CN100389383C - Method for solving writing cavity in magnetic disk redundancy array - Google Patents
Method for solving writing cavity in magnetic disk redundancy array Download PDFInfo
- Publication number
- CN100389383C CN100389383C CNB2006100836926A CN200610083692A CN100389383C CN 100389383 C CN100389383 C CN 100389383C CN B2006100836926 A CNB2006100836926 A CN B2006100836926A CN 200610083692 A CN200610083692 A CN 200610083692A CN 100389383 C CN100389383 C CN 100389383C
- Authority
- CN
- China
- Prior art keywords
- write
- master disk
- disk controller
- controller
- daily record
- 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
Abstract
The present invention discloses a method for solving the problem of writing cavities in redundant arrays of disks, which comprises the steps that a host computer sends a write command to a master disk controller, the master disk controller receives the write command and generates a checking diary, and then, the master disk controller writes the checking diary in a memory and executes write operation; if the master disk controller breaks down in the process of the write operation, a backup disk controller reads the checking diary from the memory to reestablish check data subblocks. The method of the present invention effectively stores the checking diary, rapidly reestablishes the check data subblocks, and effectively solves the problem of writing cavities and the bottleneck of the system performance in redundant arrays of disks. The method of the present invention avoids the defects that a great deal of time is wasted when the problem of writing cavities in redundant arrays of inexpensive disks is solved in the prior art and checking data sub blocks are reestablished, and a great deal of time is also wasted for waiting for information transfer between the master disk controller and the backup disk during write operation.
Description
Technical field
The present invention relates to computing machine read-write control field, be meant the method for writing cavity in a kind of solution redundant arrays of inexpensive disks (Write Hole) especially.
Background technology
At the initial stage of computing machine development, the price of " high capacity " hard disk is also quite high, the main method that solves the data storage safety issue is to use equipment such as magnetic tape station to back up, though this method can guarantee the safety of data, consults with back-up job all quite loaded down with trivial details.1987, three slip-stick artist's co-presents of University of California Berkeley be entitled as the paper of " A Case ofRedundant Array of Inexpensive Disks (Redundant Array of Inexpensive Disc scheme) ", its basic thought is exactly that many less, the relatively inexpensive hard disk drives of capacity are carried out organic assembling, form a kind of in manageability, the virtual disk that capacity, performance, reliabilty and availability are all higher makes its performance surpass the big hard disk of a costliness.This design philosophy is accepted very soon, and from then on raid-array (RAID, Redundant Array OfIndependent Disks) technology has obtained widespread use, and data storage has entered quicker, safer, the more cheap New Times.
The RAID technology is divided into several different grades.Mainly comprise several standards of RAID0～RAID7, can provide different speed respectively, security and cost performance, RAID5 is one of them.The RAID5 technology mainly is presented as on all disks access data and parity information across in read-write control on the disk.As shown in Figure 1, main frame 101 links to each other with the virtual memory of being made up of 5 disks 104 by master disk controller 102, backup diskette controller 103, and master disk controller 102 links to each other by high availability path 10 5 with backup diskette controller 103.Deblocking and checking data piecemeal cross-distribution are on each disk.In per 5 pieces 1 checking data piecemeal is arranged all, corresponding to checking data piecemeal 4, deblocking 6,7,8,9 is corresponding to checking data piecemeal 5 as data piecemeal 0,1,2,3.Generating the checking data piecemeal has two kinds of algorithms, and the checking data piecemeal that adopts two kinds of algorithms that identical deblocking is operated generation is identical.First kind of algorithm be when the deblocking that will write more for a long time, for example write 0,1,2 three data piecemeal,
The read-write of input and output for reading legacy data piecemeal 3, is write new data piecemeal 0 in proper order like this, writes new data piecemeal 1, writes new data piecemeal 2, write check deblocking 4.Second kind of algorithm be when the deblocking that will write more after a little while, for example only write 0 deblocking,
The read-write that needs input and output is like this read old checking data piecemeal 4 in proper order for reading legacy data piecemeal 0, writes new data piecemeal 0, writes new checking data piecemeal 4.When main frame 101 is write 0,1,2 deblockings of virtual disk by master disk controller 102, read, the calculation check process all finishes.If but after having write 0,1 deblocking, write in the process of the 2nd deblocking master disk controller 102 machine of delaying, and the checking data piecemeal of this moment is not also write on the virtual disk, and this checking data piecemeal still is old checking data piecemeal.Backup diskette controller 103 can start carries out write operation, and backup diskette controller 103 may use second kind of algorithm, calculates new checking data piecemeal with old checking data block data.It is wrong obtaining new checking data piecemeal like this.After this carry out data again and write fashionablely, it also is wrong generating new checking data piecemeal with wrong checking data piecemeal.If after this disk failure takes place, will cause data failure, RAID5 writing cavity effect appears.For solving the writing cavity problem, must become the verification daily record to each key element record of uncompleted write operation and preserve, rebuild the checking data piecemeal after being used for.Head it off has following two kinds of methods in the prior art.
A kind of method is to set up high availability path 10 5 between master disk controller 102 and the backup diskette controller 103, before beginning for each write operation, master disk controller 102 is sent to backup diskette controller 103 with the verification daily record by path 10 5, after master disk controller 102 waited for that backup diskette controller 103 is confirmed to receive the verification daily record, master disk controller 102 began to carry out write operation.If master disk controller 102 machine of delaying in writing process, backup diskette controller 103 is rebuild the checking data piecemeal according to the verification daily record.
Another kind method is after master disk controller 102 breaks down, and when main frame is switched to backup diskette controller 103, thinks that all checking data piecemeals are all unreliable.Checking data piecemeals all on the virtual disk is reinitialized one time.
First method need be provided at active and standby with the path 10 5 of setting up a high availability between the Magnetic Disk Controller in the prior art, and itself can not break down this passage at work, otherwise the verification daily record that the master disk controller generates can't be sent to the backup diskette controller.Master disk controller 102 all will stop write operation earlier when each write operation in the preservation process of verification daily record, wait receives after backup diskette controller 103 receives the affirmation information of verification daily record and just carries out write operation, and the efficient of write operation has been produced bigger influence.Also influenced the handling property of master disk controller in the message exchange of being carried out with the backup diskette controller.Second method is owing to will rebuild all checking data piecemeals, and for the bigger situation of data in magnetic disk amount, the time of reconstruction can be longer, to the bigger influence of normal use generation of system.
Summary of the invention
In view of this, the technical matters of the present invention's solution is to provide a kind of method that solves writing cavity in the redundant arrays of inexpensive disks.This method has not only solved the writing cavity problem, and the wait when having avoided the checking data piecemeal all to rebuild a large amount of time of waste and write operation at interval.
For addressing the above problem, the invention provides a kind of method that solves writing cavity in the redundant arrays of inexpensive disks, comprising:
A, main frame send write order to the master disk controller;
B, master disk controller are received write order, generate the verification daily record;
C, master disk controller are with in the verification daily record write store and carry out write operation;
If quit work verification daily record in the backup diskette controller readout memory in the d master disk controller write operation process;
E, backup diskette controller are rebuild the checking data piecemeal according to the verification daily record;
F, main frame overwriting data piecemeal.
Wherein, among the described step c in write store and the described steps d readout memory be in a block RAM, to write, read.
Wherein, among the described step c in write store and the described steps d readout memory be in two block RAMs, to write, read in the block RAM therein.
Described two block RAMs are independently-powered respectively.
Wherein, described step c further comprises: if the master disk controller completes successfully write operation, then send the information of writing successfully to main frame.
The present invention also provides a kind of method that solves writing cavity in the redundant arrays of inexpensive disks, it is characterized in that, comprising:
A, main frame send write order to storage control module;
B, storage control module are sent to the master disk controller with write order, and the verification daily record that self generates is sent to the backup diskette controller;
C, master disk controller are carried out write operation according to write order, and the backup diskette controller is written to the verification daily record in the internal memory of self;
If d master disk controller quits work in the process of write operation, the backup diskette controller is rebuild the checking data piecemeal according to the verification daily record, and the verification deblocking is write on the disk again.
Wherein, described step c further comprises:
C71, master disk controller are carried out write operation success back and are sent the information of writing successfully to storage control module, and backup diskette controller write check daily record success back sends the information of writing successfully to storage control module;
If c72, storage control module send the information of writing successfully to main frame after receiving the information of writing successfully of master disk controller and backup diskette controller, write failure information otherwise send to main frame.
Wherein, described storage control module is software module in the main frame.
Wherein, described storage control module is independent equipment.
The present invention compares with prior art, preserves the verification daily record effectively, and rebuilds the checking data piecemeal apace, solves writing cavity problem and system performance bottleneck in the redundant arrays of inexpensive disks efficiently.The stand-by period that information is transmitted during at write operation between plenty of time of having wasted when having avoided prior art to rebuild the checking data piecemeal when solving redundant arrays of inexpensive disks writing cavity problem and the primary, spare Magnetic Disk Controller.
Description of drawings
Fig. 1 is the system construction drawing that prior art solves the RAID5 writing cavity;
Fig. 2 is the process flow diagram of the inventive method;
Fig. 3 is the process flow diagram of the inventive method embodiment one;
Fig. 4 is the process flow diagram of the inventive method embodiment two.
Embodiment
Method of the present invention can be preserved the verification daily record effectively, rebuilds the checking data piecemeal by the verification daily record and solves the writing cavity problem, improves the service efficiency of system efficiently.Avoid the wait interval of read/writing control method when all rebuilding a large amount of time of checking data piecemeal waste and write operation of prior art.
Referring to Fig. 2, method of the present invention comprises:
Step S201: main frame sends write request;
Step S202: master disk controller write check daily record is to shared drive and carry out write operation;
Step S203: the master disk controller is delayed behind the machine, the reader check daily record and rebuild the checking data piecemeal according to the verification daily record from shared drive of backup diskette controller.
Describe method workflow of the present invention in detail below in conjunction with Fig. 3.Suppose that each itemize has 5 piecemeals in the present RAID5 system, the branch block size is 30k, data itemize 0={ deblocking 0, deblocking 1, deblocking 2, deblocking 3, checking data piecemeal 4}={1,1,1,1,0};
Step S301: main frame is initiated { start address=0, size=30k, the write request of deblocking 0=0} to the master disk controller.
Step S302: the master disk controller generates verification daily record, { start address=0, size=30k}
Step S303: the master disk controller writes the verification daily record among the RAM;
Step S304: the master disk controller is carried out write operation.
Master disk controller calculation check deblocking,
According to write request computational data itemize.Because the checking data piecemeal is invisible to the user, data itemize size=minute block size 30k * (5-1)=120k.Data itemize=start address 0/ itemize size=0/120=0 then.
The master disk controller finds data itemize 0, carry out write operation and write that { at this moment the piecemeal situation { 0 for deblocking 0=0, checking data piecemeal=1}, 1,1,1,1}, behind the intact write operation of master disk controller correct execution, execution in step S305, machine, then execution in step S306 if the master disk controller is delayed in carrying out the write operation process.
Step S305: the master disk controller returns the information write successfully to main frame.
Step S306: the backup diskette controller is read the verification daily record among the RAM.
Suppose that the master disk controller write the machine of delaying behind the deblocking 0, this moment, the checking data piecemeal was not also write on the disk, this moment, the piecemeal situation was: { 0,1,1,1,0}, this moment, the checking data piecemeal was wrong, so the master disk controller can not send the information write successfully to main frame, and the backup diskette controller is the reader check daily record from shared drive.
Step S307: the backup diskette controller is rebuild the checking data piecemeal according to the verification daily record.
The backup diskette controller recomputates the verification itemize according to the verification daily record, finds the checking data piecemeal, recomputates the checking data piecemeal,
At this moment the piecemeal situation 0,1,1,1,1}, this moment the checking data piecemeal be again correct.
Step S308: main frame overwriting data piecemeal.
Because main frame is not received the information write successfully, can retransmit write request, the backup diskette controller can be according to write request overwriting data piecemeal 0 and checking data piecemeal 4.
For RAM storer among this embodiment can be one or two, when adopting two independently-powered RAM to work, the master disk controller is written to the verification daily record in two block RAMs, and after the master disk controller quit work, the backup diskette controller is the reader check daily record from a block RAM.If in the middle of this process, when a block RAM broke down cisco unity malfunction, an other block RAM can be protected the verification daily record effectively, and the backup diskette controller still can the reader check daily record be rebuild the checking data piecemeal from an other block RAM.
The present invention is a kind of incessantly in the method that solves writing cavity by preservation verification daily record, also can realize by adding the memory controller module, describes another embodiment of method of the present invention in detail below in conjunction with accompanying drawing 4.Suppose that each itemize has 5 piecemeals in the present RAID5 system, the branch block size is 30k, data itemize 0={ deblocking 0, deblocking 1, deblocking 2, deblocking 3, checking data piecemeal 4}={1,1,1,1,0};
Step S401: main frame sends { start address=0, size=30k, deblocking 0=0} write request to storage control module;
Step S402: { start address=0, size=30k} were sent to the master disk controller with write order, simultaneously the verification daily record are sent to the backup diskette controller in generation verification daily record after storage control module was received write order;
Step S403: master disk control receives that write order carries out write operation, and the backup diskette controller receives that the verification daily record writes in the internal memory simultaneously.
Master disk controller calculation check deblocking,
According to write request computational data itemize.Because the checking data piecemeal is invisible to the user, data itemize size=minute block size 30k * (5-1)=120k.Data itemize=start address 0/ itemize size=0/120k=0 then.
The master disk controller finds data itemize 0, carries out write operation, will deblocking 0=0, checking data piecemeal 4=1} are written in the disk array, and at this moment the piecemeal situation be 0,1,1,1,1}.
{ deblocking 0=0} is saved in the local internal memory backup diskette controller for start address=0, size=30k with the verification daily record.
After master disk controller and backup diskette controller are received the information of storage control module transmission, concurrent working simultaneously, order in no particular order.
Step S404: the master disk controller machine of delaying.
Step S405: the backup diskette controller is rebuild the checking data piecemeal.
The machine if the master disk controller is delayed in the process of work, then the backup diskette controller is rebuild the checking data piecemeal according to the verification daily record of receiving.
The backup diskette controller recomputates the verification itemize according to the verification daily record, finds the checking data piecemeal, recomputates the checking data piecemeal.
At this moment the piecemeal situation 0,1,1,1,1}, this moment the checking data piecemeal be again correct.
Step S406: master disk control is returned and is write success, and the backup diskette controller returns writes success.
If primary, spare Magnetic Disk Controller has all completed successfully write operation, then primary, spare Magnetic Disk Controller returns the information write successfully to storage control module;
Step S407: have only storage control module to receive the information of writing successfully of master disk controller and backup diskette controller, just return the information write successfully to main frame; Otherwise return to main frame and to write failure information, write request can take place in main frame again.
For method of the present invention, the form of realization is diversified.Within the spirit and principles in the present invention all, any modification of being done, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (9)
1. a method that solves writing cavity in the redundant arrays of inexpensive disks is characterized in that, comprising:
A, main frame send write order to the master disk controller;
B, master disk controller are received write order, generate the verification daily record;
C, master disk controller are with in the verification daily record write store and carry out write operation;
If quit work verification daily record in the backup diskette controller readout memory in the d master disk controller write operation process;
E, backup diskette controller are rebuild the checking data piecemeal according to the verification daily record;
F, main frame overwriting data piecemeal.
2. method according to claim 1 is characterized in that, among the described step c in write store and the described steps d readout memory be in a block RAM, to write, read.
3. method according to claim 1 is characterized in that, among the described step c in write store and the described steps d readout memory be in two block RAMs, to write, read in the block RAM therein.
4. according to the described method of claim 3, it is characterized in that described two block RAMs are independently-powered respectively.
5. method according to claim 1 is characterized in that, described step c further comprises: if the master disk controller completes successfully write operation, then send the information of writing successfully to main frame.
6. a method that solves writing cavity in the redundant arrays of inexpensive disks is characterized in that, comprising:
A, main frame send write order to storage control module;
B, storage control module are sent to the master disk controller with write order, and the verification daily record that self generates is sent to the backup diskette controller;
C, master disk controller are carried out write operation according to write order, and the backup diskette controller is written to the verification daily record in the internal memory of self;
If d master disk controller quits work in the process of write operation, the backup diskette controller is rebuild the checking data piecemeal according to the verification daily record, and the verification deblocking is write on the disk again.
7. method according to claim 6 is characterized in that, described step c further comprises:
C71, master disk controller are carried out write operation success back and are sent the information of writing successfully to storage control module, and backup diskette controller write check daily record success back sends the information of writing successfully to storage control module;
If c72, storage control module send the information of writing successfully to main frame after receiving the information of writing successfully of master disk controller and backup diskette controller, write failure information otherwise send to main frame.
8. method according to claim 6 is characterized in that, described storage control module is software module in the main frame.
9. method according to claim 6 is characterized in that, described storage control module is independent equipment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100836926A CN100389383C (en) | 2006-06-02 | 2006-06-02 | Method for solving writing cavity in magnetic disk redundancy array |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100836926A CN100389383C (en) | 2006-06-02 | 2006-06-02 | Method for solving writing cavity in magnetic disk redundancy array |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1862478A CN1862478A (en) | 2006-11-15 |
| CN100389383C true CN100389383C (en) | 2008-05-21 |
Family
ID=37389922
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100836926A Active CN100389383C (en) | 2006-06-02 | 2006-06-02 | Method for solving writing cavity in magnetic disk redundancy array |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100389383C (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5029513B2 (en) * | 2008-06-30 | 2012-09-19 | ソニー株式会社 | Information processing apparatus, information processing apparatus control method, and program |
| CN101488077B (en) * | 2009-02-24 | 2011-08-24 | 浪潮电子信息产业股份有限公司 | Redundancy magnetic disk control server system |
| CN102541471B (en) * | 2011-12-28 | 2014-12-10 | 创新科软件技术（深圳）有限公司 | Storage system with multiple controllers |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5604902A (en) * | 1995-02-16 | 1997-02-18 | Hewlett-Packard Company | Hole plugging garbage collection for a data storage system |
| US5819109A (en) * | 1992-12-07 | 1998-10-06 | Digital Equipment Corporation | System for storing pending parity update log entries, calculating new parity, updating the parity block, and removing each entry from the log when update is complete |
| US5860091A (en) * | 1996-06-28 | 1999-01-12 | Symbios, Inc. | Method and apparatus for efficient management of non-aligned I/O write request in high bandwidth raid applications |
| US6523087B2 (en) * | 2001-03-06 | 2003-02-18 | Chaparral Network Storage, Inc. | Utilizing parity caching and parity logging while closing the RAID5 write hole |
| US6629211B2 (en) * | 2001-04-20 | 2003-09-30 | International Business Machines Corporation | Method and system for improving raid controller performance through adaptive write back/write through caching |
| US20050066124A1 (en) * | 2003-09-24 | 2005-03-24 | Horn Robert L. | Method of RAID 5 write hole prevention |
| CN1704910A (en) * | 2004-06-03 | 2005-12-07 | 华为技术有限公司 | Write handling method for disc array arrangement |
-
2006
- 2006-06-02 CN CNB2006100836926A patent/CN100389383C/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5819109A (en) * | 1992-12-07 | 1998-10-06 | Digital Equipment Corporation | System for storing pending parity update log entries, calculating new parity, updating the parity block, and removing each entry from the log when update is complete |
| US5604902A (en) * | 1995-02-16 | 1997-02-18 | Hewlett-Packard Company | Hole plugging garbage collection for a data storage system |
| US5860091A (en) * | 1996-06-28 | 1999-01-12 | Symbios, Inc. | Method and apparatus for efficient management of non-aligned I/O write request in high bandwidth raid applications |
| US6523087B2 (en) * | 2001-03-06 | 2003-02-18 | Chaparral Network Storage, Inc. | Utilizing parity caching and parity logging while closing the RAID5 write hole |
| US6629211B2 (en) * | 2001-04-20 | 2003-09-30 | International Business Machines Corporation | Method and system for improving raid controller performance through adaptive write back/write through caching |
| US20050066124A1 (en) * | 2003-09-24 | 2005-03-24 | Horn Robert L. | Method of RAID 5 write hole prevention |
| CN1704910A (en) * | 2004-06-03 | 2005-12-07 | 华为技术有限公司 | Write handling method for disc array arrangement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1862478A (en) | 2006-11-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7809979B2 (en) | Storage control apparatus and method | |
| JP3742494B2 (en) | Mass storage device | |
| CN103136074B (en) | Data storage method and data storage system of multiple disk array systems | |
| CN100407121C (en) | Information processing system, primary storage device, and computer readable recording medium recorded thereon logical volume restoring program | |
| US20090271659A1 (en) | Raid rebuild using file system and block list | |
| CN101351760A (en) | Method and system for accessing auxiliary data in power-efficient high-capacity scalable storage | |
| CN102207895B (en) | Data reconstruction method and device of redundant array of independent disk (RAID) | |
| CN100349127C (en) | Embedded system data back-up device and method | |
| CN101567211A (en) | Method for improving usability of disk and disk array controller | |
| CN104035830A (en) | Method and device for recovering data | |
| CN1746854A (en) | The device, method and the program that are used for control store | |
| CN101436149A (en) | Method for rebuilding data of magnetic disk array | |
| US20070101188A1 (en) | Method for establishing stable storage mechanism | |
| CN102177496A (en) | System and method for transferring data between different RAID data storage types for current data and replay data | |
| CN103049400A (en) | Disk reconfiguration method and disk reconfiguration device | |
| CN103049220A (en) | Storage control method, storage control device and solid-state storage system | |
| CN103019893A (en) | Multi-disk fault-tolerant two-dimensional hybrid disk RAID4 system architecture and read-write method thereof | |
| US7260739B2 (en) | Method, apparatus and program storage device for allowing continuous availability of data during volume set failures in a mirrored environment | |
| CA2694819C (en) | Redundancy in array storage system based on minimum distances of donor/recipient arrays | |
| CN100389383C (en) | Method for solving writing cavity in magnetic disk redundancy array | |
| JP2003316525A (en) | Disk array controller | |
| JP4698710B2 (en) | Storage system and power saving method thereof | |
| JPH1069357A (en) | Storage network control method | |
| CN101794246A (en) | Backup storage system, backup system, data backup method and recovery method | |
| CN102541472A (en) | Method and device for reconstructing RAID (Redundant Array of Independent Disks) |
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 | ||
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: 310052 Binjiang District Changhe Road, Zhejiang, China, No. 466, No. Patentee after: Xinhua three Technology Co., Ltd. Address before: 310053 Hangzhou hi tech Industrial Development Zone, Zhejiang province science and Technology Industrial Park, No. 310 and No. six road, HUAWEI, Hangzhou production base Patentee before: Huasan Communication Technology Co., Ltd. |