficacy member disk 1 stick D1, D4 and the stick D9, directly be copied in the HotSpare disk 1 of this inefficacy member disk 1 correspondence, if in copy procedure, I/O write operation visit stick D9 is arranged, so directly the data with the I/O write operation write in the HotSpare disk 1, if less than stick D9, then needing to copy other partial datas that the I/O write operation does not take place in the D9, the part of I/O write operation replenishes; And after copy is finished, stick D1, D4 in the inefficacy member disk 1 and stick D9 are labeled as disarmed state;
For the data in the stick P2 that lost efficacy in the inefficacy member disk 1 (stick P2 is owing to the wrong failure state that is marked as of I/O read-write), reading the data that belong in other normal member's disks 0, member's disk 2, the member's disk 3 among the RAID in the band 2 corresponding sticks respectively carries out being written in the HotSpare disk 1 of this inefficacy member disk 1 correspondence after Xor calculates.
Yet, suppose in inefficacy member disk 1 that the copying data in stick D1, D4, the D9 is finished and mark after, the data Xor in the stick P2 calculates when not beginning, member's disk 2 is marked as inefficacy (shown in " * " in the square frame of representing member's disk 2 among Fig. 3) because the I/O read-write of its stick D10 is wrong.
At this moment,, belong to different bands with band 2 under the inefficacy stick P2 in the inefficacy member disk 1 because the inefficacy stick D10 of inefficacy member disk 2 belongs to band 3, promptly not overlapping, therefore:
For the data in the failed stick P2 in the inefficacy member disk 1, still can read data in the corresponding stick that belongs to band 2 among the RAID in other normal member's disks 0 and member's disk 3 and inefficacy member disk 2 respectively and carry out Xor and calculate, and be written in the HotSpare disk 1 of this inefficacy member disk 1 correspondence.
For the stick D2, the P1 that did not lose efficacy in the inefficacy member disk 2, the data in the D7, directly be copied to respectively in the HotSpare disk 2 of this inefficacy member disk 2 correspondences, and after copy is finished, stick D2, P1, D7 in the inefficacy member disk 2 be labeled as disarmed state;
For the data in the failed stick D10 in the inefficacy member disk 2, reading data in the corresponding stick that belongs to band 3 among the RAID in other normal member's disks 0 and member's disk 3 and HotSpare disk 1 (the stick D9 that belongs to band 3 in the inefficacy member disk 1 has been marked as failure state) respectively carries out being written in the HotSpare disk 2 of this inefficacy member disk 2 correspondences after Xor calculates.
Certainly, before all data in inefficacy member disk 1 and member's disk 2 are finished recovery, if have stick D0 in member's disk 0 to lose efficacy again and/or member's disk 3 in stick D5 lost efficacy, perhaps, there are stick D3 and/or the stick P0 in member's disk 3 in member's disk 0 to lose efficacy again, then still can continue to rebuild.
As seen, the redundance of RAID 5 as shown in Figure 3 is 1 member's disk, and the fault tolerance degree is then more than 1 member's disk.
More than be to realizing the detailed description of the method that RAID rebuilds in the present embodiment.Below, again to realizing in the present embodiment that the RAID controller among RAID and this RAID describes.
Can realize in the present embodiment that the RAID that rebuilds comprises RAID controller and a plurality of member's disk that condenses together.Wherein, the RAID controller is according to realizing in the present embodiment that method that RAID rebuilds carries out corresponding operating and can realize that RAID rebuilds.
Referring to Fig. 4, the RAID controller in the present embodiment comprises: main control unit 401 and copy cell 402 and computing unit 403.
In the practical application, because the data computation of each inefficacy member disk also can need the data in other inefficacies member disk, and the corresponding data in other inefficacies member disk may be resumed to HotSpare disk, therefore, except main control unit 401 will cause owing to I/O read-write is wrong truly the fail address zone marker be the failure state, can be disarmed state further also by copy cell 402 with the not fail address zone marker of having finished copy (comprise and copy two kinds of situations of I/O write data) in each inefficacy member disk from the address area copy;
Like this, computing unit 403 not during the data in the zone, fail address, if this is not marked as disarmed state in the zone, fail address, then further reads the data in the corresponding address zone in the corresponding HotSpare disk in reading RAID.
In order to realize to zone, true fail address, and the mark of finishing the zone, not fail address of copy, can further comprise map unit 404 in the RAID controller, be stored as the corresponding mapping MAP table that each member's disk is provided with respectively, in the MAP table of each member's disk correspondence, each element is respectively to an address area in should member's disk, and can take place to be labeled as inefficacy after the I/O read-write mistake, and have finished copy in this address area (comprising) from address area copy and two kinds of situations of copy I/O write data in this address area, or it is invalid by being marked as after the calculating renewal.Specifically, each element can only be made of dibit, wherein a bit represent the corresponding address zone whether be marked as owing to the I/O read-write is wrong inefficacys, promptly represent zone, real fail address, another bit represent the corresponding address zone whether by since finished copy (comprising) from address area copy and two kinds of situations of copy I/O write data or by the calculating renewal be marked as invalid, be the zone, not fail address that data have been finished renewal.
In addition, main control unit 401 is further kicked out of RAID with inefficacy member disk after reconstruction is finished, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk.For the situation that realizes mark with the MAP table, main control unit 401 is after kicking out of RAID with inefficacy member disk, also should further delete the MAP table of this inefficacy member disk correspondence, and with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk after, further corresponding new MAP table is set for newcomer's disk.
Identical with the method that realizes the RAID reconstruction in the present embodiment, for RAID and the RAID controller thereof realizing rebuilding, address area in member's disk can be divided with stick, and correspondingly, each zone, fail address is one or more stick that includes the inefficacy sector; Perhaps, the address area in member's disk also can be divided with the sector, and correspondingly, each zone, fail address is one or more inefficacy sector.
The foregoing description mainly is to be the explanation that example is carried out with RAID 5, in the practical application, technical scheme in the foregoing description also is applicable to RAID 6, and only to be the RAID algorithm that adopts of RAID 6 slightly have any different and inefficacy member number of disks is slightly had any different compared to RAID 5.
The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to and replace and improvement etc., all should be included within protection scope of the present invention.
Claims (21)
1, a kind of method for reconstructing of redundancy array in disc RAID is characterized in that, being in has other member's disk failure again among the RAID of degrading state and the zone, fail address between member's disk of respectively losing efficacy when not overlapping, and this method for reconstructing comprises the steps:
For each inefficacy member disk data in the zone, fail address not, directly be copied in the HotSpare disk of this member's disk correspondence;
For the data in each inefficacy member disk zone, fail address, read the corresponding data in other normal member's disks among the RAID and other inefficacies member disk, and be written to after calculating in the HotSpare disk of this member's disk correspondence.
2, method for reconstructing as claimed in claim 1 is characterized in that, this method for reconstructing further will copy in each inefficacy member disk or be disarmed state by calculating the not fail address zone marker that upgrades;
And, during the corresponding data of this method for reconstructing in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
3, method for reconstructing as claimed in claim 2, it is characterized in that, in copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, this method for reconstructing directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
4, method for reconstructing as claimed in claim 3 is characterized in that, this method for reconstructing further is respectively the mapping MAP table that each member's disk is provided with a correspondence; In the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and I/O read-write error flag can take place in this address area for losing efficacy, and the data in this address area have copied or by calculate be marked as after upgrading invalid.
5, method for reconstructing as claimed in claim 4, it is characterized in that, each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by the calculating renewal.
As claim 4 or 5 described method for reconstructing, it is characterized in that 6, an address area is one or more sticks or is one or more sector.
7, as claim 4 or 5 described method for reconstructing, it is characterized in that, this method for reconstructing is after reconstruction is finished, further inefficacy member disk is kicked out of the MAP table of RAID, deletion inefficacy member disk correspondence, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
8, a kind of redundancy array in disc RAID of rebuilding realized, comprise: RAID controller and a plurality of member's disk that condenses together, it is characterized in that, when this RAID is in failure state, other member's disk failure are arranged again and the zone, fail address of respectively losing efficacy between member's disk not overlapping;
For each inefficacy member disk data in the zone, fail address not, the RAID controller directly is copied in the HotSpare disk of this member's disk correspondence;
For the data in each inefficacy member disk zone, fail address, the RAID controller reads the corresponding data in other normal member's disks among the RAID and other inefficacies member disk, and is written to after calculating in the HotSpare disk of this member's disk correspondence.
9, RAID as claimed in claim 8 is characterized in that, the RAID controller further will copy in each inefficacy member disk or be disarmed state by calculating the not fail address zone marker that upgrades;
And, during the corresponding data of RAID controller in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
10, RAID as claimed in claim 8, it is characterized in that, in copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, the RAID controller directly directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
11, RAID as claimed in claim 10 is characterized in that, further is respectively arranged with the mapping MAP table of a correspondence in the RAID controller for each member's disk; In the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and I/O read-write error flag can take place in this address area for losing efficacy, and the data in this address area have copied or by calculate be marked as after upgrading invalid.
12, RAID as claimed in claim 11, it is characterized in that, each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by the calculating renewal.
As claim 11 or 12 described RAID, it is characterized in that 13, an address area is one or more sticks or is one or more sector.
14, as claim 11 or 12 described RAID, it is characterized in that, the RAID controller is after reconstruction is finished, further inefficacy member disk is kicked out of the MAP table of RAID, deletion inefficacy member disk correspondence, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
15, a kind of RAID controller is connected in each member's disk and HotSpare disk among the RAID, it is characterized in that this RAID controller comprises:
Main control unit has other member's disk failure again and the zone, fail address between member's disk of respectively losing efficacy when not overlapping in being in the RAID of degrading state, keep enabling copy cell and computing unit;
Copy cell with inefficacy member disk data in the zone, fail address not, directly is copied in the HotSpare disk of this member's disk correspondence;
Computing unit, at the data in inefficacy member disk zone, fail address, read the data in the corresponding address zone in other normal member's disks among the RAID and other inefficacies member disk, and be written to after calculating in the HotSpare disk of this member's disk correspondence.
16, RAID controller as claimed in claim 15 is characterized in that, copy cell is a disarmed state with copied not fail address zone marker in each inefficacy member disk further; Computing unit will be a disarmed state by calculating the not fail address zone marker that upgrades in each inefficacy member disk further;
And, during the corresponding data of computing unit in reading zone, fail address not,, then further read in the corresponding HotSpare disk and these corresponding data in zone, fail address not if this data place is not marked as disarmed state in the zone, fail address.
17, RAID controller as claimed in claim 16, it is characterized in that, in copy procedure, have the not fail address zone in the inefficacy member disk that the I/O write operation takes place, copy cell directly is copied to the I/O write data in the HotSpare disk of this member's disk correspondence, and with this not the fail address zone marker be disarmed state.
18, RAID controller as claimed in claim 17, it is characterized in that, further comprise map unit in the RAID controller, be stored as the corresponding mapping MAP table that each member's disk is provided with respectively, in the MAP table of each member's disk correspondence, each element is respectively to address area in should member's disk, and can take place be labeled as inefficacy after the I/O read-write mistake in this address area, and finished in this address area copy or by calculate be marked as after upgrading invalid.
19, RAID controller as claimed in claim 18, it is characterized in that, each element is made of dibit, and wherein a bit represents whether the corresponding address zone is labeled as inefficacy owing to the I/O read-write is wrong, another bit represents whether the corresponding address zone is invalid owing to having copied or being marked as by calculating renewal back.
As claim 18 or 19 described RAID controllers, it is characterized in that 20, an address area is one or more sticks or is one or more sector.
21, as claim 18 or 19 described RAID controllers, it is characterized in that, main control unit is after reconstruction is finished, further inefficacy member disk is kicked out of the MAP table of RAID, deletion inefficacy member disk correspondence, and further with the HotSpare disk of inefficacy member disk correspondence as newcomer's disk, corresponding MAP table is set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2009100894183A CN101609420A (en) | 2009-07-17 | 2009-07-17 | Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2009100894183A CN101609420A (en) | 2009-07-17 | 2009-07-17 | Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101609420A true CN101609420A (en) | 2009-12-23 |
Family
ID=41483180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2009100894183A Pending CN101609420A (en) | 2009-07-17 | 2009-07-17 | Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101609420A (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833422A (en) * | 2010-03-08 | 2010-09-15 | 成都市华为赛门铁克科技有限公司 | Data processing method and device for disk array |
| CN101859265A (en) * | 2010-06-02 | 2010-10-13 | 杭州华三通信技术有限公司 | Disk array reconstruction method and device |
| CN101916173A (en) * | 2010-08-27 | 2010-12-15 | 杭州华三通信技术有限公司 | RAID (Redundant Array of Independent Disks) based data reading and writing method and system thereof |
| CN102147714A (en) * | 2011-03-17 | 2011-08-10 | 杭州宏杉科技有限公司 | Management method and management device of network storage system |
| CN102385542A (en) * | 2011-12-13 | 2012-03-21 | 云海创想信息技术（无锡）有限公司 | Method and device for RAID0 data backup |
| CN102981778A (en) * | 2012-11-15 | 2013-03-20 | 浙江宇视科技有限公司 | Redundant array of independent disks (RAID) array reconstruction method and device thereof |
| CN103064635A (en) * | 2012-12-19 | 2013-04-24 | 华为技术有限公司 | Distributed storage method and device |
| CN103488547A (en) * | 2013-09-24 | 2014-01-01 | 浪潮电子信息产业股份有限公司 | Rapid reconstruction method of RAID group fault hard disk |
| CN103534688A (en) * | 2013-05-29 | 2014-01-22 | 华为技术有限公司 | Data recovery method, storage equipment and storage system |
| CN103677669A (en) * | 2013-11-29 | 2014-03-26 | 华为技术有限公司 | Method and device for IO (input-output) dispatching |
| CN103729268A (en) * | 2014-01-15 | 2014-04-16 | 浪潮电子信息产业股份有限公司 | Data recovery method for RAID5 with two disks lost |
| WO2015058542A1 (en) * | 2013-10-21 | 2015-04-30 | 华为技术有限公司 | Reconstruction method and device for redundant array of independent disks |
| CN105094712A (en) * | 2015-09-30 | 2015-11-25 | 浙江宇视科技有限公司 | Data processing method and device |
| CN106371947A (en) * | 2016-09-14 | 2017-02-01 | 郑州云海信息技术有限公司 | Multi-fault disk data recovery method for RAID (Redundant Arrays of Independent Disks) and system thereof |
| CN106610788A (en) * | 2015-10-26 | 2017-05-03 | 华为技术有限公司 | Hard disk array control method and device |
| WO2019056948A1 (en) * | 2017-09-22 | 2019-03-28 | 华为技术有限公司 | Storage medium management method, device and readable storage medium |
| CN112000480A (en) * | 2020-08-25 | 2020-11-27 | 深圳忆联信息系统有限公司 | Method, device, equipment and medium for improving SSD full-disk scanning efficiency |
| CN114968129A (en) * | 2022-07-28 | 2022-08-30 | 苏州浪潮智能科技有限公司 | Disk array redundancy method, system, computer device and storage medium |
-
2009
- 2009-07-17 CN CNA2009100894183A patent/CN101609420A/en active Pending
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101833422A (en) * | 2010-03-08 | 2010-09-15 | 成都市华为赛门铁克科技有限公司 | Data processing method and device for disk array |
| CN101859265A (en) * | 2010-06-02 | 2010-10-13 | 杭州华三通信技术有限公司 | Disk array reconstruction method and device |
| CN101859265B (en) * | 2010-06-02 | 2012-09-26 | 杭州华三通信技术有限公司 | Disk array reconstruction method and device |
| CN101916173B (en) * | 2010-08-27 | 2013-08-28 | 杭州华三通信技术有限公司 | RAID (Redundant Array of Independent Disks) based data reading and writing method and system thereof |
| CN101916173A (en) * | 2010-08-27 | 2010-12-15 | 杭州华三通信技术有限公司 | RAID (Redundant Array of Independent Disks) based data reading and writing method and system thereof |
| CN102147714A (en) * | 2011-03-17 | 2011-08-10 | 杭州宏杉科技有限公司 | Management method and management device of network storage system |
| CN102147714B (en) * | 2011-03-17 | 2015-12-16 | 杭州宏杉科技有限公司 | A kind of management method of network store system and device |
| CN102385542A (en) * | 2011-12-13 | 2012-03-21 | 云海创想信息技术（无锡）有限公司 | Method and device for RAID0 data backup |
| CN102981778A (en) * | 2012-11-15 | 2013-03-20 | 浙江宇视科技有限公司 | Redundant array of independent disks (RAID) array reconstruction method and device thereof |
| CN102981778B (en) * | 2012-11-15 | 2016-11-16 | 浙江宇视科技有限公司 | A kind of RAID array method for reconstructing and device |
| CN103064635A (en) * | 2012-12-19 | 2013-04-24 | 华为技术有限公司 | Distributed storage method and device |
| CN103064635B (en) * | 2012-12-19 | 2016-08-24 | 华为技术有限公司 | Distributed storage method and distributed storage devices |
| CN103534688A (en) * | 2013-05-29 | 2014-01-22 | 华为技术有限公司 | Data recovery method, storage equipment and storage system |
| CN103534688B (en) * | 2013-05-29 | 2016-03-09 | 华为技术有限公司 | Data reconstruction method, memory device and storage system |
| CN103488547A (en) * | 2013-09-24 | 2014-01-01 | 浪潮电子信息产业股份有限公司 | Rapid reconstruction method of RAID group fault hard disk |
| WO2015058542A1 (en) * | 2013-10-21 | 2015-04-30 | 华为技术有限公司 | Reconstruction method and device for redundant array of independent disks |
| CN103677669A (en) * | 2013-11-29 | 2014-03-26 | 华为技术有限公司 | Method and device for IO (input-output) dispatching |
| CN103677669B (en) * | 2013-11-29 | 2016-08-17 | 华为技术有限公司 | The method and apparatus of scheduling IO |
| CN103729268A (en) * | 2014-01-15 | 2014-04-16 | 浪潮电子信息产业股份有限公司 | Data recovery method for RAID5 with two disks lost |
| CN105094712A (en) * | 2015-09-30 | 2015-11-25 | 浙江宇视科技有限公司 | Data processing method and device |
| CN105094712B (en) * | 2015-09-30 | 2019-01-11 | 浙江宇视科技有限公司 | A kind of data processing method and device |
| CN106610788A (en) * | 2015-10-26 | 2017-05-03 | 华为技术有限公司 | Hard disk array control method and device |
| CN106371947A (en) * | 2016-09-14 | 2017-02-01 | 郑州云海信息技术有限公司 | Multi-fault disk data recovery method for RAID (Redundant Arrays of Independent Disks) and system thereof |
| CN106371947B (en) * | 2016-09-14 | 2019-07-26 | 郑州云海信息技术有限公司 | A kind of multiple faults disk data reconstruction method and its system for RAID |
| WO2019056948A1 (en) * | 2017-09-22 | 2019-03-28 | 华为技术有限公司 | Storage medium management method, device and readable storage medium |
| US11237929B2 (en) | 2017-09-22 | 2022-02-01 | Huawei Technologies Co., Ltd. | Method and apparatus, and readable storage medium |
| US11714733B2 (en) | 2017-09-22 | 2023-08-01 | Huawei Technologies Co., Ltd. | Method and apparatus, and readable storage medium |
| CN112000480A (en) * | 2020-08-25 | 2020-11-27 | 深圳忆联信息系统有限公司 | Method, device, equipment and medium for improving SSD full-disk scanning efficiency |
| CN112000480B (en) * | 2020-08-25 | 2023-12-05 | 深圳忆联信息系统有限公司 | Method, device, equipment and medium for improving SSD full-disk scanning efficiency |
| CN114968129A (en) * | 2022-07-28 | 2022-08-30 | 苏州浪潮智能科技有限公司 | Disk array redundancy method, system, computer device and storage medium |
| CN114968129B (en) * | 2022-07-28 | 2022-12-06 | 苏州浪潮智能科技有限公司 | Disk array redundancy method, system, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101609420A (en) | Realize method and the redundant arrays of inexpensive disks and the controller thereof of rebuilding of disc redundant array | |
| US8307159B2 (en) | System and method for providing performance-enhanced rebuild of a solid-state drive (SSD) in a solid-state drive hard disk drive (SSD HDD) redundant array of inexpensive disks 1 (RAID 1) pair | |
| CN104035830B (en) | A kind of data reconstruction method and device | |
| CN103049222B (en) | A kind of RAID5 writes IO optimized treatment method | |
| CN104246707B (en) | storage system and storage device | |
| US8839028B1 (en) | Managing data availability in storage systems | |
| KR100701563B1 (en) | Storage control apparatus and method | |
| CN101916173B (en) | RAID (Redundant Array of Independent Disks) based data reading and writing method and system thereof | |
| US8356292B2 (en) | Method for updating control program of physical storage device in storage virtualization system and storage virtualization controller and system thereof | |
| CN102508724B (en) | Disk bad block processing method based on soft RAID (redundant array of independent disks) | |
| CN102799533B (en) | Method and apparatus for shielding damaged sector of disk | |
| CN102981778B (en) | A kind of RAID array method for reconstructing and device | |
| JPH06504863A (en) | Storage array with copyback cache | |
| JPH05505264A (en) | Non-volatile memory storage of write operation identifiers in data storage devices | |
| CN105531677A (en) | Raid parity stripe reconstruction | |
| CN102508620B (en) | Method for processing RAID5 (Redundant Array of Independent Disks) bad sector | |
| CN102520890B (en) | RS (Reed-Solomon) - DRAID( D redundant array of independent disk) system based on GPUs (graphic processing units) and method for controlling data of memory devices | |
| US20190243553A1 (en) | Storage system, computer-readable recording medium, and control method for system | |
| CN101840360A (en) | Rapid reconstruction method and device of RAID (Redundant Array of Independent Disk) system | |
| CN110187830A (en) | A kind of method and system accelerating disk array reconstruction | |
| TW201017404A (en) | System and method for loose coupling between RAID volumes and drive groups | |
| US10409682B1 (en) | Distributed RAID system | |
| CN102164165B (en) | Management method and device for network storage system | |
| CN111124262A (en) | Management method, apparatus and computer readable medium for Redundant Array of Independent Disks (RAID) | |
| CN103544995B (en) | A kind of bad track repairing method and bad track repairing device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20091223 |