CN102012849A - Flash memory-based database restoring method - Google Patents
Flash memory-based database restoring method Download PDFInfo
- Publication number
- CN102012849A CN102012849A CN 201010552789 CN201010552789A CN102012849A CN 102012849 A CN102012849 A CN 102012849A CN 201010552789 CN201010552789 CN 201010552789 CN 201010552789 A CN201010552789 A CN 201010552789A CN 102012849 A CN102012849 A CN 102012849A
- Authority
- CN
- China
- Prior art keywords
- flash memory
- log
- checkpoint
- record
- page
- 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
Images
Abstract
The invention relates to a flash memory-based database restoring method. The method is characterized in that: an internal memory, a flash memory, a check point management module, a system startup module and a system shutdown module are provided, wherein a data page buffer area, a log buffer area and a log area abstract are stored in the internal memory; a log area, a data area, a page table and a bit diagram for identifying a flash memory space allocation status; the check point management module sets up check points in the log area; the log area needs to be scanned when the system startup module is started, so that acquired logs and data are combined; a series of logs need to be inserted into the log area when the system shutdown module is started, and a series of operation is performed to finish system shutdown; and part of logs are recorded in the log buffer area, and the logs are written into the flash memory. The flash memory-based database restoring method can reduce erasing to the flash memory and fully exert the advantages of the flash memory, and can be widely applied to the restoring of system databases.
Description
Technical field
The present invention relates to a kind of database recovery method, particularly about a kind of database recovery method based on flash memory.
Background technology
Along with being extensive use of of flash memory, more and more data is stored on the flash memory device.The flash data management becomes a new research field.Database technology based on flash memory is faced with huge challenge, has caused that industry more and more pays close attention to.If existing database technology at the disk storage device design is run directly on the flash memory, the speed of reading in query processing has only improved 1～10 times, and the speed that writes has only improved 1～3 times.And in fact, the read or write speed of flash memory is the microsecond level, is about 100 times of disk read-write speed.This shows that existing Database Systems towards traditional magnetic disk can not be given full play to the high speed readwrite bandwidth of flash memory compared to disk if run directly on the flash memories, the excellent readwrite performance of flash memory is had a greatly reduced quality.Its reason is that there is huge difference in the physical characteristics of flash memory and disk.For example, when the enterprising line data of flash memory upgrades, owing to flash memory is must wipe whole before read-write unit and covering are write with the page or leaf, even the fraction data in one page are upgraded, for the original place of supporting existing database new model more, also need the data of this page place piece are all read in internal memory, revise this page data then, wipe whole again, the data with the monoblock in the internal memory write back in the flash memory at last.This has greatly reduced more new capability, thereby makes the more new capability of database sharply descend, and the affairs throughput of system does not obtain due raising.
The restoration methods of system database generally has two parts to constitute, and the one, when the affairs normal process, can keep enough information and can be used for fault recovery; The 2nd, the measure that will take after fault takes place to conforming state, keeps atomicity and persistent state of affairs with database recovery.These two aspects are closely linked systems, and the type of the information of maintenance and degree have determined the type of the step that the fault generation will be taked afterwards and the size of cost.The size of overhead when operation types after equally, fault takes place and cost size are also determining the affairs normal process.Existing database recovery method has following several: (1) is widely adopted in disk based on the recovery technology of daily record.This method has produced the way of various log and recovery, has various committing protocol instructing the process of log and recovery.Under the different agreements, the design of log record, the management of daily record/data buffer, checkpointing mechanism, the process of log and recovery is all very different.(2) shadow page is another famous transaction recovery technology, and it does not need explicit rolling back action.The main thought of shadow page technology is the more new model that adopts the strange land to upgrade, and safeguards two page tables in the life cycle of affairs: current page table (Current Page Table) and shadow page table (Shadow Page Table).When affairs began, these two page tables were identical.Shadow page table never changes in the affairs implementation, and the new address of affairs after to the renewal of database writes in the current page table.When affairs were submitted to, current page table was written in the nonvolatile memory, and replaced current page table and shadow page table with an atomic operation, then the current page telogenesis for shadow page table, next affairs are carried out.If current page table is lost during system crash, database can recover with shadow page table.(3) framework all need be dealt with various faults and handle the problem of recovering from disaster at various system softwares on the operating system (as the data base management system) and application software by the interface the affairs (Transaction).If the interface of affairs (Transaction) can be provided in the file system aspect, will reduce on it software greatly in order to deal with the redundant cost that the system failure is paid separately.
A lot of previous work are devoted to utilize copy-on-write (copy-on-write) and daily record technology that such transaction interface is provided on disk file system.But, because the data slot (data fragmentation) that the data that copy-on-write (copy-on-write) technology can make script cluster on disk to be deposited become and very disperse.Data write efficient discrete on the such plant equipment of disk is very low.In order to address this problem, checkpoint (checkpoint) and regular removing (cleaning) the mechanism discrete data of recombinating is adopted in a lot of work.Studies show that such reorganization expense is very expensive, make and support the file system of affairs to lose more than gain.The physical characteristics of flash memories has been fit to the requirement of transactional file system well, supports that on flash memory file system transaction interface becomes good selection.The affairs flash memory (Tansactional Flash) provide transaction interface on flash memory file system, it has adopted metadata and circle committing protocol based on every page to reduce performance and the space expense that extra submission (commit) is write down.Its main way is that the page or leaf that each transactions modify is crossed is coupled together with the metadata pointer, forms a circle.Like this, if affairs are normally submitted to, its redaction will be in a circle, if the abnormal of affairs submitted to, its amended data page just can not form a circle.The circle committing protocol provides transaction interface in the file system aspect with very little expense, but this interface is not perfect, and garbage reclamation, rollback and system recovery also need the support of complexity logic, and its expense is also relatively large.
Summary of the invention
At the problems referred to above, the purpose of this invention is to provide a kind of can the minimizing, and can make flash memory give full play to the database recovery method based on flash memory of its efficient read write attribute the flash memory erase operation.
For achieving the above object, the present invention takes following technical scheme: a kind of database recovery method based on flash memory, it is characterized in that: it comprises an internal memory, a flash memory, a checkpoint administration module, a system start-up module and a system closing module, stores data page buffer zone, log buffer district and log area summary in the described internal memory; Store log area, data field, page table in the described flash memory and be used for identifying described flash memory space allocation situation the position diagrammatic sketch; Described checkpoint administration module is set up the checkpoint in described log area; Described system start-up module need scan described log area when starting, and daily record and the data that obtain are merged; Need in described log area, insert a series of daily records when described system closing module starts, carry out a series of operation simultaneously and finish system closing; Described log buffer district writes down a part of daily record, and daily record is write with a brush dipped in Chinese ink on the described flash memory.
The establishment step of the checkpoint of described checkpoint administration module is as follows: (1) inserts a record＜checkpoint numbering in log area, active transaction tabulation, beginning〉indicate the beginning of this checkpoint; (2) etc. after all of active transaction tabulation the inside are enlivened all explicit submission of affairs or cancelled, insert an other record＜checkpoint numbering, finish in log area, indicate the normal termination of this checkpoint.
The start-up course of described system start-up module is as follows: the afterbody in (1) audit log district, judge whether that the checkpoint end record＜checkpoint of sign normal shutdown is numbered, finish exist; If there is not this record, then judge enlivening of shutdown checkpoint and whether have the affairs of submitting to or cancelling in the transaction list, exist then and number in the one＜affairs of afterbody interpolation of log area, abandon next explicit this affairs of cancelling of record, and then insertion＜checkpoint numbering, finish record indicates the end of this checkpoint; On the contrary, then direct insertion＜checkpoint numbering finishes〉record indicates the end of this checkpoint; (2) scanning log area is rebuild the log area summary in internal memory.
Step when described system closing module starts is as follows: (1) inserts record＜checkpoint numbering that the sign checkpoint begins, active transaction tabulation, beginning 〉, the operation of beginning system closing; (2) data with the data page buffer zone scan out in the data field of flash memory, and add relevant log record in the log buffer district; (3) log record with the page table log area scans out in the log area of flash memory; (4) insert the record＜checkpoint numbering that finishes the sign checkpoint, finish 〉.
The management of described data page buffer zone is when described data page buffer zone is filled with, scan out data page according to replacing algorithm, and the afterbody in described log buffer district increases a page table log record＜affairs numbering, logical page address, physical page address〉write down the mapping relations between this page logical address and the physical address.
In the management in described log buffer district, the log area of described log buffer district and flash memory all requirement writes in proper order and scans out in proper order.
The step of searching page address in the described page table is as follows: (1) searches the log buffer district, has judged whether successfully to submit transactions modify to logical page (LPAGE) to be found, if having, then returns the physical address that nearest successful submission office identifies; Otherwise, enter next step; (2) search the log area summary, judge whether logical page address to be found, return if having then; Otherwise, enter next step; (3) search log area, judged whether successfully to submit transactions modify to logical page (LPAGE) to be found, if having, then return the physical address that nearest successful submission office represents; Otherwise, enter next step; (4) search in the page table the inside, finish the back page address.
When described log area is filled with, described page table and log area are merged the big or small S of described log area
rSatisfy following formula:
Wherein, S
fRepresent the size of described flash memory, S
pRepresent the size of the page in the described flash memory, S
lRepresent the size of page table log record.
The present invention is owing to take above technical scheme, and it has the following advantages: 1, the present invention is owing to adopt log-structured based on flash memory, and what given full play to flash memory is the operating characteristic of unit with the page or leaf, has reduced wiping flash memory simultaneously.Therefore the affairs throughput of system is greatly improved, and the superior function of flash memory is not fully exerted.2, the present invention is being better than traditional restoration methods at disk database owing to adopt the physical characteristics that can adapt to flash memories based on the database recovery method of flash memory well aspect resume speed and the restore data.3, the present invention does not change the two-layer storage organization of traditional internal memory and flash memory, and supports complete transaction interface simultaneously, and its versatility is guaranteed, and helps the quick application of flash memories.4, the present invention can operate on the flash memory device easily owing to be based on the application of disk file system.The present invention can be widely used in the recovery of system database.
Description of drawings
Fig. 1 is an one-piece construction synoptic diagram of the present invention;
Fig. 2 is that schematic flow sheet is set up in the checkpoint in the administration module of checkpoint of the present invention;
Fig. 3 is that system start-up module of the present invention starts schematic flow sheet;
Fig. 4 is that system closing module of the present invention starts schematic flow sheet;
Fig. 5 is that the present invention searches page address schematic flow sheet in the page table.
Embodiment
Below in conjunction with drawings and Examples the present invention is described in detail.
The more new model that the present invention adopts the strange land to upgrade removes to manage free space on the flash memory with a page table.The new page table mapping address page table that do not write direct, but the mode record by daily record.By the feasible modification for page table of the mode of daily record is atom, obtains the atomicity of affairs with this.
As shown in Figure 1, the present invention includes an internal memory 1, a flash memory 2, a checkpoint administration module 3, a system start-up module 4 and a system closing module 5, wherein, store data page buffer zone 6, log buffer district 7 and log area summary 8 in the internal memory 1; Store log area 9, data field 10, page table 11 in the flash memory 2 and be used for identifying flash memory 2 allocation of space situations the position diagrammatic sketch 12; Checkpoint administration module 3 is mainly set up the checkpoint in log area 9, promptly insert a checkpoint daily record, in order to finish recovery operation more efficiently; System start-up module 4 need scan log area 9 when starting, and daily record and the data that obtain are merged; Need in log area 9, insert a series of daily records when system closing module 5 starts, carry out a series of operation simultaneously and finish system closing; 7 records a part of daily record in log buffer district is write with a brush dipped in Chinese ink daily record on the flash memory 2 then, reaches the purpose of permanent storage.The present invention searches according to log content, owing in the log content logical address of record data and the mapping relations of physical address are arranged, and therefore by searching the actual position that daily record can be found data.
Log area 9 constitutes by two row the record of each page table 11, and first classifies logical page address as, and secondary series is this logical page address corresponding physical page address, and then this page table log record has following two types:
(1)＜and the affairs numbering, transaction status 〉;
(2)＜and the affairs numbering, logical page address, physical page address 〉.
Wherein, the affairs numbering is the ID numbering of affairs, is the unique identification of affairs.The value of transaction status can be " submission " or " abandoning ".Article one,＜and affairs numberings, logical page address, physical page address〉record transforms into a corresponding physical page address to a logical page address.
As shown in Figure 2, in the foregoing description, the checkpoint of checkpoint administration module 3 is used for reducing maintenance log district summary 8 in log area 9, and the page table log record number that will scan when merging page table 11 and log area 9.When system was closed with log area 9 long enoughs, the checkpoint was inserted in the log area 9, and then the establishment step of checkpoint is as follows:
(1) in log area 9, inserts a record＜checkpoint numbering, the active transaction tabulation, beginning〉indicate the beginning of this checkpoint, wherein the checkpoint numbering is the unique identification of this checkpoint, active transaction tabulation the inside comprises this moment still at active affairs ID numbered list;
(2) etc. after all of active transaction tabulation the inside are enlivened all explicit submission of affairs or cancelled, insert an other record＜checkpoint numbering, finish in log area 9, indicate the normal termination of this checkpoint.
As shown in Figure 3, in the various embodiments described above, when system start-up module 4 restarts, check at first whether shutdown last time is normal, whether start this time is to restart from the primary system collapse, does corresponding processing then; Must scan the log area summary 8 in the log area 9 reconstruction internal memories 1 at last.The start-up course of system start-up module 4 is as follows:
(1) afterbody in audit log district 9 judges whether that the checkpoint end record＜checkpoint of sign normal shutdown is numbered, and finishes〉exist; If this record not, illustrative system is from once restarting the collapse, then must finish the checkpoint process of not finishing last time, promptly for those enlivening in the transaction list in the shutdown checkpoint, but during the affairs that also do not have explicit submission or cancel when collapse takes place, then the afterbody in log area 9 adds a log record＜affairs numbering, abandons〉next explicit this affairs of cancelling, and then insertion＜checkpoint numbering, finish record indicates the end of this checkpoint; Otherwise, if when collapse takes place explicit submission or the affairs cancelled, then direct insertions＜checkpoint numbering finishes〉write down the end that indicates this checkpoint;
(2) scanning log area 9 is rebuild log area summary 8 in internal memory.
As shown in Figure 4, in the various embodiments described above, when system closing module 5 started, all data in data buffer 6 and log buffer district 7 were scanned out respectively in the data field 10 and log area 9 on the flash memory 2.Meanwhile, a checkpoint is inserted into log area 9 afterbodys to indicate this normal shutdown.Step when system closing module 5 starts is as follows:
(1) insert record＜checkpoint numbering that the sign checkpoint begins, active transaction tabulation, beginning 〉, the operation of beginning system closing;
(2) data with data page buffer zone 6 scan out in the data field 10 of flash memory 2, and add relevant log record in log buffer district 7;
(3) log record with page table log area 9 scans out in the log area 9 of flash memory 2;
(4) insert the record＜checkpoint numbering that finishes the sign checkpoint, finish 〉.
In the above-mentioned steps, if system failure abnormal shutdown, the checkpoint during normal shutdown then can not find the record of corresponding sign checkpoint normal termination with imperfect at the end of log area 9.
In the various embodiments described above, the management of data page buffer zone 6 is when not having enough space requirements to eliminate data page when data page buffer zone 6 is filled with, scan out data page according to replacing algorithm, and the afterbody in log buffer district 9 increases a page table log record＜affairs numbering, logical page address, physical page address〉write down the mapping relations between this page logical address and the physical address.
In the various embodiments described above, in the management in log buffer district 7, the log area 9 of log buffer district 7 and flash memory 2 all requirement writes in proper order and scans out in proper order, if the log record＜affairs numbering of the explicit submission affairs of sign is arranged in the daily record that log buffer district 7 scans out, abandon 〉, then will be from tail to head scanning log area 9, to the last the page table address that this submission transactions modify is crossed is upgraded in the starting position of a checkpoint in log area summary 8.
As shown in Figure 5, in the various embodiments described above, in searching page table 11, during the process of page address, search a logical page address at every turn and all must scan whole page table journal file from the beginning to the end.Sao Miao log record is too much if desired, and this search procedure might become the new bottleneck of system.All scan whole journal file for unnecessary at every turn,, promptly deposit recently the page address that frequentation is asked in the log area summary 8 page table address buffering in the log area summary 8 of internal memory 1 that frequentation is asked.Therefore, a logical page (LPAGE) correctly timely the physical page address just may have four places: among log buffer district 7, log area summary 8, log area 9 or the page table 11.So it is as follows to search the step of page address in the page table 11:
(1) searches log buffer district 7, judged whether successfully to submit transactions modify to logical page (LPAGE) to be found,, then return the physical address of nearest successful submission office sign if having; If do not find, then enter next step;
(2) search log area summary 8, judge whether logical page address to be found, if having then return; If do not find then enter next step;
(3) search log area 9, judged whether successfully to submit transactions modify to logical page (LPAGE) to be found,, then return the physical address that nearest successful submission office represents if having; If do not find, then enter next step;
(4) search in page table 11 the insides, finish the back page address.
In the above steps, because log buffer district 9 and log area summary 8 is all in internal memory 1, if the hitting area of the locality of visit and log area summary 8 is unlikely to too poor, the page address search procedure just can be returned after preceding two steps.This search procedure will be enough fast like this, and be unlikely to become the bottleneck of system.Show that through experimental analysis as long as spent very little extra cost, the hit rate of log area summary 8 just can reach more than 90%.
In the various embodiments described above, when log area 9 is filled with, then page table 11 and log area 9 to be merged, and then produce new page table, wipe log area 9 then and deposit the page table log record once more.Regularly carry out such merging process and also can reduce the page table daily record number of the log area 9 that when log area summary 8 does not hit, need scan, accelerate the production process of page address.Scrap in order to prevent log area 9 from reaching the erasing times restriction before the serviceable life of whole flash memory 2, the big or small Sr of log area 9 need satisfy following formula:
Wherein, S
fThe size of sign flash memory 2, S
pThe size of the page in the sign flash memory 2, S
lThe size of sign page table log record.
The merging process of page table 11 and log area 9 also can carry out on the backstage, and is similarly machine-processed in employing of the present invention and shadow page and the flash translation layer (FTL): as to preserve two versions of page table 11 when merging beginning, for example PT0 and PT1; Keep PT0 constant then in the process that merges, merge PT1 and log area 9 on the backstage, new page table writes the position of original PT1, merges the pointed PT1 that successfully points to page table 11 with an atomic operation modification afterwards and makes it to become new effective page table; Wipe log area 9 then, finish the merging process of page table 11 and log area 9.Atomicity after can effectively guaranteeing to merge like this.
The various embodiments described above only are used to illustrate the present invention, and wherein each step all can change to some extent, and every equivalents of carrying out on the basis of technical solution of the present invention and improvement all should not got rid of outside protection scope of the present invention.
Claims (8)
1. database recovery method based on flash memory, it is characterized in that: it comprises an internal memory, a flash memory, a checkpoint administration module, a system start-up module and a system closing module, stores data page buffer zone, log buffer district and log area summary in the described internal memory; Store log area, data field, page table in the described flash memory and be used for identifying described flash memory space allocation situation the position diagrammatic sketch; Described checkpoint administration module is set up the checkpoint in described log area; Described system start-up module need scan described log area when starting, and daily record and the data that obtain are merged; Need in described log area, insert a series of daily records when described system closing module starts, carry out a series of operation simultaneously and finish system closing; Described log buffer district writes down a part of daily record, and daily record is write with a brush dipped in Chinese ink on the described flash memory.
2. a kind of database recovery method as claimed in claim 1 based on flash memory, it is characterized in that: the establishment step of the checkpoint of described checkpoint administration module is as follows:
(1) in log area, insert a record＜checkpoint numbering, active transaction tabulation, beginning〉indicate the beginning of this checkpoint;
(2) etc. after all of active transaction tabulation the inside are enlivened all explicit submission of affairs or cancelled, insert an other record＜checkpoint numbering, finish in log area, indicate the normal termination of this checkpoint.
3. a kind of database recovery method as claimed in claim 1 based on flash memory, it is characterized in that: the start-up course of described system start-up module is as follows:
(1) afterbody in audit log district judges whether that the checkpoint end record＜checkpoint of sign normal shutdown is numbered, and finishes〉exist; If there is not this record, then judge enlivening of shutdown checkpoint and whether have the affairs of submitting to or cancelling in the transaction list, exist then and number in the one＜affairs of afterbody interpolation of log area, abandon next explicit this affairs of cancelling of record, and then insertion＜checkpoint numbering, finish record indicates the end of this checkpoint; On the contrary, then direct insertion＜checkpoint numbering finishes〉record indicates the end of this checkpoint;
(2) scanning log area is rebuild the log area summary in internal memory.
4. a kind of database recovery method based on flash memory as claimed in claim 1 is characterized in that: the step when described system closing module starts is as follows:
(1) insert record＜checkpoint numbering that the sign checkpoint begins, active transaction tabulation, beginning 〉, the operation of beginning system closing;
(2) data with the data page buffer zone scan out in the data field of flash memory, and add relevant log record in the log buffer district;
(3) log record with the page table log area scans out in the log area of flash memory;
(4) insert the record＜checkpoint numbering that finishes the sign checkpoint, finish 〉.
5. a kind of database recovery method as claimed in claim 1 based on flash memory, it is characterized in that: the management of described data page buffer zone is when described data page buffer zone is filled with, scan out data page according to replacing algorithm, and the afterbody in described log buffer district increases a page table log record＜affairs numbering, logical page address, physical page address〉write down the mapping relations between this page logical address and the physical address.
6. a kind of database recovery method based on flash memory as claimed in claim 1 is characterized in that: in the management in described log buffer district, the log area of described log buffer district and flash memory all requirement writes in proper order and scans out in proper order.
7. a kind of database recovery method as claimed in claim 1 based on flash memory, it is characterized in that: the step of searching page address in the described page table is as follows:
(1) searches the log buffer district, judged whether successfully to submit transactions modify to logical page (LPAGE) to be found, if having, then return the physical address that nearest successful submission office identifies; Otherwise, enter next step;
(2) search the log area summary, judge whether logical page address to be found, return if having then; Otherwise, enter next step;
(3) search log area, judged whether successfully to submit transactions modify to logical page (LPAGE) to be found, if having, then return the physical address that nearest successful submission office represents; Otherwise, enter next step;
(4) search in the page table the inside, finish the back page address.
8. as claim 1 or 5 described a kind of database recovery methods, it is characterized in that: when described log area is filled with, described page table and log area are merged the big or small S of described log area based on flash memory
rSatisfy following formula:
Wherein, S
fRepresent the size of described flash memory, S
pRepresent the size of the page in the described flash memory, S
lRepresent the size of page table log record.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010552789A CN102012849B (en) | 2010-11-19 | 2010-11-19 | Flash memory-based database restoring method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010552789A CN102012849B (en) | 2010-11-19 | 2010-11-19 | Flash memory-based database restoring method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102012849A true CN102012849A (en) | 2011-04-13 |
| CN102012849B CN102012849B (en) | 2012-10-24 |
Family
ID=43843025
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010552789A Expired - Fee Related CN102012849B (en) | 2010-11-19 | 2010-11-19 | Flash memory-based database restoring method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102012849B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102750317A (en) * | 2012-05-02 | 2012-10-24 | 华为技术有限公司 | Method and device for data persistence processing and data base system |
| WO2013091349A1 (en) * | 2011-12-21 | 2013-06-27 | 华为技术有限公司 | Method and device for recovering description information and caching data in database |
| CN104537037A (en) * | 2014-12-23 | 2015-04-22 | 杭州华为数字技术有限公司 | Method and device for processing database logs |
| CN105677879A (en) * | 2016-01-12 | 2016-06-15 | 诸葛晴凤 | Data organizing method and accessing method for relational database in memory |
| 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 |
| US9684596B2 (en) | 2015-02-25 | 2017-06-20 | Microsoft Technology Licensing, Llc | Application cache replication to secondary application(s) |
| US9684607B2 (en) | 2015-02-25 | 2017-06-20 | Microsoft Technology Licensing, Llc | Automatic recovery of application cache warmth |
| CN107515827A (en) * | 2017-08-21 | 2017-12-26 | 湖南国科微电子股份有限公司 | Storage method, device and the SSD of the self-defined daily records of PCIE SSD |
| CN107562567A (en) * | 2016-06-30 | 2018-01-09 | 华为技术有限公司 | A kind of data processing method and data processing equipment |
| CN110019280A (en) * | 2013-03-15 | 2019-07-16 | 亚马逊科技公司 | System scope checkpoint for distributed data base system avoids |
| CN113821382A (en) * | 2021-11-24 | 2021-12-21 | 西安热工研究院有限公司 | Real-time database data processing method, system and equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1851672A (en) * | 2006-04-05 | 2006-10-25 | 北京飞天诚信科技有限公司 | Flashmemory safety read-write method |
| US20090287874A1 (en) * | 2008-05-13 | 2009-11-19 | Microsoft Corporation | Flash Recovery Employing Transaction Log |
-
2010
- 2010-11-19 CN CN201010552789A patent/CN102012849B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1851672A (en) * | 2006-04-05 | 2006-10-25 | 北京飞天诚信科技有限公司 | Flashmemory safety read-write method |
| US20090287874A1 (en) * | 2008-05-13 | 2009-11-19 | Microsoft Corporation | Flash Recovery Employing Transaction Log |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013091349A1 (en) * | 2011-12-21 | 2013-06-27 | 华为技术有限公司 | Method and device for recovering description information and caching data in database |
| US8521959B2 (en) | 2011-12-21 | 2013-08-27 | Huawei Technologies Co., Ltd. | Method and device for recovering description information, and method and device for caching data in database |
| CN102750317B (en) * | 2012-05-02 | 2015-01-21 | 华为技术有限公司 | Method and device for data persistence processing and data base system |
| CN102750317A (en) * | 2012-05-02 | 2012-10-24 | 华为技术有限公司 | Method and device for data persistence processing and data base system |
| CN110019280B (en) * | 2013-03-15 | 2024-01-19 | 亚马逊科技公司 | System-wide checkpoint avoidance for distributed database systems |
| CN110019280A (en) * | 2013-03-15 | 2019-07-16 | 亚马逊科技公司 | System scope checkpoint for distributed data base system avoids |
| CN104537037B (en) * | 2014-12-23 | 2018-01-23 | 杭州华为数字技术有限公司 | A kind of method and device of processing data storehouse daily record |
| CN104537037A (en) * | 2014-12-23 | 2015-04-22 | 杭州华为数字技术有限公司 | Method and device for processing database logs |
| US9684596B2 (en) | 2015-02-25 | 2017-06-20 | Microsoft Technology Licensing, Llc | Application cache replication to secondary application(s) |
| US9684607B2 (en) | 2015-02-25 | 2017-06-20 | Microsoft Technology Licensing, Llc | Automatic recovery of application cache warmth |
| US10204048B2 (en) | 2015-02-25 | 2019-02-12 | Microsoft Technology Licensing, Llc | Replicating a primary application cache within a secondary application cache |
| US10114765B2 (en) | 2015-02-25 | 2018-10-30 | Microsoft Technology Licensing, Llc | Automatic recovery of application cache warmth |
| CN105677879A (en) * | 2016-01-12 | 2016-06-15 | 诸葛晴凤 | Data organizing method and accessing method for relational database in memory |
| CN105677879B (en) * | 2016-01-12 | 2019-10-18 | 诸葛晴凤 | The data organization and access method of relationship memory database |
| WO2017190604A1 (en) * | 2016-05-06 | 2017-11-09 | 华为技术有限公司 | Transaction recovery method in database system and 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 |
| CN107562567A (en) * | 2016-06-30 | 2018-01-09 | 华为技术有限公司 | A kind of data processing method and data processing equipment |
| CN107515827A (en) * | 2017-08-21 | 2017-12-26 | 湖南国科微电子股份有限公司 | Storage method, device and the SSD of the self-defined daily records of PCIE SSD |
| CN107515827B (en) * | 2017-08-21 | 2021-07-27 | 湖南国科微电子股份有限公司 | PCIE SSD custom log storage method and device and SSD |
| CN113821382A (en) * | 2021-11-24 | 2021-12-21 | 西安热工研究院有限公司 | Real-time database data processing method, system and equipment |
| CN113821382B (en) * | 2021-11-24 | 2022-03-01 | 西安热工研究院有限公司 | Real-time database data processing method, system and equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102012849B (en) | 2012-10-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102012849B (en) | Flash memory-based database restoring method | |
| US20190073277A1 (en) | Transaction Recovery Method in Database System, and Database Management System | |
| Prabhakaran et al. | Transactional Flash. | |
| US8312326B2 (en) | Delta checkpoints for a non-volatile memory indirection table | |
| US9183236B2 (en) | Low level object version tracking using non-volatile memory write generations | |
| CN1205550C (en) | A method and system for highly-parallel logging and recovery operation in main-memory transaction processing systems | |
| US8239356B2 (en) | Methods and apparatuses for data protection | |
| CN103577121B (en) | A kind of highly reliable linear file access method based on Nand Flash | |
| Graefe | A survey of B-tree logging and recovery techniques | |
| EP2590078B1 (en) | Shadow paging based log segment directory | |
| CN105159818A (en) | Log recovery method in memory data management and log recovery simulation system in memory data management | |
| Eich | Mars: The design of a main memory database machine | |
| JPH07175700A (en) | Database management system | |
| US8200627B2 (en) | Journaling database changes using a bit map for zones defined in each page | |
| US8108356B2 (en) | Method for recovering data in a storage system | |
| CN103577574A (en) | High-reliability linear file system based on nand flash | |
| CN104516959A (en) | Method and device for managing database logs | |
| CN113515501B (en) | Nonvolatile memory database management system recovery method and device and electronic equipment | |
| Graefe et al. | Instant recovery with write-ahead logging | |
| TWI737189B (en) | Method for facilitating recovery from crash of solid-state storage device, computer system, and solid-state storage device | |
| Xu et al. | Building a fast and efficient LSM-tree store by integrating local storage with cloud storage | |
| Do et al. | Fast peak-to-peak behavior with SSD buffer pool | |
| KR101419428B1 (en) | Apparatus for logging and recovering transactions in database installed in a mobile environment and method thereof | |
| Svobodova | Management of object histories in the SWALLOW repository | |
| CN114816224A (en) | Data management method and data management device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121024 Termination date: 20151119 |
|
| EXPY | Termination of patent right or utility model |