CN107220145A - A kind of method of flash memory database data recovery - Google Patents
A kind of method of flash memory database data recovery Download PDFInfo
- Publication number
- CN107220145A CN107220145A CN201710358273.7A CN201710358273A CN107220145A CN 107220145 A CN107220145 A CN 107220145A CN 201710358273 A CN201710358273 A CN 201710358273A CN 107220145 A CN107220145 A CN 107220145A
- Authority
- CN
- China
- Prior art keywords
- data
- log
- affairs
- log recording
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1471—Saving, restoring, recovering or retrying involving logging of persistent data for recovery
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0602—Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
- G06F3/0608—Saving storage space on storage systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0628—Interfaces specially adapted for storage systems making use of a particular technique
- G06F3/0638—Organizing or formatting or addressing of data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/06—Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
- G06F3/0601—Interfaces specially adapted for storage systems
- G06F3/0668—Interfaces specially adapted for storage systems adopting a particular infrastructure
- G06F3/0671—In-line storage system
- G06F3/0673—Single storage device
- G06F3/0679—Non-volatile semiconductor memory device, e.g. flash memory, one time programmable memory [OTP]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Human Computer Interaction (AREA)
- Quality & Reliability (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of method of flash memory database data recovery, wherein, including：Scanning log file, confirms whether have still effective affairs to submit log recording or transaction rollback daily record in journal file；The state current according to journal file is respectively processed：If submitting log recording in the presence of effective affairs, show that all changes for being related to the affairs are not yet completely written in data file, need to complete the rewriting of data file according to log recording；If being recorded in the presence of effective rewind journal, all log recordings are searched according to affairs ID, and carry out rolling back action；If submitting log recording in the absence of effective affairs, also in the absence of effective rewind journal, but there is still effective log recording, show that the affairs are not yet completed, check for partial log record written data file, and part situation about not writing, as existed, then rolling back action is carried out.
Description
Technical field
The present invention relates to a kind of data storage technology of flash memory, more particularly to a kind of side of flash memory database data recovery
Method.
Background technology
Recovery technology based on daily record is widely used in traditional data in magnetic disk base management system.In different agreements
Under, design, the management of daily record/data buffer zone, checkpointing mechanism, log and the process of recovery of log recording are not
Equally.By taking undo daily records as an example, when affairs T needs database element X (original value is v) value being changed to v1, undo days
Will will be formed such as<T,X,v1>Log recording and store on disk.When needing to carry out rolling back action to affairs T, then shape
Cheng Ru<T,X,v>Log recording, element X value is reset into v.
Flash memory (Flash Memory) is a kind of long-life nonvolatile memory.Compared with disk, flash memory, which has, to be read
Writing rate is fast, antidetonation, power saving, small volume the advantages of.Undo journal recovery processes are reappeared in a flash memory：
(1) tri- elements of A, B, C are included shown in tables of data original state such as table 1 (a), in table, initial value be respectively va,
Vb, vc, now flag bit be set to 1, represent that three element values are effective；
(2) when the A values in tables of data are revised as in va ', one new record of insertion, such as table 1 (b) most by affairs T from va
A line afterwards, now the flag bit of the first row A values be set to 0, represent that A value is changed, initial value is no longer valid；
(3) if affairs T needs rollback, then a new record is write, A values is reverted into va, shown in such as table 1 (c).
Table 1 (a)
| Value | Flag bit | |
| A | va | 1 |
| B | vb | 1 |
| C | vc | 1 |
| … | … | … |
Table 1 (b)
| Value | Flag bit | |
| A | va | 0 |
| B | vb | 1 |
| C | vc | 1 |
| A | va’ | 1 |
| … | … | … |
Table 1 (c)
| Value | Flag bit | |
| A | va | 0 |
| B | vb | 1 |
| C | vc | 1 |
| A | va’ | 0 |
| A | va | 1 |
| … | … | … |
As can be seen that the last item record records identical with first from table 1 (c), that is to say, that last column
It is redundancy in this case.Therefore generally there is the old version of substantial amounts of data in flash memory, and traditional data recovery
Method constantly writes the data record existed again, so not only wasting space but also has lost time.
The content of the invention
It is an object of the invention to provide a kind of method of flash memory database data recovery, for solving above-mentioned prior art
The problem of.
A kind of method of flash memory database data recovery of the present invention, wherein, including：Scanning log file, confirms daily record text
Whether there are still effective affairs to submit log recording or transaction rollback daily record in part；The state difference current according to journal file
It is handled as follows：If submitting log recording in the presence of effective affairs, show that all changes for being related to the affairs are not yet write completely
Enter into data file, need to complete the rewriting of data file according to log recording；If being recorded in the presence of effective rewind journal,
All log recordings are searched according to affairs ID, and carry out rolling back action；If submitting log recording in the absence of effective affairs, also not
In the presence of effective rewind journal, but there is still effective log recording, show that the affairs are not yet completed, check for portion
Point log recording written data file, and the situation that part does not write, such as exist, then carry out rolling back action.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, rolling back action includes：First
The still effective log recording of the corresponding affairs of all about affairs ID, Ran Houyi are found out from journal file according to affairs ID
Secondary judgement logging operations type, according to different operating type, completes following operation：If insertion operation, then delete current
The data element of insertion；If deletion action, then deleted data element is reinserted；If updating operation, then by data
The value of element returns to original state.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, in addition to：Chain type daily record,
Including a chained list and a tables of data, the daily record note for the different pieces of information operation that each affairs of sequential storage are produced in chained list
Include and included in the different editions data of each data member in a log recording, the log recording of chained list in record, tables of data
Specify the address of tables of data；Log recording in tables of data includes affairs ID, data member, preceding member and data address table four
Field, wherein affairs ID are No. ID of affairs belonging to current log is recorded；Data member is that current data operates the data pointed to
The title of member；Preceding member is to produce the preceding member of change in Current transaction；Data address table is all version numbers of data member
According to address, each edition data is attached with chained list in chronological order, when affairs are submitted or after rollback, the data address word
Section is set to sky.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, log recording builds daily record
The method of file includes：When a new affairs unlatching, and it is the data member of first renewal when producing first data variation
Insert a log recording；When producing the change of other data members in affairs, then a new log recording is inserted, before it
Member is set to the data record of a change；For insertion and deletion action, come in data address table using multiple indications
Distinguish the data of different editions；If affairs are submitted or rollback, insert one and submit or rewind journal record, preceding member sets
It is set to the member updated before；All data members for producing change and its specific data manipulation in database of record, and unite
One is stored in journal file.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, in addition to：In journal file
Middle write-in checkpoint, it is checkpoint to show current location；Daily record is scanned one by one in units of log blocks, when running into inspection
During point, the validity of all log recordings before checkpoint is confirmed, still effective log recording is write into new free block；
After the scanning to log blocks is completed, erasing operation is carried out to it, and be transferred to next log blocks not scanned；Work as the free time
Block is full, then applies for new free block, and new effective record is write to new free block.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, in addition to：By current log
The method that effective record in block dumps to new log blocks includes：Unloading flag bit is set to 1, and creates a new daily record
Block；First log recording is obtained, whether judge current unloading flag bit is 1, if not 1, then wipe old log blocks and knot
Beam；If 1, then judge whether current log record effective, if still effectively if by current log record copies to new daily record
Block, and obtain next log recording and recorded as current log；Directly obtained if invalid next log recording as work as
Preceding log recording；If new log blocks are full, next free block is obtained as new log blocks；If the current log obtained
Sky is recorded as, then unloading flag bit is set to 0, shows that unloading operation has terminated.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, the time interval of checkpoint
Dynamically adjusted according to application characteristic.
According to an embodiment of the method for the flash memory database data recovery of the present invention, wherein, the data of chain type daily record are deposited
Method for storing：Log recording is first write into internal memory or caching, log recording is write into journal file, data variation is write into data text
Corresponding log recording in part, ineffective treatment journal file.
The present invention proposes that one kind can either save memory space based on flash memory database, and data recovering efficiency can be lifted again
Data reconstruction method.Chain type log system carries out journalizing using old version information present in flash memory database, adopts
Sequential organization is replaced with list structure to store log recording to improve data recovering efficiency.By periodic test point strategy come
Remove the log recording of failure, save flash memory space, while effectively reducing the length of journal file.
Brief description of the drawings
Fig. 1 show chain type log system configuration diagram；
Fig. 2 show log record figure；
Fig. 3 show the step flow chart of unloading；
Embodiment
To make the purpose of the present invention, content and advantage clearer, with reference to the accompanying drawings and examples, to the present invention's
Embodiment is described in further detail.
Fig. 1 show chain type log system configuration diagram, as shown in figure 1, the present invention devises a kind of new day aspiration
System, is named as chain type daily record.In chain type log system, log manager 1 is responsible for the generation and unified management of log recording；Day
Will file 2 is the set of log recording, is made up of a rule log recording；RMAN 3 is responsible for system crash or affairs are returned
The recovery management of data during rolling；DBMS4 (Database management system) is data base management system, is responsible for affairs
Management, log management, the management of data file 5 etc.；Data file 5 is the physical store place of data record.
Fig. 2 show log record figure, as shown in Fig. 2 chain type daily record stores log recording using list structure, is
System safeguards a chained list and a tables of data.The day for the different pieces of information operation that each affairs of sequential storage are produced wherein in chained list
The different editions data of each data member in a log recording are included in will, tables of data, specified number is included in chained list daily record
According to the address of table.
As shown in Fig. 2 chain type daily record is used to realize to the fast quick-recovery of flash memory database, the characteristics of being mainly based upon flash memory sets
A kind of new log record has been counted, can be quick right by operation information of the log recording quick obtaining user to database
Flash memory database is recovered.The log recording of chain type daily record includes affairs ID, data member, preceding member and data address table four
Individual field, wherein " affairs ID " refers to No. ID of the affiliated affairs of current log；" data member " refers to the number that current data operation is pointed to
According to the title of member；" preceding member " refers to the preceding member that change is produced in Current transaction, if the data member of current operation is the thing
Be engaged in first member, then preceding member fields are set to " Begin "；" data address table " refers to the ground of all edition datas of data member
Location, each edition data is attached with chained list in chronological order, and after affairs submission or rollback, the data address field is set
For " NULL ".
As shown in Figure 1 and Figure 2, journal file 2 is made up of a rule log recording, and table 2 is literary for the daily record of chain type daily record
Part structural table.Log recording (abbreviation daily record) is to record user to the operation information of database, can be learnt by log recording
User is to the operation of database and the change of data, when system crash, and database can be recovered according to log recording；Daily record
File refers to the file for depositing log recording, is made up of multiple log blocks；Journal file is stored in storage by block,
Each block is referred to as log blocks, i.e. journal file and is divided into multiple log blocks, and each log blocks store multiple log recordings；
Chain type daily record is a kind of new log system that the present invention is designed, for realizing to the fast quick-recovery of flash memory database,
The characteristics of being mainly based upon flash memory devises a kind of new log record, can pass through log recording quick obtaining user's logarithm
According to the operation information in storehouse, quickly flash memory database is recovered
Table 2
| Sequence number | Affairs ID | Data member | Preceding member | Data address table |
| 1 | T1 | X | Begin | P(X1)->P(X2)->P(X3) |
| 2 | T1 | Y | P(X) | P(Y1)->P(Y2) |
| 3 | T2 | A | Begin | P(A1)->P(A2) |
| 4 | T1 | Z | P(Y) | P(Z1)->P(Z2)->NULL |
| 5 | T2 | B | P(A) | NULL->P(B1) |
| 6 | T1 | Commit | P(Z) | NULL |
| 7 | T2 | Rollback | P(B) | NULL |
| 8 | … | … | … | … |
As shown in Figure 1 and Figure 2, log recording is unified is responsible for by log manager 1.Chain type log recording builds day
The algorithm and flow of will file 2 are as follows：
First, when a new affairs unlatching, and (insertion, modification or deletion action cause when producing first data variation
Database produces change), log manager 1 is that the data member of first renewal inserts a log recording, and by before it into
Member is set to " Begin ".As shown in the first row of table 2, affairs T1 is opened, and carries out data manipulation, sequential update to data member X
Its numerical value (from X1 to X2, then to X3), so first log recording is entered as：Affairs ID is set to T1, and data member is set to X,
Preceding member is set to " Begin ", data address table be set to P (X1)->P(X2)->P (X3) (is herein X1, X2, X3 place tables of data
In address).
Second, when producing the change of other data members in affairs, then log manager 1 inserts a new daily record note
Record, its preceding member is set to the data record of a upper change.Data member Y value is changed such as in affairs T1, by it from Y1
Y2 is changed to, then log recording is entered as：Affairs ID is still set to T1, and data member is set to Y, and preceding member is set to P (X), data
Location table be set to P (Y1)->P (Y2), as shown in the row of table 2 second.
3rd, for insertion and deletion action, different editions are distinguished using multiple indications in data address table
Data.Insertion operation is carried out such as in affairs, then first address in its data address table is set to " NULL ", entered in affairs
Row deletion action, then be set to " NULL " by last address in data address table, and fourth line represents to delete in T1 in such as table 2
Data member Z, fifth line represents to insert data member B in T2.
4th, if affairs are submitted or rollback, log manager 1 is inserted one and submitted or rewind journal record, preceding
Member is set to the member updated before, and address table is set to " NULL ".T1 affairs are submitted (Commit) in such as table 2, T2 affairs
(Rollback) is rolled back, so being respectively written into the six, the seven row log recordings.
By above-mentioned daily record algorithm, log manager 1 can be built in log recording complete one by one, database of record
All data members for producing change and its specific data manipulation, and be uniformly stored in journal file 2, form journal file
2。
As shown in figure 1, to avoid producing long journal file 2, causing the reduction of system effectiveness, present invention uses inspection
Strategy is made an inventory of, to ensure only to transmit still effective log recording.The method to set up of checkpoint is write in journal file 2
Checkpointing, it is checkpoint to show current location.
In chain type daily record using checkpoint strategy after, operating process approximately as：
First, daily record is scanned one by one in units of log blocks.When running into checkpoint, confirm before checkpoint
The validity of all log recordings, still effective log recording is write into new free block, and (idle log blocks, i.e., do not write
Enter the log blocks of log recording).
Second, after the scanning to log blocks is completed, erasing operation can be carried out to it, and be transferred to and next do not sweep
The log blocks retouched.
3rd, when free block is full, then apply for new free block, by new effective record (effective log recording.Daily record
Record has a flag bit, shows whether current log record is effective.) the new free block of write-in.
A such as log blocks can deposit 100 log recordings, wherein having 10 still effectively, then dump to 10 new
Log blocks in, old piece can wipe.
Fig. 3 show the step flow chart of unloading, as shown in Figure 1 and Figure 3, includes the step of unloading：
(1) " unloading flag bit " is set to 1, and creates a new log blocks；Wherein, unloading flag bit refers to judge to work as
The mark whether preceding daily record still effectively records soon, if 1, then show to exist effectively record, if 0, then show to be not present
Effectively record, can wipe the log blocks.
(2) first log recording is obtained, judges whether current " unloading flag bit " is 1.If not 1, then wipe old
Log blocks simultaneously terminate；If 1, then judge whether current log record effective, if still effectively if by current log record copies
To new log blocks, and obtain next log recording and recorded as current log；Next daily record is directly obtained if invalid
Record is recorded as current log；
(3) if new log blocks are full, next free block is obtained as new log blocks；
(4) if the current log obtained is recorded as sky, " unloading flag bit " is set to 0, shows that unloading operation has terminated.
Long checkpoint interval can cause to produce big journal file 2, and small checkpoint interval can shorten the life of flash memory
Life cycle, therefore, the time interval of checkpoint of the invention can dynamically adjust according to application characteristic, while using periodic test
The strategy of point.
In chain type log system, under normal circumstances the process of data storage be related to data base management system (DBMS4) and
Log manager 1, wherein DBMS4 are responsible for the generation of the normal transaction operation of database and control data file 5, log manager
1 is responsible for the management of log recording and the generation of journal file 2.The normal data storage of chain type daily record substantially flow is as follows：
First, DBMS4 carry out normal transaction operation, and the database change to generation (is related to data to insert, delete and repair
Change operation) issued transaction is done, and log manager 1 is transferred to, wherein, issued transaction is when a series of changes carried out to source data
During as an entirety to preserve or cancel, it need to be handled using affairs.Issued transaction include open affairs, submit affairs and
Rollback affairs, update all operations when submitting affairs, cancel all operations during rollback affairs.Only insert, delete and change behaviour
Affairs are needed to use, because these operations produce renewal to the data in database.Need not if only read operation
Use affairs.
Second, log manager 1 requires one rule log recording of generation according to chain type journal format, and constitutes journal file
2, complete the complete documentation of transaction operation.
3rd, affairs are normally submitted, and now data record is write data file 5 (secondary storage media) by DBMS4.
4th, the log recording involved by the affairs having been filed on is done ineffective treatment processing by log manager 1, and whole affairs are carried
Friendship process is completed.
From above-mentioned data storage procedure, the substantially flow of data storage is：Log recording is first write into internal memory or slow
Deposit->Log recording write-in journal file 2->Data variation write-in data file 5->Corresponding daily record note in ineffective treatment journal file 2
Record.And data recovery must be by means of log recording, i.e. collapse occurs before log recording write-in journal file 2, then need not
Recover, because currently without log recording；If collapse occurs after log recording ineffective treatment, without recovery, because
Current data written data file 5.In summary, data recovery is only occurred in after the generation of journal file 2, data file 5
Before completion.
With reference to Fig. 1, the data recovery procedure of the method for flash memory database data recovery of the invention includes：
(1) RMAN 3 confirms whether have still effective affairs to carry in journal file 2 by scanning log file 2
Hand over daily record (commit records) or transaction rollback daily record (rollback records).The state current according to journal file 2 is entered respectively
The following processing of row.
(2) if submitting log recording (commit records) in the presence of effective affairs, show to be related to all changes of the affairs
Not yet it is completely written in data file 5, now needs to complete the rewriting of data file 5 according to log recording.It is i.e. submitted
Whether affairs, feed back in data file 5 completely no matter updating, as long as reforming the log recording for once relating to the affairs,
Ensure that the state when data record in data file 5 is submitted for normal transaction.As in table 2, affairs T1 has been filed on, but not
Know whether be completely written in data file 5.If now collapsing, RMAN 3 is first depending on affairs ID from journal file 2
In find out all about T1 effective record the first, second of table 2 (, fourth line), then will be related to successively in log recording
Data variation re-writes data file 5, such as the first row of table 2 be by data member X value be revised as successively P (X1)->P(X2)-
>P (X3), the second row be by data member Y value be revised as successively P (Y1)->P (Y2), fourth line is first by data element Z
Value be revised as P (Z1)->P (Z2), then deletes data element Z, data recovery procedure only need to be by data element in affairs T1 most
X values (can be revised as X3, Y value is revised as Y2, deletes Z) by whole state write-in data file 5, and middle change in process can
To omit, the efficiency of journal recovery is so substantially increased.
(3) if recording (rollback daily records) in the presence of effective rewind journal, also need to search all days according to affairs ID
Will is recorded, and carries out rolling back action.Rollback flow is to be first depending on affairs ID to find out all about affairs from journal file 2
Still effective log recording, then judges logging operations type successively, according to different operating type, completes following behaviour
Make：If insertion operation (i.e. first data address of log recording is " NULL "), then delete the data element being currently inserted into；If
It is deletion action (last data address is " NULL "), then reinserts deleted data element；If updating operation
(in addition to above-mentioned two situations), then return to original state (value of i.e. first data address) by the value of data element.Such as
Affairs T2 has rewind journal in table 2, then finds still effective log recording (the third line of table 2, fifth line) according to affairs ID, so
After carry out rollback：Data element A value is set to initial value A1, data element B is deleted.
(4) if submitting daily record in the absence of effective affairs, also in the absence of effective rewind journal, but exist still effective
Log recording, shows that the affairs are not yet completed, it is understood that there may be partial log record written data file 5, and part does not write
Situation, now need display one transaction rollback of insertion.Then rolling back action is carried out according to above-mentioned 3rd step.It is false in such as table 2
Be located at fifth line data record fill in finish after collapse, now affairs T1 does not write " commit " log recording, now needs rollback
Operations all T1, i.e., one rollback record of insertion first in journal file 2, as shown in table 3.
Table 3：
| T1 | Rollback | P(Z) | NULL |
Then data element X value is reverted into P (X1), data element Y value returns to P (Y1), inserts data element
Z, and its value is set to P (Z1).
The method of the flash memory database data recovery of the present invention, devises a kind of new log record, by change
The data address of data member and its all different editions is saved as in same log recording, is easy in log rollback or submission
When according to affairs ID be quickly found out valid data record, increase data recovery efficiency.
The present invention follows WAL (Write Ahead Logging) rules to determine when refresh daily record simultaneously, i.e., local number
When being modified according to storehouse, it is necessary to have corresponding log recording, on the contrary it is not necessarily so.When system conflicts, renewal will only by
Journal file 2 is stored, without being delivered in database.Therefore when performing recovery operation, above-mentioned mechanism is able to ensure that data
The uniformity and integrality in storehouse.
The method of flash memory database data recovery of the present invention, goes through present in the flash memory database using chain type log system
History version information carries out journalizing, uses list structure to replace sequential organization to store log recording to improve data recovery effect
Rate.The log recording of failure is removed by periodic test point strategy, flash memory space is saved, while effectively reducing daily record text
The length of part.
Described above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, some improvement and deformation can also be made, these improve and deformed
Also it should be regarded as protection scope of the present invention.
Claims (8)
1. a kind of method of flash memory database data recovery, it is characterised in that including：
Scanning log file, confirms whether have still effective affairs to submit log recording or transaction rollback day in journal file
Will；
The state current according to journal file is handled as follows respectively：
If submitting log recording in the presence of effective affairs, show that all changes for being related to the affairs are not yet completely written to data text
In part, need to complete the rewriting of data file according to log recording；
If being recorded in the presence of effective rewind journal, all log recordings are searched according to affairs ID, and carry out rolling back action；
If submitting log recording in the absence of effective affairs, also in the absence of effective rewind journal, but there is still effective day
Will is recorded, and shows that the affairs are not yet completed, and checks for partial log record written data file, and part does not write
Situation, such as exist, then carry out rolling back action.
2. the method for flash memory database data recovery as claimed in claim 1, it is characterised in that rolling back action includes：First
The still effective log recording of the corresponding affairs of all about affairs ID, Ran Houyi are found out from journal file according to affairs ID
Secondary judgement logging operations type, according to different operating type, completes following operation：If insertion operation, then delete current
The data element of insertion；If deletion action, then deleted data element is reinserted；If updating operation, then by data
The value of element returns to original state.
3. the method for flash memory database data recovery as claimed in claim 1, it is characterised in that also include：Chain type daily record, bag
The log recording for the different pieces of information operation that each affairs of sequential storage are produced in a chained list and a tables of data, chained list is included,
Included in tables of data in the different editions data of each data member in a log recording, the log recording of chained list comprising specified
The address of tables of data；Log recording in tables of data includes affairs ID, data member, four words of preceding member and data address table
Section, wherein affairs ID is No. ID of affairs belonging to current log is recorded；Data member be current data operate point to data into
The title of member；Preceding member is to produce the preceding member of change in Current transaction；Data address table is all edition datas of data member
Address, each edition data is attached with chained list in chronological order, when affairs are submitted or after rollback, the data address field
It is set to sky.
4. the method for flash memory database data recovery as claimed in claim 3, it is characterised in that log recording builds daily record text
The method of part includes：
It is that the data member of first renewal inserts a day when a new affairs unlatching, and when producing first data variation
Will is recorded；
When producing the change of other data members in affairs, then a new log recording is inserted, its preceding member is set to upper one
The data record of change；
For insertion and deletion action, the data of different editions are distinguished using multiple indications in data address table；
If affairs are submitted or rollback, insert one and submit or rewind journal record, preceding member is set to what is updated before
Member；
All data members for producing change and its specific data manipulation in database of record, and uniformly it is stored in journal file
In.
5. the method for flash memory database data recovery as claimed in claim 4, it is characterised in that also include：
Checkpoint is write in journal file, it is checkpoint to show current location；
Log recording is scanned one by one in units of log blocks, when running into checkpoint, confirms own before checkpoint
The validity of log recording, new free block is write by still effective log recording；
After the scanning to log blocks is completed, erasing operation is carried out to it, and be transferred to next log blocks not scanned；
When free block is full, then apply for new free block, new effective record is write to new free block.
6. the method for flash memory database data recovery as claimed in claim 5, it is characterised in that also include：By current log
The method that effective record in block dumps to new log blocks includes：
Unloading flag bit is set to 1, and creates a new log blocks；
First log recording is obtained, whether judge current unloading flag bit is 1, if not 1, then wipe old log blocks and knot
Beam；If 1, then judge whether current log record effective, if still effectively if by current log record copies to new daily record
Block, and obtain next log recording and recorded as current log；Directly obtained if invalid next log recording as work as
Preceding log recording；
If new log blocks are full, next free block is obtained as new log blocks；
If the current log obtained is recorded as sky, unloading flag bit is set to 0, shows that unloading operation has terminated.
7. the method for flash memory database data recovery as claimed in claim 6, it is characterised in that the time interval root of checkpoint
Dynamically adjusted according to application characteristic.
8. the method for flash memory database data recovery as claimed in claim 3, it is characterised in that the data storage of chain type daily record
Method：
Log recording is first write into internal memory or caching, log recording is write into journal file, data variation is write into data file,
Corresponding log recording in ineffective treatment journal file.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710358273.7A CN107220145B (en) | 2017-05-19 | 2017-05-19 | Method for recovering data of flash memory database |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710358273.7A CN107220145B (en) | 2017-05-19 | 2017-05-19 | Method for recovering data of flash memory database |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107220145A true CN107220145A (en) | 2017-09-29 |
| CN107220145B CN107220145B (en) | 2021-06-01 |
Family
ID=59945356
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710358273.7A Active CN107220145B (en) | 2017-05-19 | 2017-05-19 | Method for recovering data of flash memory database |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107220145B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109101368A (en) * | 2018-08-20 | 2018-12-28 | 郑州云海信息技术有限公司 | A kind of data processing method and device |
| CN109885422A (en) * | 2018-12-29 | 2019-06-14 | 贝壳技术有限公司 | The data rewind method and apparatus of configuration platform in a kind of house prosperity transaction system |
| CN109960601A (en) * | 2017-12-26 | 2019-07-02 | 航天信息股份有限公司 | A kind of restoration methods of business datum, device and electronic equipment |
| CN110096389A (en) * | 2019-04-30 | 2019-08-06 | 上海达梦数据库有限公司 | A kind of starting method, apparatus, equipment and the storage medium of database |
| CN111026772A (en) * | 2019-11-26 | 2020-04-17 | 中科驭数（北京）科技有限公司 | Method and device for ensuring permanence of database modification |
| CN111240891A (en) * | 2018-11-29 | 2020-06-05 | 中国移动通信集团内蒙古有限公司 | Data recovery method and device based on data consistency among multiple tables of database |
| WO2020143240A1 (en) * | 2019-01-10 | 2020-07-16 | 江苏华存电子科技有限公司 | Method for quickly recovering data in flash memory database |
| CN111654453A (en) * | 2020-04-23 | 2020-09-11 | 平安国际智慧城市科技股份有限公司 | Form data offline caching method and device, terminal and storage medium |
| CN111694893A (en) * | 2020-04-23 | 2020-09-22 | 武汉达梦数据库有限公司 | Partial rollback analysis method based on log analysis and data synchronization system |
| CN112612648A (en) * | 2020-12-29 | 2021-04-06 | 厦门市美亚柏科信息股份有限公司 | SQL Server database recovery method, terminal device and storage medium |
| CN114064664A (en) * | 2022-01-17 | 2022-02-18 | 北京奥星贝斯科技有限公司 | Method and device for inquiring transaction modification content in database |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060047925A1 (en) * | 2004-08-24 | 2006-03-02 | Robert Perry | Recovering from storage transaction failures using checkpoints |
| CN101286127A (en) * | 2008-05-08 | 2008-10-15 | 华中科技大学 | Multi-fork diary memory continuous data protecting and restoration method |
| CN101470629A (en) * | 2007-12-30 | 2009-07-01 | 英特尔公司 | Mechanism for strong atomicity in a transactional memory system |
| CN102156720A (en) * | 2011-03-28 | 2011-08-17 | 中国人民解放军国防科学技术大学 | Method, device and system for restoring data |
| US8037112B2 (en) * | 2007-04-23 | 2011-10-11 | Microsoft Corporation | Efficient access of flash databases |
| CN102436490A (en) * | 2010-10-28 | 2012-05-02 | 微软公司 | Versatile in-memory database recovery |
| US20140095765A1 (en) * | 2012-09-28 | 2014-04-03 | Apple Inc. | Flash translation layer (ftl) database journaling schemes |
| CN105786955A (en) * | 2015-01-08 | 2016-07-20 | 国际商业机器公司 | Data replication in a database management system |
| CN105930500A (en) * | 2016-05-06 | 2016-09-07 | 华为技术有限公司 | Transaction recovery method in database system, and database management system |
| US20160292250A1 (en) * | 2013-12-12 | 2016-10-06 | Huawei Technologies Co., Ltd. | Data replication method and storage system |
-
2017
- 2017-05-19 CN CN201710358273.7A patent/CN107220145B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060047925A1 (en) * | 2004-08-24 | 2006-03-02 | Robert Perry | Recovering from storage transaction failures using checkpoints |
| US8037112B2 (en) * | 2007-04-23 | 2011-10-11 | Microsoft Corporation | Efficient access of flash databases |
| CN101470629A (en) * | 2007-12-30 | 2009-07-01 | 英特尔公司 | Mechanism for strong atomicity in a transactional memory system |
| CN101286127A (en) * | 2008-05-08 | 2008-10-15 | 华中科技大学 | Multi-fork diary memory continuous data protecting and restoration method |
| CN102436490A (en) * | 2010-10-28 | 2012-05-02 | 微软公司 | Versatile in-memory database recovery |
| CN102156720A (en) * | 2011-03-28 | 2011-08-17 | 中国人民解放军国防科学技术大学 | Method, device and system for restoring data |
| US20140095765A1 (en) * | 2012-09-28 | 2014-04-03 | Apple Inc. | Flash translation layer (ftl) database journaling schemes |
| US20160292250A1 (en) * | 2013-12-12 | 2016-10-06 | Huawei Technologies Co., Ltd. | Data replication method and storage system |
| CN105786955A (en) * | 2015-01-08 | 2016-07-20 | 国际商业机器公司 | Data replication in a database management system |
| CN105930500A (en) * | 2016-05-06 | 2016-09-07 | 华为技术有限公司 | Transaction recovery method in database system, and database management system |
Non-Patent Citations (4)
| Title |
|---|
| SILBERSCHATZ,A.等著，杨冬青等译: "《数据库系统概念（原书第6版）》", 31 March 2012, 机械工业出版社 * |
| 卢泽萍，孟小峰，周大: "HV-Recovery :一种闪存数据库的高效恢复方法", 《计算机学报》 * |
| 林春: "分布式内存数据库的恢复", 《航空计算技术》 * |
| 王曼丽等: "一种高效的闪存数据库故障恢复方法MMR", 《计算机技术与发展》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109960601A (en) * | 2017-12-26 | 2019-07-02 | 航天信息股份有限公司 | A kind of restoration methods of business datum, device and electronic equipment |
| CN109960601B (en) * | 2017-12-26 | 2023-03-24 | 航天信息股份有限公司 | Service data recovery method and device and electronic equipment |
| CN109101368B (en) * | 2018-08-20 | 2022-04-22 | 郑州云海信息技术有限公司 | Data processing method and device |
| CN109101368A (en) * | 2018-08-20 | 2018-12-28 | 郑州云海信息技术有限公司 | A kind of data processing method and device |
| CN111240891A (en) * | 2018-11-29 | 2020-06-05 | 中国移动通信集团内蒙古有限公司 | Data recovery method and device based on data consistency among multiple tables of database |
| CN109885422A (en) * | 2018-12-29 | 2019-06-14 | 贝壳技术有限公司 | The data rewind method and apparatus of configuration platform in a kind of house prosperity transaction system |
| CN109885422B (en) * | 2018-12-29 | 2024-01-12 | 贝壳技术有限公司 | Data rollback method and device for configuration platform in real estate transaction system |
| WO2020143240A1 (en) * | 2019-01-10 | 2020-07-16 | 江苏华存电子科技有限公司 | Method for quickly recovering data in flash memory database |
| CN110096389A (en) * | 2019-04-30 | 2019-08-06 | 上海达梦数据库有限公司 | A kind of starting method, apparatus, equipment and the storage medium of database |
| CN111026772A (en) * | 2019-11-26 | 2020-04-17 | 中科驭数（北京）科技有限公司 | Method and device for ensuring permanence of database modification |
| CN111026772B (en) * | 2019-11-26 | 2020-11-27 | 中科驭数（北京）科技有限公司 | Method and device for ensuring permanence of database modification |
| CN111694893A (en) * | 2020-04-23 | 2020-09-22 | 武汉达梦数据库有限公司 | Partial rollback analysis method based on log analysis and data synchronization system |
| CN111654453B (en) * | 2020-04-23 | 2023-04-18 | 平安国际智慧城市科技股份有限公司 | Form data offline caching method and device, terminal and storage medium |
| CN111694893B (en) * | 2020-04-23 | 2023-04-25 | 武汉达梦数据库股份有限公司 | Partial rollback analysis method and data synchronization system based on log analysis |
| CN111654453A (en) * | 2020-04-23 | 2020-09-11 | 平安国际智慧城市科技股份有限公司 | Form data offline caching method and device, terminal and storage medium |
| CN112612648A (en) * | 2020-12-29 | 2021-04-06 | 厦门市美亚柏科信息股份有限公司 | SQL Server database recovery method, terminal device and storage medium |
| CN112612648B (en) * | 2020-12-29 | 2022-11-08 | 厦门市美亚柏科信息股份有限公司 | SQL Server database recovery method, terminal equipment and storage medium |
| CN114064664A (en) * | 2022-01-17 | 2022-02-18 | 北京奥星贝斯科技有限公司 | Method and device for inquiring transaction modification content in database |
| WO2023134518A1 (en) * | 2022-01-17 | 2023-07-20 | 北京奥星贝斯科技有限公司 | Method and apparatus for querying modified content of transaction in database |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107220145B (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107220145A (en) | A kind of method of flash memory database data recovery | |
| CN104537037B (en) | A kind of method and device of processing data storehouse daily record | |
| US7552150B2 (en) | Reason-based database changes | |
| CN101183322B (en) | Method for deferred logging and apparatus thereof | |
| EP3441890A1 (en) | Transaction recovery method in database system and database management system | |
| CN100430902C (en) | Distributing real-time data bank fault recovering method capable of supporting serving and recovering simultaneously | |
| CN107992269B (en) | Transaction writing method based on deduplication SSD | |
| CN103744936A (en) | Multi-version concurrency control method in database and database system | |
| US20140236911A1 (en) | Low Level Object Version Tracking Using Non-Volatile Memory Write Generations | |
| US9280572B2 (en) | Managing product information versions | |
| CN107851037A (en) | The consistency protocol that hardware transactional in the shared memory using locking and unlocked nonvolatile storage stores | |
| CN110555770B (en) | Block chain world state checking and recovering method based on incremental hash | |
| WO2020143240A1 (en) | Method for quickly recovering data in flash memory database | |
| CN104899117B (en) | Memory database parallel logging method towards Nonvolatile memory | |
| CN106095613A (en) | Use the rollback database of WAL | |
| KR102195839B1 (en) | Technique for log records management in database management system | |
| CN110096509A (en) | Realize that historical data draws the system and method for storage of linked list modeling processing under big data environment | |
| CN106681848A (en) | Data consistency guarantee method and system of erasure code RAID | |
| CN111414320B (en) | Method and system for constructing disk cache based on nonvolatile memory of log file system | |
| US9411692B2 (en) | Applying write elision | |
| KR20110046118A (en) | Adaptive logging apparatus and method | |
| CN103678608B (en) | Blog management method and device | |
| CN105556486B (en) | Method and system for managing the memory dynamically distributed in computing unit automatically | |
| US20110113017A1 (en) | Supporting Internal Consistency Checking with Consistency Coded Journal File Entries | |
| CN112597070B (en) | Object recovery method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |