CN101976176B - Method for constructing disk array of horizontal grouping parallel distribution check - Google Patents
Method for constructing disk array of horizontal grouping parallel distribution check Download PDFInfo
- Publication number
- CN101976176B CN101976176B CN201010256699XA CN201010256699A CN101976176B CN 101976176 B CN101976176 B CN 101976176B CN 201010256699X A CN201010256699X A CN 201010256699XA CN 201010256699 A CN201010256699 A CN 201010256699A CN 101976176 B CN101976176 B CN 101976176B
- Authority
- CN
- China
- Prior art keywords
- data
- block
- disk
- band
- sub
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Abstract
The invention relates to a method for constructing a disk array of horizontal grouping parallel distribution check, which is suitable for continuous data storage and belongs to the technical field of redundant arrays of independent disks. Aiming at the characteristics of continuous data storage, a disk array SHE-RAID 5 of the horizontal grouping parallel distribution check is designed. The method mainly comprises the steps of: distributing storage data in the SHF-RAID 5, reading and writing the storage data, checking a generation mode of the data, buffering, prereading and optimizing the data when the data is written, switching and scheduling work states of a disk, and the like. The disk array SHE-RAID 5 has a better redundant data protecting mechanism, and can remarkably reduce the power consumption and prolong the service life.
Description
Technical field
The present invention relates to a kind of Redundant Array of Independent Disks (RAID) (Redundant Arrays of Independent Disks, RAID) construction method, be particularly related to the construction method of the disk array of a kind of horizontal type grouping parallel distributed verification, be applicable to the continuous data storage, belong to the Redundant Array of Independent Disks (RAID) technical field.
Background technology
In modern field of storage, for reliability that improves the storage data and input, the output performance that improves storage system, people have designed the several data storage scheme, the normally various types of Redundant Array of Independent Disks (RAID) of these data storage scheme (Redundant Arrays of Independent Disks, RAID).By using specific hardware or software, RAID joins together a plurality of physical storage devices such as disk, forms a unified logical memory device.
Below technical term commonly used among the RAID is made an explanation:
Band: be called Stripe again; Being the set of the relevant piecemeal in position on the different disk of disk array, is the unit that organizes stick on the different disk.
Striping: be called Striping again; Be meant the data block that one section continuous data is divided into identical size, every segment data be written to the method on the different disk of disk array respectively.
Disk mirroring is meant copy source data to one or more disks,
Error correction is meant utilizes certain computing, as XOR, generates and the preservation redundant data, can utilize redundant data, the data of losing or makeing mistakes on the regeneration disk.
XOR computing: XOR.
Relatively Chang Yong RAID has RAID0, RAID1, RAID5, RAID6, RAID10 etc.Wherein RAID0 does not have redundant ability, and RAID1 has just done mirror image to disk.Other 3 kinds of arrays are made up of a plurality of disks respectively, and they are with the disk write data of mode in array of band, and parity data leaves on each disk in the array.Each band of RAID5 contains 1 check block, supports to damage one of them disk arbitrarily, come data reconstruction by the parity block on other disk; Each band of RAID6 contains 2 check blocks, supports to damage arbitrarily wherein two disks, comes restore data by the parity block on other disk; RAID10 carries out mirror image to disk groups earlier, again disk is carried out striping, so it does not contain parity block, after a disk failures, pass through its corresponding mirror drive restore data, 50% disk failures under maximum permission diverse locations under the perfect condition, the situation of worst is to damage simultaneously with a pair of mirrored disk.
In RAID 5, a band comprises one group of data block that logic is adjacent, and these data block store are in array on the different memory device, the data that RAID 5 can correct or regenerate and store on the disk, and because a plurality of disks are united the redundant data of using a disk size, so the required redundant data amount of unit valid data is less, especially when comprising a plurality of disk among the RAID 5, this advantage is more obvious.As the RAID 5 that 8 disks constitute, redundant data accounts for 1 disk size, for all storing 1/8 of data.And because the data block that RAID 5 each band are distributed on the different disk is bigger, guaranteed that most of I/O operation can both drop in the data block of a disk, checking data also is a distributed store, increases the concurrent possibility of many I/O so greatly.Because RAID 5 has outstanding combination property, for most of commerce server provide the cost performance of optimizing, so obtained widespread use.
Yet; in the continuous data field of storage; in application such as video monitoring, VTL (VTL), continuous data protection (CDP); continuous data stream is written in the disk in proper order; what carry out is sequential storage; the address of promptly storing data is continuous, satisfies the principle of locality of storage space visit, does not therefore have the I/O concurrency.Disk array commonly used, in order to improve data access speed and to guarantee the I/O concurrency, continuous data is disperseed to store on the different disks, though the local adjacent disk of one or a few logic of each only visit, most disks do not have task must dally yet in the array, so not only waste a large amount of energy, also increased the invalid line duration of disk simultaneously greatly, seriously shortened the serviceable life of disk.
Along with the disk development of technology; the continuous readwrite bandwidth of monolithic disk is significantly improved; store for continuous data; if can be when having the redundant data protection mechanism; make full use of the transmission bandwidth of monolithic disk; according to the actual task demand; make one or a few disk work in the array; finish store tasks; other disk scheduling that does not have task is to halted state; be that disk spindle and magnetic head motor quit work, not only can save a large amount of energy, and can also effectively utilize the serviceable life of disk.
The energy consumption of disk under the different operating state sees Table 1.
The energy consumption of table 1 disk under the different operating state
| The disk duty | Read-write operation | Idling conditions | Halted state |
| Average power consumption | 11.5W | 8.5W | Be about 0W |
Summary of the invention
The objective of the invention is to overcome the deficiency that prior art exists, satisfying under the prerequisite of setting memory property, propose the construction method of the disk array of a kind of horizontal type grouping parallel distributed verification.The disk array of horizontal type grouping parallel distributed verification abbreviates disk array SHE-RAID 5 as.
The objective of the invention is to be achieved through the following technical solutions.
The construction method of a kind of disk array SHE-RAID 5 is applied in the scene that disk array comprises N disk, and wherein N 〉=3 and N are positive integer; N disk is configured to the disk matrix of 1 row * N row; Whole disks are carried out band to be divided, be divided into N band and also be each band serial number, N storage block arranged on each band, wherein 1 is check block, N-1 is data block in addition, check block is by drawing by XOR with the data block of the N-1 in the band, and N check block is distributed on the different disk of array in proper order.
(wherein j represents the sequence number of this storage block place disk for i, a j) storage block among the expression disk array SHE-RAID 5, and (i j) is positioned on the disk j storage block X with X; I represents this storage block place band; 1≤i, j≤N; Then (i N+1-i) is check block to storage block X, with P (i, j) expression; Other storage block is a data block, and (wherein u represents the sequence number of data block place band for u, v) expression, and 1≤u≤N, v represent the sequence number (ignoring check block) of the data block of this data block in affiliated band, and sequence number is since 1,1≤v≤N-1 with L.
Data block L (u, v) the corresponding relation with storage block satisfies formula 1:
In order to improve continuous memory bandwidth, and obtain different continuous memory bandwidths, to satisfy the demand of different memory rates, N-1 on each a band data block is divided into p (p 〉=2 and p are positive integer) group, every group comprises the individual data block of q (q 〉=1 and q are positive integer, and the q value is determined the demand of continuous bandwidth according to the actual storage task); And relation: N-1=p * q below satisfying.
Each data block is divided into K equal-sized data sub-block, and K 〉=2 and K are positive integer, are followed successively by the sub-block number of total data according to the band order then, and the data sub-block in the same band is according to the group serial number.That is: at first be that the 1st the sub-block number of data is 1 in the 1st data block of the 1st group of first band; Be in first band in the 1st group the 2nd data block the 1st the sub-block number of data be 2; Be in first band in the 1st group q the data block the 1st the sub-block number of data be q; Be then in first band in the 1st group the 1st data block the 2nd the sub-block number of data be q+1; Be that the 2nd the sub-block number of data is q+2 in the 2nd data block of the 1st group of first band; Be in first band in the 1st group q the data block the 2nd the sub-block number of data be 2 * q; By that analogy, be in first band in the 1st group the 1st data block K the sub-block number of data be (K-1) * q+1; Be that K the sub-block number of data is (K-1) * q+2 in the 2nd data block of the 1st group of first band; Be in first band in the 1st group q the data block K the sub-block number of data be K * q;
Be that the 1st the sub-block number of data is K * q+1 in the 1st data block of the 2nd group of first band then; Be in first band in the 2nd group the 2nd data block the 1st the sub-block number of data be K * q+2; Be in first band in the 2nd group q the data block the 1st the sub-block number of data be (K+1) * q; Be then in first band in the 2nd group the 1st data block the 2nd the sub-block number of data be (K+1) * q+1; Be that the 2nd the sub-block number of data is (K+1) * q+2 in the 2nd data block of the 2nd group of first band; Be in first band in the 2nd group q the data block the 2nd the sub-block number of data be (K+2) * q; By that analogy, be in first band in the 2nd group the 1st data block K the sub-block number of data be 2 * K * q-q+1; Be that K the sub-block number of data is 2 * K * q-q+2 in the 2nd data block of the 2nd group of first band; Be in first band in the 2nd group q the data block K the sub-block number of data be 2 * K * q;
Be that the 1st the sub-block number of data is (p-1) * K * q+1 in the 1st data block of first band p group then; Be that the 1st the sub-block number of data is (p-1) * K * q+2 in the 2nd data block that p organizes in first band; Be that the 1st the sub-block number of data is (p-1) * K * q+q in q the data block that p organizes in first band; Be that the 2nd the sub-block number of data is (p-1) * K * q+q+1 in the 1st data block that p organizes in first band then; Be that the 2nd the sub-block number of data is (p-1) * K * q+q+2 in the 2nd data block of first band p group; Be that the 2nd the sub-block number of data is (p-1) * K * q+q * 2 in q the data block that p organizes in first band; By that analogy, be that K the sub-block number of data is p * K * q-q+1 in the 1st data block that p organizes in first band; Be that K the sub-block number of data is p * K * q-q+2 in the 2nd data block of first band p group; Be that K the sub-block number of data is p * K * q in q the data block that p organizes in first band;
By that analogy, according to identical numbering principle, the data sub-block that brings to the N band for second is numbered.Number adjacent data sub-block, its logical address is adjacent.
The adjacent group of sequence number in the same band, its logical address is adjacent; In the adjacent ribbons, last group of last band is adjacent with first group logical address of back one band.
Each check block also is divided into K equal-sized syndrome piece, the syndrome piece is got by deviation post is identical in the band of place N-1 data sub-disk array piece XOR, and the syndrome piece that such N-1 data sub-block and XOR thereof generate is formed a verification bar.
Above-mentioned layout type, but q data block in the concurrent access group makes store tasks be distributed in one group again, and all the other p-1 group data block place disks do not have task, being convenient to like this dispatch does not have the disk of task to enter into halted state, to reach the purpose energy-conservation and loss of reduction disk.
Described method further comprises: the method for control disk duty.
The disk duty comprises: stop, operation, ready three kinds of states.
During the disk halted state, disk spindle stops the rotation, magnetic head stops tracking;
During the disk running status, disk is being carried out read-write operation, comprises rotating shaft rotation, magnetic head tracking;
During the disk ready state, expression will be read and write, and disk spindle rotation this moment, magnetic head be tracking not;
When carrying out continuous write operation, in the overwhelming majority time, have only one group of data block place disk and 1 check block place disk (q+1 disk) to be in running status, do not have the disk of work to be scheduled for halted state, to reach purpose of energy saving, disk duty conversion scheduling strategy is specially:
For being provided with a semaphore, each disk (represents that with Sem i 1≤i≤N), the initial value of signalization amount Sem i is 0.
The 1st step:, calculate affiliated band u of this visit data place data block and affiliated group (representing) with r according to the logical address (representing) of visit data with LBA.Obtain by formula 2:
Wherein, stripe size is represented the big or small sum of the size of each data block in the band, and the group size is the big or small sum of each data block in the group, and MOD represents modular arithmetic, adds 1 expression band numbering and group # all since 1.
Then, use formula 1 further to obtain interior q the data block L (u of u band r group, r * q-q+1), L (u, r * q-q+2) ..., L (u, the affiliated disk of r * q-q+q), and be N+1-u with disk under the check block in the band, and the place disk and the disk N+1-u of q data block in the u band r group handled respectively: be in halted state as if this disk, then be dispatched to ready state.Band sequence number u and group sequence number r are respectively applied for the band sequence number and the group sequence number of record current accessed operation.
The 2nd step: when the place disk of q data block in u band r organizes and disk N+1-u were in ready state, when this q+1 disk carried out write operation, the disk that carries out write operation transferred running status to by ready state respectively;
The 3rd step: if this visit finishes in u band r group, after then this visit finishes, disk reaches and transfers ready state with disk N+1-u under the check block in the band to by running status under q the data block in the u band r group, withdraws from this scheduling then; Otherwise, the t moment (t is that disk transfers the needed time of ready state to by halted state) before u band r group access finishes, determine band numbering (representing) and the group # (representing) of adjacent set with R with I, and disk N+1-I under the check block in disk and the I band under q data block in the I band R group handled respectively: be in halted state as if this disk, then be dispatched to ready state; If this disk is in running status, then the disk semaphore of this disk adds 1.After u band r group access finishes, disk N+1-u under the check block in disk and the u band under q the data block in the u band r group is handled respectively: the semaphore of this disk correspondence is subtracted 1, if the semaphore of this disk is less than 0, then transfer this disk to halted state by running status, and the semaphore of resetting this disk is 0; Otherwise, transfer this disk to ready state by running status.
The 4th step: the u value of record current accessed place band sequence number is updated to I, the value that writes down the r of current accessed place group sequence number is updated to R, repeat the 2nd and went on foot for the 4th step.
When carrying out continuous read operation, in the overwhelming majority times, also have only one group of data block place disk and 1 check block place disk (q+1 disk) to be in running status, disk duty conversion scheduling strategy is identical with continuous write operation.
Preferably, when carrying out continuous read operation, check block place disk quits work.
For each write operation, owing to only upgrade affiliated q piece disk of one group of data block and the data on the check block dish, being equivalent to reading among the RAID 5 rewrites, promptly be written in parallel to new data D1, D2 ... during Dq, need usefulness formula 3 and formula 4 generation new data D1, D2 ..., the new checking data P under the Dq on the verification bar:
Xi＝(Di)XOR(Di′) (3)
P＝(X1)XOR(X2)XOR…XOR(Xq)XOR?P′ (4)
Wherein, Di ' is the legacy data on the new data Di writing position in the disk, and 1≤i≤q, P ' are the old checking datas on the legacy data Di ' place verification bar, also are the old checking datas on the new checking data P writing position, and XOR represents XOR.
In order to generate new checking data P, (time was very short when hardware was realized except carrying out XOR, can ignore) outside, must reach old checking data P ' by first parallel read-out legacy data Di ', for disk, each write operation all comprises to be read legacy data earlier and writes two operations of new data again, in order to reduce the switching times of magnetic head reading and writing data, make full use of the disk bandwidth, described method further comprises: the optimization process method of write operation is specially:
Set up a new data buffer zone, being used for the buffer memory application program writes the new data of disk array SHE-RAID 5, data pre-head district, q Geju City and Geju City checking data are read the district in advance, be used for respectively depositing and from disk array SHE-RAID 5, read in advance, legacy data corresponding and old checking data with new data to be written, each reads to distinguish equal and opposite in direction in advance, and much larger than the new data buffer zone.The sub-piece of legacy data of the sub-piece correspondence of each new data and the sub-piece of old checking data are read in advance before write operation and are read the district in advance.
Carry out following operation steps then:
The 2.1st step: when new data buffer zone when be empty, from the new data buffer zone, take out q data sub-block at every turn, be split as belong to the data sub-block D1, the D2 that organize interior q piece disk ..., behind the Dq, be written in parallel to q piece disk, be sent to the XOR device simultaneously;
The 2.2nd step: the corresponding sub-piece D1 ' of legacy data, D2 ' in the data pre-head district, q Geju City ..., Dq ' is sent to the XOR device, carry out XOR with the data sub-block that the 2.1st step was sent here, the sub-piece X1 of generation variance data, X2 ..., Xq, wherein Xi=(Di) XOR (Di ');
The 2.3rd step: the sub-piece X1 of variance data, X2 ... Xq is sent to the XOR device, the sub-piece P ' of old checking data corresponding during old checking data is read to distinguish in advance is sent to the XOR device, and XOR generates the sub-piece P of new checking data, P=(X1) XOR (X2) XOR ... XOR (Xq) XOR (P ');
The 2.4th step: the sub-piece P renewal of new checking data data sub-block D1, D2 ..., the syndrome piece under the Dq in the verification bar.
If write data initial logical address LBA can not be divided exactly by the size of q data sub-block, and during the individual data sub-block of surplus k (k is a positive integer), or in the new data buffer zone when difference k and not enough q data sub-block, then only get this k data sub-block and write corresponding disk, and utilize the legacy data and the old checking data of same verification bar correspondence, generate new checking data according to above method and get final product, all the other q-k disk does not have writing task on the same group.
Owing to can directly read sub-piece of legacy data and the sub-piece of old checking data from reading the district in advance, disk just can be carried out write operation continuously, after having only data sub-block in reading in advance to distinguish to use up, just once read in advance to fill up and read the district in advance, by reading in advance, the read operation of a plurality of dispersions is converted into a continuous read operation, has effectively reduced the switching times of magnetic head, can be provided with according to actual conditions and read the district in advance for enough big.
According to above analysis,, estimate that easily the bandwidth of writing continuously of this disk array SHE-RAID 5 is about qBW when reading the district in advance enough greatly the time
Disk/ 2, wide about qBR continuously reads tape
Disk, BW wherein
DiskWrite bandwidth, BR continuously for the monolithic disk
DiskThe continuous tape reading of monolithic disk is wide, and q is disk number in the group.
Described method further comprises: the disposal route of failed disk being carried out reading and writing data; Be specially:
Break down as disk, in the time of can't carrying out the normal data read-write, activate disk array SHE-RAID this moment
All disks enter ready state in 5, suspend the duty of using described disk duty control method control disk array SHE-RAID 5; Being divided into following 2 kinds of situations then handles:
Situation 1: if check block place disk breaks down, the data on the piece place disk that then continues to read and write data, but disk array SHE-RAID this moment 5 is in irredundant data check state; Behind the new building to be replaced,, use all the other storage blocks of this storage block place band, carry out XOR and regenerate each storage block on the faulty disk.
Situation 2: if data block place disk breaks down, then according to the reading and writing operation is handled respectively:
Read operation: use other storage block of this data block place band to carry out XOR, generate this data block; Behind the new building to be replaced,, use other storage blocks of this piecemeal place band to carry out XOR, recover the data on this dish each storage block on the faulty disk.
Write operation: 2 kinds of processing modes are arranged:
The 1st kind of processing mode: the check block of directly write data being write with band also writes down the check block that writes, and behind the faulty disk to be replaced, carries out reduction of data, data reproduction and regeneration checking data.Reduction of data promptly is moved back into the origin-location to the data block that is deposited with in the check block disk; Data reproduction promptly recovers non-reduced storage block on the new data dish, utilizes all the other storage blocks of place band to carry out the XOR realization.The regeneration checking data is the check block that writes that regenerates record, utilizes all the other storage block XORs of place band to realize.
The 2nd kind of processing mode: use standby dish to keep in to write the data on the failed disk, after changing failed disk, carry out reduction of data, data reproduction and regeneration checking data.
After changing normal condition over to, enable described disk duty control method once more the duty of disk array SHE-RAID 5 is controlled.
Beneficial effect
The construction method of disk array SHE-RAID 5 of the present invention compared with the prior art, has the following advantages:
1, significantly reduces energy resource consumption.Its data layout has good redundant data protection mechanism; take grouping strategy simultaneously; each has only one group of in running order and working time long enough of disk constantly, is fit to all the other off working state disks of scheduling to halted state, to reach purpose of energy saving.
2, can select continuous memory bandwidth according to mission requirements.After taking grouping strategy, can set the q value according to mission requirements, the q piece disk in the concurrent access group adopt data buffering and the measure of reading in advance to improve simultaneously to read to rewrite to writing the influence of bandwidth, so it provides the continuous readwrite bandwidth that meets mission requirements.
3, fully increase the service life.Because the inoperative disk halted state that is scheduled in the array, stopped the high speed rotating of disk spindle, and disk starts, stand-by time is fully long at interval, so can reduce probability that disk breaks down and the serviceable life that fully prolongs it.
Description of drawings
Fig. 1 is disk, band, data block, the check block numbering synoptic diagram among the disk array SHE-RAID 5 in the specific embodiment of the present invention;
Fig. 2 is the synoptic diagram that divides into groups among the disk array SHE-RAID 5 in the specific embodiment of the present invention;
Fig. 3 is data sub-block and a syndrome block number synoptic diagram among the disk array SHE-RAID 5 in the specific embodiment of the present invention;
Fig. 4 is that the disk duty in the specific embodiment of the present invention shifts synoptic diagram;
Fig. 5 is that the write operation in the specific embodiment of the present invention is optimized synoptic diagram.
Embodiment
The present invention is described in detail below in conjunction with the drawings and specific embodiments.
In the present embodiment, disk array SHE-RAID 5 is made up of 5 Seagate Cheetah 10K disks, and the capacity of monolithic disk is 300GB, and its correlation parameter sees Table 2.5 disks are carried out band divide, as shown in Figure 1, be divided into 5 bands, be numbered 1～5.5 storage blocks are arranged on each band, and wherein 1 storage block is a check block, and other 4 storage blocks are data block; The 5th data block of the 4th data block of the 3rd data block of the 2nd data block of the 1st data block of the 5th disk, the 4th disk, the 3rd disk, the 2nd disk, the 1st disk is check block; Other storage block is a data block; Check block is drawn by XOR by 4 data blocks in this check block place band.
The correlation parameter of table 2 Seagate Cheetah 10K disk
In order to improve continuous memory bandwidth, 4 data blocks on each band are divided into 2 groups, every group comprises 2 data blocks.The grouping situation of the data block on each band as shown in Figure 2.
Data sub-block is designed to 4KB, so each data block is divided into K=19,660,800 data sub-block are followed successively by the sub-block number of total data according to the band order then, and the data sub-block in the same band is according to the group serial number, as shown in Figure 3.Promptly at first be that the 1st the sub-block number of data is 1 in the 1st data block of the 1st group of first band; Be in first band in the 1st group the 2nd data block the 1st the sub-block number of data be 2; Be then in first band in the 1st group the 1st data block the 2nd the sub-block number of data be 3; Be that the 2nd the sub-block number of data is 4 in the 2nd data block of the 1st group of first band; By that analogy, be in first band in the 1st group the 1st data block K the sub-block number of data be (K-1) * 2+1; Be that K the sub-block number of data is (K-1) * 2+2 in the 2nd data block of the 1st group of first band;
Be that the 1st the sub-block number of data is K * 2+1 in the 1st data block of the 2nd group of first band then; Be in first band in the 2nd group the 2nd data block the 1st the sub-block number of data be K * 2+2; Be then in first band in the 2nd group the 1st data block the 2nd the sub-block number of data be (K+1) * 2+1; Be that the 2nd the sub-block number of data is (K+1) * 2+2 in the 2nd data block of the 2nd group of first band; By that analogy, be in first band in the 2nd group the 1st data block K the sub-block number of data be 4 * K-1; Be that K the sub-block number of data is 4 * K in the 2nd data block of the 2nd group of first band;
By that analogy, according to identical numbering principle, the data sub-block that brings to the 5th band for second is numbered.Number adjacent data sub-block, its logical address is adjacent.
The adjacent group of sequence number in the same band, its logical address is adjacent; In the adjacent ribbons, last group of last band is adjacent with first group logical address of back one band.
Each check block also is divided into K equal-sized syndrome piece, and the syndrome piece get by 4 identical data sub-disk array piece XORs of deviation post in the band of place, verification bar of syndrome piece composition that such 4 data sub-block and XOR thereof generate.
Above-mentioned layout type, but 2 data blocks in the concurrent access group make store tasks be distributed in one group again, and all the other 1 group of data block place disks do not have task, being convenient to like this dispatch does not have the disk of task to enter into halted state, to reach the purpose energy-conservation and loss of reduction disk.
Each check block is divided into K=19,660,800 equal-sized syndrome pieces, and be 1～5 according to band order number consecutively; Described syndrome piece is drawn by XOR by 4 identical data sub-block of deviation post in each data block in this syndrome piece place band.
When carrying out continuous read-write operation, in the overwhelming majority time, have only 3 disks (2 data block place disks and 1 check block place disk) to be in running status, do not have the disk of work to be scheduled for halted state, to reach purpose of energy saving, disk duty conversion scheduling strategy is specially:
For each disk is provided with a semaphore Sem i, 1≤i≤5, initial value is 0.
When to disk array SHE-RAID 5 visit datas, the state exchange of disk as shown in Figure 4: the data block that calculates the visit data place is the 1st group of the 4th band.On disk 1 and disk 3, the check block on the 4th band if disk 1,2 or 3 is in halted state, then is dispatched to ready state at the 2nd disk to 2 data blocks in the 1st group of the 4th band respectively, as among Fig. 4 1. shown in.When on disk 1,2,3, reading and writing data, disk 1,2,3 transfer running status to by ready state, as among Fig. 4 2. shown in.By calculating, this visit finishes in the 2nd group of the 4th band, and therefore t determines the 2nd group of interior 2 data blocks of adjacent set the 4th band respectively on the 4th, 5 disks constantly before the 4th band the 1st group access finishes, and the check block of the 4th band is on the 2nd disk; The 4th, 5 disks are dispatched to ready state by halted state, the semaphore of the 2nd disk are added 1, i.e. Sem 2=1.After the 4th band the 1st group access finishes, the semaphore of disk 1,2,3 is subtracted 1, i.e. Sem 1=-1, Sem 2=0, therefore Sem 3=-1 transfers disk 1 and 3 to halted state by running status, as among Fig. 4 4. shown in, and the semaphore of initial correspondence is 0 again, Sem 1=0, Sem 3=0; Disk 2 transfers ready state to by running status, as among Fig. 4 3. shown in.When reading and writing data on disk 2,4,5, disk 2,4,5 transfers running status to by ready state, as among Fig. 4 2. shown in.This visit finishes in the 2nd group of the 4th band, therefore when to disk 2,4,5 visit end, transfers disk 2,4,5 to ready state by running status.
For normal continuous data storage, in the overwhelming majority time, disk array SHE-RAID 5 has only 3 disk work, has only the fraction time, have 5 disks to be in read-write or idling conditions (adding 1 group of disk that starts in advance), so its power consumption is about same disc and count 3/5 of RAID 5.
In internal memory, set up 4 buffer zones, be respectively 1 new data buffer zone, data pre-head district, 2 Geju City and 1 Geju City checking data and read the district in advance.Wherein, legacy data read in advance the district and old checking data read to distinguish equal and opposite in direction in advance; Legacy data is read in advance to distinguish with old checking data and is read to distinguish 10 times that are the new data buffer zone in advance.The sub-piece of legacy data of the sub-piece correspondence of new data and the sub-piece of old checking data are read in advance before write operation and are read the district in advance.
Carry out following operation steps then:
The 2.1st step: when new data buffer zone when be empty, 2 data sub-block of taking-up from the new data buffer zone at every turn, be split as belong to data sub-block D1, the D2 that organizes interior 2 disks after, be written in parallel to 2 disks, be sent to the XOR device simultaneously, as the 1. process among Fig. 5;
The 2.2nd step: the sub-piece D1 ' of legacy data, the D2 ' of correspondence in the data pre-head district, 2 Geju City, be sent to the XOR device, carry out XOR with the data sub-block that the 2.1st step was sent here, generate the sub-piece X1 of variance data, X2, Xi=(Di) XOR (Di ') wherein, i=1,2, as the 2. process among Fig. 5;
The 2.3rd step: the sub-piece X1 of variance data, X2 are sent to the XOR device, as the 3. process among Fig. 5, the sub-piece P ' of old checking data corresponding during old checking data is read to distinguish in advance is sent to the XOR device, as the 4. process among Fig. 5, XOR generates the sub-piece P of new checking data, P=(X1) XOR (X2) XOR (P ');
The 2.4th step: the old syndrome piece that upgrades in the corresponding verification bar is the sub-piece P of new checking data, as the 5. process among Fig. 5.
By data buffering with read in advance to handle, can make the bandwidth of writing continuously of disk array SHE-RAID 5 be about q.BW
Disk/ 2, wide about q.BR continuously reads tape
Disk, BW wherein
DiskWrite bandwidth, BR continuously for the monolithic disk
DiskWide for the continuous tape reading of monolithic disk, q is disk number in the group.The disk of present embodiment is Seagate Cheetah10K, its wide 85MB/s that is that reads tape continuously, and writing bandwidth continuously is 84MB/s.The bandwidth of writing continuously that can get disk array SHE-RAID5 is about 84MB/s, and the wide 170MB/s of being about that reads tape continuously writes when promptly can satisfy the high definition video steaming of 42 road 2MB/s, reads in the time of 85 road such video flowings.
Break down as disk, can't carry out normal data when read-write, activate that all disks enter ready state among the disk array SHE-RAID5 this moment, suspend the duty of using described disk duty control method control disk array SHE-RAID 5; The situation that is divided into then in following 2 is handled:
Break down as disk, in the time of can't carrying out the normal data read-write, activate disk array SHE-RAID this moment
All disks enter ready state in 5, suspend the duty of using described disk duty control method control disk array SHE-RAID 5; The situation that is divided into then in following 2 is handled:
Situation 1: if check block place disk breaks down, the data on the piece place disk that then continues to read and write data, but this moment, disk array was in irredundant data check state; Behind the new building to be replaced,, use all the other storage blocks of this storage block place band, carry out XOR and regenerate each storage block on the faulty disk.
Situation 2: if data block place disk breaks down, then according to the reading and writing operation is handled respectively:
Read operation: use other storage block of this data block place band to carry out XOR, generate this data block; Behind the new building to be replaced,, use other storage blocks of this piecemeal place band to carry out XOR, recover the data on this dish each storage block on the faulty disk.
Write operation: 2 kinds of processing modes are arranged:
The 1st kind of processing mode: the check block of directly write data being write with band also writes down the check block that writes, and behind the faulty disk to be replaced, carries out reduction of data, data reproduction and regeneration checking data.Reduction of data promptly is moved back into the origin-location to the data block that is deposited with in the check block disk; Data reproduction promptly recovers non-reduced storage block on the new data dish, utilizes all the other storage blocks of place band to carry out the XOR realization.The regeneration checking data is the check block that writes that regenerates record, utilizes all the other storage block XORs of place band to realize.
The 2nd kind of processing mode: use standby dish to keep in to write the data on the failed disk, after changing failed disk, carry out reduction of data, data reproduction and regeneration checking data.
After changing normal condition over to, enable described disk duty control method once more the duty of disk array SHE-RAID 5 is controlled.
The above only is a preferred implementation of the present invention; should be understood that; for those skilled in the art; under the prerequisite that does not break away from the principle of the invention; can also make some improvement; perhaps part technical characterictic wherein is equal to replacement, these improvement and replace and also should be considered as protection scope of the present invention.
Claims (5)
1. the construction method of the disk array of horizontal type grouping parallel distributed verification; The disk array of horizontal type grouping parallel distributed verification abbreviates disk array SHE-RAID 5 as; It is characterized in that:
Be applied in the scene that disk array comprises N disk, wherein N 〉=3 and N are positive integer; N disk is configured to the disk matrix of 1 row * N row; Whole disks are carried out band to be divided, be divided into N band and also be each band serial number, N storage block arranged on each band, wherein 1 is check block, N-1 is data block in addition, check block is by drawing by XOR with the data block of the N-1 in the band, and N check block is distributed on the different disk of array in proper order;
(wherein j represents the sequence number of this storage block place disk for i, a j) storage block among the expression disk array SHE-RAID 5, and (i j) is positioned on the disk j storage block X with X; I represents this storage block place band; 1≤i, j≤N; Then (i N+1-i) is check block to storage block X, with P (i, j) expression; Other storage block is a data block, and (wherein u represents the sequence number of data block place band for u, v) expression, and 1≤u≤N, v represent the sequence number of the data block of this data block in affiliated band, and sequence number is since 1,1≤v≤N-1 with L;
Data block L (u, v) the corresponding relation with storage block satisfies formula 1:
In order to improve continuous memory bandwidth, and obtain different continuous memory bandwidths, to satisfy the demand of different memory rates, N-1 on each a band data block is divided into the p group, p 〉=2 and p are positive integer, every group comprises q data block, and q 〉=1 and q are positive integer, and the q value is determined the demand of continuous bandwidth according to the actual storage task; And relation: N-1=p * q below satisfying;
Each data block is divided into K equal-sized data sub-block, and K 〉=2 and K are positive integer, are followed successively by the sub-block number of total data according to the band order then, and the data sub-block in the same band is according to the group serial number; That is: at first be that the 1st the sub-block number of data is 1 in the 1st data block of the 1st group of first band; Be in first band in the 1st group the 2nd data block the 1st the sub-block number of data be 2; Be in first band in the 1st group q the data block the 1st the sub-block number of data be q; Be then in first band in the 1st group the 1st data block the 2nd the sub-block number of data be q+1; Be that the 2nd the sub-block number of data is q+2 in the 2nd data block of the 1st group of first band; Be in first band in the 1st group q the data block the 2nd the sub-block number of data be 2 * q; By that analogy, be in first band in the 1st group the 1st data block K the sub-block number of data be (K-1) * q+1; Be that K the sub-block number of data is (K-1) * q+2 in the 2nd data block of the 1st group of first band; Be in first band in the 1st group q the data block K the sub-block number of data be K * q;
Be that the 1st the sub-block number of data is K * q+1 in the 1st data block of the 2nd group of first band then; Be in first band in the 2nd group the 2nd data block the 1st the sub-block number of data be K * q+2; Be in first band in the 2nd group q the data block the 1st the sub-block number of data be (K+1) * q; Be then in first band in the 2nd group the 1st data block the 2nd the sub-block number of data be (K+1) * q+1; Be that the 2nd the sub-block number of data is (K+1) * q+2 in the 2nd data block of the 2nd group of first band; Be in first band in the 2nd group q the data block the 2nd the sub-block number of data be (K+2) * q; By that analogy, be in first band in the 2nd group the 1st data block K the sub-block number of data be 2 * K * q-q+1; Be that K the sub-block number of data is 2 * K * q-q+2 in the 2nd data block of the 2nd group of first band; Be in first band in the 2nd group q the data block K the sub-block number of data be 2 * K * q;
Be that the 1st the sub-block number of data is (p-1) * K * q+1 in the 1st data block of first band p group then; Be that the 1st the sub-block number of data is (p-1) * K * q+2 in the 2nd data block that p organizes in first band; Be that the 1st the sub-block number of data is (p-1) * K * q+q in q the data block that p organizes in first band; Be that the 2nd the sub-block number of data is (p-1) * K * q+q+1 in the 1st data block that p organizes in first band then; Be that the 2nd the sub-block number of data is (p-1) * K * q+q+2 in the 2nd data block of first band p group; Be that the 2nd the sub-block number of data is (p-1) * K * q+q * 2 in q the data block that p organizes in first band; By that analogy, be that K the sub-block number of data is p * K * q-q+1 in the 1st data block that p organizes in first band; Be that K the sub-block number of data is p * K * q-q+2 in the 2nd data block of first band p group; Be that K the sub-block number of data is p * K * q in q the data block that p organizes in first band;
By that analogy, according to identical numbering principle, the data sub-block that brings to the N band for second is numbered; Number adjacent data sub-block, its logical address is adjacent;
The adjacent group of sequence number in the same band, its logical address is adjacent; In the adjacent ribbons, last group of last band is adjacent with first group logical address of back one band;
Each check block also is divided into K equal-sized syndrome piece, the syndrome piece is got by deviation post is identical in the band of place N-1 data sub-disk array piece XOR, and the syndrome piece that such N-1 data sub-block and XOR thereof generate is formed a verification bar;
Above-mentioned layout type, but q data block in the concurrent access group makes store tasks be distributed in one group again, and all the other p-1 group data block place disks do not have task, being convenient to like this dispatch does not have the disk of task to enter into halted state, to reach the purpose energy-conservation and loss of reduction disk;
Described method further comprises: the method for control disk duty;
The disk duty comprises: stop, operation, ready three kinds of states;
During the disk halted state, disk spindle stops the rotation, magnetic head stops tracking;
During the disk running status, disk is being carried out read-write operation, comprises rotating shaft rotation, magnetic head tracking;
During the disk ready state, expression will be read and write, and disk spindle rotation this moment, magnetic head be tracking not;
When carrying out continuous write operation, in the overwhelming majority time, have only one group of data block place disk and 1 check block place disk to be in running status, do not have the disk of work to be scheduled for halted state, to reach purpose of energy saving, disk duty conversion scheduling strategy is specially:
For each disk is provided with a semaphore, represent with Sem i, 1≤i≤N, the initial value of signalization amount Sem i is 0;
The 1st step: according to the logical address of visit data, represent, calculate affiliated band u of this visit data place data block and affiliated group, represent with r with LBA; Obtain by formula 2:
Wherein, stripe size is represented the big or small sum of each data block in the band, and the group size is the big or small sum of each data block in the group, and MOD represents modular arithmetic, adds 1 expression band numbering and group # all since 1;
Then, use formula 1 further to obtain interior q the data block L (u of u band r group, r * q-q+1), L (u, r * q-q+2) ..., L (u, the affiliated disk of r * q-q+q), and be N+1-u with disk under the check block in the band, and the place disk and the disk N+1-u of q data block in the u band r group handled respectively: be in halted state as if this disk, then be dispatched to ready state; Band sequence number u and group sequence number r are respectively applied for the band sequence number and the group sequence number of record current accessed operation;
The 2nd step: when the place disk of q data block in u band r organizes and disk N+1-u were in ready state, when this q+1 disk carried out write operation, the disk that carries out write operation transferred running status to by ready state respectively;
The 3rd step: if this visit finishes in u band r group, after then this visit finishes, disk reaches and transfers ready state with disk N+1-u under the check block in the band to by running status under q the data block in the u band r group, withdraws from this scheduling then; Otherwise, t constantly before u band r group access finishes, t is that disk transfers the needed time of ready state to by halted state, determine the band numbering of adjacent set, represent with I, and group #, represent with R, and disk N+1-I under the check block in disk and the I band under q data block in the I band R group handled respectively: be in halted state as if this disk, then be dispatched to ready state; If this disk is in running status, then the disk semaphore of this disk adds 1; After u band r group access finishes, disk N+1-u under the check block in disk and the u band under q the data block in the u band r group is handled respectively: the semaphore of this disk correspondence is subtracted 1, if the semaphore of this disk is less than 0, then transfer this disk to halted state by running status, and the semaphore of resetting this disk is 0; Otherwise, transfer this disk to ready state by running status;
The 4th step: the u value of record current accessed place band sequence number is updated to I, the value that writes down the r of current accessed place group sequence number is updated to R, repeat the 2nd and went on foot for the 4th step;
When carrying out continuous read operation, in the overwhelming majority times, also have only one group of data block place disk and 1 check block place disk to be in running status, disk duty conversion scheduling strategy is identical with continuous write operation.
2. the construction method of the disk array of a kind of horizontal type grouping parallel distributed as claimed in claim 1 verification; It is characterized in that:
Preferably, when carrying out continuous read operation, check block place disk quits work.
3. the construction method of the disk array of a kind of horizontal type grouping parallel distributed as claimed in claim 1 verification; It is characterized in that:
Described method further comprises: the optimization process method of write operation is specially:
Set up a new data buffer zone, being used for the buffer memory application program writes the new data of disk array SHE-RAID 5, data pre-head district, q Geju City and Geju City checking data are read the district in advance, be used for respectively depositing and from disk array SHE-RAID 5, read in advance, the legacy data corresponding and old checking data with new data to be written; The sub-piece of legacy data of the sub-piece correspondence of each new data and the sub-piece of old checking data are read in advance before write operation and are read the district in advance;
Carry out following operation steps then:
The 2.1st step: when new data buffer zone when be empty, from the new data buffer zone, take out q data sub-block at every turn, be split as belong to the data sub-block D1, the D2 that organize interior q piece disk ..., behind the Dq, be written in parallel to q piece disk, be sent to the XOR device simultaneously;
The 2.2nd step: the corresponding sub-piece D1 ' of legacy data, D2 ' in the data pre-head district, q Geju City ..., Dq ' is sent to the XOR device, carry out XOR with the data sub-block that the 2.1st step was sent here, the sub-piece X1 of generation variance data, X2 ..., Xq, wherein Xi=(Di) XOR (Di ');
The 2.3rd step: the sub-piece X1 of variance data, X2 ... Xq is sent to the XOR device, the sub-piece P ' of old checking data corresponding during old checking data is read to distinguish in advance is sent to the XOR device, and XOR generates the sub-piece P of new checking data, P=(X1) XOR (X2) XOR ... XOR (Xq) XOR (P ');
The 2.4th step: the sub-piece P renewal of new checking data data sub-block D1, D2 ..., the syndrome piece under the Dq in the verification bar;
If write data initial logical address LBA can not be divided exactly by the size of q data sub-block, and during surplus k data sub-block, k is a positive integer, or in the new data buffer zone when difference k and not enough q data sub-block, then only get this k data sub-block and write corresponding disk, and utilize the legacy data and the old checking data of same verification bar correspondence, and generating new checking data according to above method and get final product, all the other q-k disk does not have writing task on the same group;
Owing to can directly read sub-piece of legacy data and the sub-piece of old checking data from reading the district in advance, disk just can be carried out write operation continuously, after having only data sub-block in reading in advance to distinguish to use up, just once read in advance to fill up and read the district in advance, by reading in advance, the read operation of a plurality of dispersions is converted into a continuous read operation, has effectively reduced the switching times of magnetic head, can be provided with according to actual conditions and read the district in advance for enough big.
4. the construction method of the disk array of a kind of horizontal type grouping parallel distributed as claimed in claim 3 verification; It is characterized in that:
Preferably, legacy data is read the district in advance and old checking data reads to distinguish equal and opposite in direction in advance, and much larger than the new data buffer zone.
5. as the construction method of the disk array of the described a kind of horizontal type grouping of one of claim 1 to 4 parallel distributed verification; It is characterized in that:
Described method further comprises: the disposal route of failed disk being carried out reading and writing data; Be specially:
Break down as disk, can't carry out normal data when read-write, activate that all disks enter ready state among the disk array SHE-RAID5 this moment, suspend the duty of using described disk duty control method control disk array SHE-RAID 5; Being divided into following 2 kinds of situations then handles:
Situation 1: if check block place disk breaks down, the data on the piece place disk that then continues to read and write data, but disk array SHE-RAID this moment 5 is in irredundant data check state; Behind the new building to be replaced,, use all the other storage blocks of this storage block place band, carry out XOR and regenerate each storage block on the faulty disk;
Situation 2: if data block place disk breaks down, then according to the reading and writing operation is handled respectively:
Read operation: use other storage block of this data block place band to carry out XOR, generate this data block; Behind the new building to be replaced,, use other storage blocks of this piecemeal place band to carry out XOR, recover the data on this dish each storage block on the faulty disk;
Write operation: 2 kinds of processing modes are arranged:
The 1st kind of processing mode: the check block of directly write data being write with band also writes down the check block that writes, and behind the faulty disk to be replaced, carries out reduction of data, data reproduction and regeneration checking data; Reduction of data promptly is moved back into the origin-location to the data block that is deposited with in the check block disk; Data reproduction promptly recovers non-reduced storage block on the new data dish, utilizes all the other storage blocks of place band to carry out the XOR realization; The regeneration checking data is the check block that writes that regenerates record, utilizes all the other storage block XORs of place band to realize;
The 2nd kind of processing mode: use standby dish to keep in to write the data on the failed disk, after changing failed disk, carry out reduction of data, data reproduction and regeneration checking data;
After changing normal condition over to, enable described disk duty control method once more the duty of disk array SHE-RAID 5 is controlled.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010256699XA CN101976176B (en) | 2010-08-19 | 2010-08-19 | Method for constructing disk array of horizontal grouping parallel distribution check |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010256699XA CN101976176B (en) | 2010-08-19 | 2010-08-19 | Method for constructing disk array of horizontal grouping parallel distribution check |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101976176A CN101976176A (en) | 2011-02-16 |
| CN101976176B true CN101976176B (en) | 2011-12-14 |
Family
ID=43576064
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010256699XA Active CN101976176B (en) | 2010-08-19 | 2010-08-19 | Method for constructing disk array of horizontal grouping parallel distribution check |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101976176B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102830942B (en) * | 2012-06-28 | 2016-06-22 | 记忆科技(深圳)有限公司 | The method of solid state hard disc disk array mapping and solid state hard disc thereof |
| CN102789370B (en) * | 2012-06-29 | 2015-11-25 | 浙江宇视科技有限公司 | A kind of RAID array synchronous method and device |
| CN104866239B (en) * | 2015-05-26 | 2018-09-14 | 浙江宇视科技有限公司 | A kind of video storage method of high availability |
| CN107832168B (en) * | 2017-10-13 | 2020-10-16 | 记忆科技(深圳)有限公司 | Solid state disk data protection method |
| WO2019090756A1 (en) * | 2017-11-13 | 2019-05-16 | 清华大学 | Raid mechanism-based data storage system for sharing resources globally |
| CN108334280B (en) * | 2017-12-28 | 2021-01-08 | 深圳创新科技术有限公司 | RAID5 disk group fast reconstruction method and device |
| CN111858189A (en) * | 2019-04-29 | 2020-10-30 | 伊姆西Ip控股有限责任公司 | Handling of storage disk offline |
| CN111857552A (en) * | 2019-04-30 | 2020-10-30 | 伊姆西Ip控股有限责任公司 | Storage management method, electronic device and computer program product |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101098476A (en) * | 2007-07-12 | 2008-01-02 | 刘西红 | Method of optimizing video information output of data reading in magnetic disk array |
| CN101174196A (en) * | 2006-10-31 | 2008-05-07 | 英业达股份有限公司 | Method for magnetic disk array grade in changing operation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6640282B2 (en) * | 2000-01-25 | 2003-10-28 | Hewlett-Packard Development Company, L.P. | Hot replace power control sequence logic |
-
2010
- 2010-08-19 CN CN201010256699XA patent/CN101976176B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101174196A (en) * | 2006-10-31 | 2008-05-07 | 英业达股份有限公司 | Method for magnetic disk array grade in changing operation |
| CN101098476A (en) * | 2007-07-12 | 2008-01-02 | 刘西红 | Method of optimizing video information output of data reading in magnetic disk array |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101976176A (en) | 2011-02-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101976175B (en) | Method for constructing disk array by horizontal grouping parallel concentrated verification | |
| CN101976174B (en) | Method for constructing energy-saving disk array of vertical configuration distribution check | |
| CN101976177B (en) | Method for constructing vertical grouped disk array capable of being subject to parallel centralized check | |
| CN101976176B (en) | Method for constructing disk array of horizontal grouping parallel distribution check | |
| CN102033716B (en) | Method for constructing energy-saving type disc array with double discs for fault tolerance | |
| CN102023820B (en) | Method for constructing double-disk fault-tolerant vertical packet parallel access disk array | |
| CN102023819B (en) | Method for constructing double-disk fault tolerance horizontal grouping and parallel access disk array | |
| CN101976178B (en) | Method for constructing vertically-arranged and centrally-inspected energy-saving disk arrays | |
| CN103049222B (en) | A kind of RAID5 writes IO optimized treatment method | |
| CN102122235B (en) | RAID4 (redundant array of independent disks) system and data reading and writing method thereof | |
| CN101976179B (en) | Construction method of vertical-type grouping parallel-distributed and checked disk array | |
| CN111984200B (en) | Mass video data storage system | |
| CN103793182A (en) | Scalable storage protection | |
| CN101093434A (en) | Method of improving input and output performance of raid system using matrix stripe cache | |
| CN102147713B (en) | Method and device for managing network storage system | |
| CN104956309A (en) | Storage system and data management method | |
| CN101625586A (en) | Method, equipment and computer for managing energy conservation of storage device | |
| CN101470640B (en) | Method and system for providing deferred maintenance on storage subsystems | |
| CN103870352A (en) | Method and system for data storage and reconstruction | |
| CN102164165A (en) | Management method and device for network storage system | |
| CN105930097A (en) | Distributed verification redundant array of independent disks capable of eliminating local parallel read-modify-write operation | |
| CN101414244A (en) | A kind of methods, devices and systems of processing data under network environment | |
| CN100470507C (en) | Method for rewriting in magnetic disc array structure | |
| CN104063184A (en) | Disk Recording Method, Nonvolatile Storage Medium Thereof And Electronic Device | |
| Otoo et al. | Dynamic data reorganization for energy savings in disk storage systems |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| DD01 | Delivery of document by public notice |
Addressee: Beijing has to fly with Ji Polytron Technologies Inc responsible for collection Document name: Notification that Application Deemed not to be Proposed |